diff --git "a/groundtruth/parser/font_11.pdf.page_no_12.py.json.word.txt" "b/groundtruth/parser/font_11.pdf.page_no_12.py.json.word.txt" deleted file mode 100644--- "a/groundtruth/parser/font_11.pdf.page_no_12.py.json.word.txt" +++ /dev/null @@ -1,927 +0,0 @@ -(039.69, 020.40) (268.42, 020.40) (268.42, 027.41) (039.69, 027.41) /T1_0 NatureMachineIntelligence|Volume8|June2026|984-996 <|special_separator|> -(547.11, 019.82) (562.68, 019.82) (562.68, 026.80) (547.11, 026.80) /T1_1 995 <|special_separator|> -(039.69, 757.67) (071.22, 757.67) (071.22, 766.42) (039.69, 766.42) /T1_0 Article <|special_separator|> -(391.20, 758.00) (569.67, 758.00) (569.67, 764.99) (391.20, 764.99) /T1_1 https://doi.org/10.1038/s42256-026-01242-8 <|special_separator|> -(039.69, 734.06) (049.30, 734.06) (049.30, 741.07) (039.69, 741.07) /T1_2 15. <|special_separator|> -(056.69, 734.06) (067.27, 734.06) (067.27, 741.07) (056.69, 741.07) /T1_2 Qi, <|special_separator|> -(068.87, 734.06) (076.28, 734.06) (076.28, 741.07) (068.87, 741.07) /T1_2 X. <|special_separator|> -(077.88, 734.06) (085.16, 734.06) (085.16, 741.07) (077.88, 741.07) /T1_2 et <|special_separator|> -(086.76, 734.06) (095.47, 734.06) (095.47, 741.07) (086.76, 741.07) /T1_2 al. <|special_separator|> -(097.07, 734.06) (134.51, 734.06) (134.51, 741.07) (097.07, 741.07) /T1_2 Predicting <|special_separator|> -(136.11, 734.06) (189.19, 734.06) (189.19, 741.07) (136.11, 741.07) /T1_2 transcriptional <|special_separator|> -(190.79, 734.06) (228.18, 734.06) (228.18, 741.07) (190.79, 741.07) /T1_2 responses <|special_separator|> -(229.78, 734.06) (237.35, 734.06) (237.35, 741.07) (229.78, 741.07) /T1_2 to <|special_separator|> -(238.94, 734.06) (259.07, 734.06) (259.07, 741.07) (238.94, 741.07) /T1_2 novel <|special_separator|> -(260.67, 734.06) (294.37, 734.06) (294.37, 741.07) (260.67, 741.07) /T1_2 chemical <|special_separator|> -(056.69, 723.30) (106.33, 723.30) (106.33, 730.32) (056.69, 730.32) /T1_2 perturbations <|special_separator|> -(107.93, 723.30) (127.94, 723.30) (127.94, 730.32) (107.93, 730.32) /T1_2 using <|special_separator|> -(129.54, 723.30) (148.35, 723.30) (148.35, 730.32) (129.54, 730.32) /T1_2 deep <|special_separator|> -(149.95, 723.30) (188.77, 723.30) (188.77, 730.32) (149.95, 730.32) /T1_2 generative <|special_separator|> -(190.37, 723.30) (213.97, 723.30) (213.97, 730.32) (190.37, 730.32) /T1_2 model <|special_separator|> -(215.57, 723.30) (225.78, 723.30) (225.78, 730.32) (215.57, 730.32) /T1_2 for <|special_separator|> -(227.38, 723.30) (244.86, 723.30) (244.86, 730.32) (227.38, 730.32) /T1_2 drug <|special_separator|> -(246.46, 723.30) (283.37, 723.30) (283.37, 730.32) (246.46, 730.32) /T1_2 discovery. <|special_separator|> -(056.69, 712.55) (108.65, 712.55) (108.65, 719.57) (056.69, 719.57) /T1_3 Nat.Commun. <|special_separator|> -(110.22, 712.57) (118.45, 712.57) (118.45, 719.57) (110.22, 719.57) /T1_4 15 <|special_separator|> -(118.32, 712.55) (120.57, 712.55) (120.57, 719.57) (118.32, 719.57) /T1_2 , <|special_separator|> -(122.17, 712.55) (140.82, 712.55) (140.82, 719.57) (122.17, 719.57) /T1_2 9256 <|special_separator|> -(142.42, 712.55) (168.22, 712.55) (168.22, 719.57) (142.42, 719.57) /T1_2 (2024). <|special_separator|> -(039.69, 701.80) (049.68, 701.80) (049.68, 708.81) (039.69, 708.81) /T1_2 16. <|special_separator|> -(056.69, 701.80) (069.09, 701.80) (069.09, 708.81) (056.69, 708.81) /T1_2 He, <|special_separator|> -(070.68, 701.80) (077.61, 701.80) (077.61, 708.81) (070.68, 708.81) /T1_2 S. <|special_separator|> -(079.21, 701.80) (086.49, 701.80) (086.49, 708.81) (079.21, 708.81) /T1_2 et <|special_separator|> -(088.09, 701.80) (096.81, 701.80) (096.81, 708.81) (088.09, 708.81) /T1_2 al. <|special_separator|> -(098.41, 701.80) (128.81, 701.80) (128.81, 708.81) (098.41, 708.81) /T1_2 Squidiff: <|special_separator|> -(130.41, 701.80) (168.22, 701.80) (168.22, 708.81) (130.41, 708.81) /T1_2 predicting <|special_separator|> -(169.82, 701.80) (197.31, 701.80) (197.31, 708.81) (169.82, 708.81) /T1_2 cellular <|special_separator|> -(198.91, 701.80) (247.73, 701.80) (247.73, 708.81) (198.91, 708.81) /T1_2 development <|special_separator|> -(249.33, 701.80) (263.24, 701.80) (263.24, 708.81) (249.33, 708.81) /T1_2 and <|special_separator|> -(056.69, 691.05) (094.08, 691.05) (094.08, 698.06) (056.69, 698.06) /T1_2 responses <|special_separator|> -(095.68, 691.05) (103.25, 691.05) (103.25, 698.06) (095.68, 698.06) /T1_2 to <|special_separator|> -(104.85, 691.05) (154.48, 691.05) (154.48, 698.06) (104.85, 698.06) /T1_2 perturbations <|special_separator|> -(156.08, 691.05) (176.10, 691.05) (176.10, 698.06) (156.08, 698.06) /T1_2 using <|special_separator|> -(177.70, 691.05) (182.02, 691.05) (182.02, 698.06) (177.70, 698.06) /T1_2 a <|special_separator|> -(183.62, 691.05) (215.26, 691.05) (215.26, 698.06) (183.62, 698.06) /T1_2 diffusion <|special_separator|> -(216.86, 691.05) (242.61, 691.05) (242.61, 698.06) (216.86, 698.06) /T1_2 model. <|special_separator|> -(244.21, 691.05) (258.80, 691.05) (258.80, 698.06) (244.21, 698.06) /T1_3 Nat. <|special_separator|> -(260.32, 691.05) (291.73, 691.05) (291.73, 698.06) (260.32, 698.06) /T1_3 Methods <|special_separator|> -(056.69, 680.31) (066.38, 680.31) (066.38, 687.31) (056.69, 687.31) /T1_4 23 <|special_separator|> -(066.25, 680.30) (068.50, 680.30) (068.50, 687.31) (066.25, 687.31) /T1_2 , <|special_separator|> -(070.10, 680.30) (091.38, 680.30) (091.38, 687.31) (070.10, 687.31) /T1_2 65-77 <|special_separator|> -(092.98, 680.30) (119.00, 680.30) (119.00, 687.31) (092.98, 687.31) /T1_2 (2026). <|special_separator|> -(039.69, 669.54) (048.33, 669.54) (048.33, 676.56) (039.69, 676.56) /T1_2 17. <|special_separator|> -(056.69, 669.54) (071.27, 669.54) (071.27, 676.56) (056.69, 676.56) /T1_2 Cui, <|special_separator|> -(072.87, 669.54) (080.86, 669.54) (080.86, 676.56) (072.87, 676.56) /T1_2 H. <|special_separator|> -(082.46, 669.54) (089.74, 669.54) (089.74, 676.56) (082.46, 676.56) /T1_2 et <|special_separator|> -(091.34, 669.54) (100.05, 669.54) (100.05, 676.56) (091.34, 676.56) /T1_2 al. <|special_separator|> -(101.65, 669.54) (126.50, 669.54) (126.50, 676.56) (101.65, 676.56) /T1_2 scGPT: <|special_separator|> -(128.10, 669.54) (153.68, 669.54) (153.68, 676.56) (128.10, 676.56) /T1_2 toward <|special_separator|> -(155.28, 669.54) (185.56, 669.54) (185.56, 676.56) (155.28, 676.56) /T1_2 building <|special_separator|> -(187.16, 669.54) (191.48, 669.54) (191.48, 676.56) (187.16, 676.56) /T1_2 a <|special_separator|> -(193.08, 669.54) (232.91, 669.54) (232.91, 676.56) (193.08, 676.56) /T1_2 foundation <|special_separator|> -(234.51, 669.54) (258.11, 669.54) (258.11, 676.56) (234.51, 676.56) /T1_2 model <|special_separator|> -(259.71, 669.54) (269.92, 669.54) (269.92, 676.56) (259.71, 676.56) /T1_2 for <|special_separator|> -(056.69, 658.79) (094.69, 658.79) (094.69, 665.81) (056.69, 665.81) /T1_2 single-cell <|special_separator|> -(096.29, 658.79) (139.73, 658.79) (139.73, 665.81) (096.29, 665.81) /T1_2 multi-omics <|special_separator|> -(141.33, 658.79) (161.34, 658.79) (161.34, 665.81) (141.33, 665.81) /T1_2 using <|special_separator|> -(162.94, 658.79) (201.76, 658.79) (201.76, 665.81) (162.94, 665.81) /T1_2 generative <|special_separator|> -(203.36, 658.79) (213.02, 658.79) (213.02, 665.81) (203.36, 665.81) /T1_2 AI. <|special_separator|> -(214.62, 658.79) (229.20, 658.79) (229.20, 665.81) (214.62, 665.81) /T1_3 Nat. <|special_separator|> -(230.72, 658.79) (262.14, 658.79) (262.14, 665.81) (230.72, 665.81) /T1_3 Methods <|special_separator|> -(263.81, 658.81) (271.87, 658.81) (271.87, 665.81) (263.81, 665.81) /T1_4 21 <|special_separator|> -(271.73, 658.79) (273.99, 658.79) (273.99, 665.81) (271.73, 665.81) /T1_2 , <|special_separator|> -(056.69, 648.04) (095.45, 648.04) (095.45, 655.05) (056.69, 655.05) /T1_2 1470-1480 <|special_separator|> -(097.05, 648.04) (122.85, 648.04) (122.85, 655.05) (097.05, 655.05) /T1_2 (2024). <|special_separator|> -(039.69, 637.29) (049.58, 637.29) (049.58, 644.30) (039.69, 644.30) /T1_2 18. <|special_separator|> -(056.69, 637.29) (077.00, 637.29) (077.00, 644.30) (056.69, 644.30) /T1_2 Yang, <|special_separator|> -(078.60, 637.29) (084.40, 637.29) (084.40, 644.30) (078.60, 644.30) /T1_2 F. <|special_separator|> -(086.00, 637.29) (093.28, 637.29) (093.28, 644.30) (086.00, 644.30) /T1_2 et <|special_separator|> -(094.88, 637.29) (103.60, 637.29) (103.60, 644.30) (094.88, 644.30) /T1_2 al. <|special_separator|> -(105.20, 637.29) (131.97, 637.29) (131.97, 644.30) (105.20, 644.30) /T1_2 scBERT <|special_separator|> -(133.57, 637.29) (141.63, 637.29) (141.63, 644.30) (133.57, 644.30) /T1_2 as <|special_separator|> -(143.23, 637.29) (147.55, 637.29) (147.55, 644.30) (143.23, 644.30) /T1_2 a <|special_separator|> -(149.15, 637.29) (189.51, 637.29) (189.51, 644.30) (149.15, 644.30) /T1_2 large-scale <|special_separator|> -(191.11, 637.29) (229.13, 637.29) (229.13, 644.30) (191.11, 644.30) /T1_2 pretrained <|special_separator|> -(230.73, 637.29) (249.54, 637.29) (249.54, 644.30) (230.73, 644.30) /T1_2 deep <|special_separator|> -(251.14, 637.29) (285.51, 637.29) (285.51, 644.30) (251.14, 644.30) /T1_2 language <|special_separator|> -(056.69, 626.54) (080.29, 626.54) (080.29, 633.55) (056.69, 633.55) /T1_2 model <|special_separator|> -(081.89, 626.54) (092.10, 626.54) (092.10, 633.55) (081.89, 633.55) /T1_2 for <|special_separator|> -(093.70, 626.54) (107.18, 626.54) (107.18, 633.55) (093.70, 633.55) /T1_2 cell <|special_separator|> -(108.78, 626.54) (125.10, 626.54) (125.10, 633.55) (108.78, 633.55) /T1_2 type <|special_separator|> -(126.70, 626.54) (166.18, 626.54) (166.18, 633.55) (126.70, 633.55) /T1_2 annotation <|special_separator|> -(167.78, 626.54) (175.27, 626.54) (175.27, 633.55) (167.78, 633.55) /T1_2 of <|special_separator|> -(176.87, 626.54) (214.87, 626.54) (214.87, 633.55) (176.87, 633.55) /T1_2 single-cell <|special_separator|> -(216.47, 626.54) (247.89, 626.54) (247.89, 633.55) (216.47, 633.55) /T1_2 RNA-seq <|special_separator|> -(249.49, 626.54) (267.86, 626.54) (267.86, 633.55) (249.49, 633.55) /T1_2 data. <|special_separator|> -(269.46, 626.54) (284.05, 626.54) (284.05, 633.55) (269.46, 633.55) /T1_3 Nat. <|special_separator|> -(056.69, 615.78) (078.69, 615.78) (078.69, 622.80) (056.69, 622.80) /T1_3 Mach. <|special_separator|> -(080.21, 615.78) (100.30, 615.78) (100.30, 622.80) (080.21, 622.80) /T1_3 Intell. <|special_separator|> -(101.88, 615.80) (107.12, 615.80) (107.12, 622.80) (101.88, 622.80) /T1_4 4 <|special_separator|> -(106.99, 615.78) (109.24, 615.78) (109.24, 622.80) (106.99, 622.80) /T1_2 , <|special_separator|> -(110.84, 615.78) (142.86, 615.78) (142.86, 622.80) (110.84, 622.80) /T1_2 852-866 <|special_separator|> -(144.46, 615.78) (170.04, 615.78) (170.04, 622.80) (144.46, 622.80) /T1_2 (2022). <|special_separator|> -(039.68, 605.03) (049.37, 605.03) (049.37, 612.05) (039.68, 612.05) /T1_2 19. <|special_separator|> -(056.69, 605.03) (087.62, 605.03) (087.62, 612.05) (056.69, 612.05) /T1_2 Noutahi, <|special_separator|> -(089.22, 605.03) (096.07, 605.03) (096.07, 612.05) (089.22, 612.05) /T1_2 E. <|special_separator|> -(097.67, 605.03) (104.95, 605.03) (104.95, 612.05) (097.67, 612.05) /T1_2 et <|special_separator|> -(106.55, 605.03) (115.26, 605.03) (115.26, 612.05) (106.55, 612.05) /T1_2 al. <|special_separator|> -(116.86, 605.03) (140.74, 605.03) (140.74, 612.05) (116.86, 612.05) /T1_2 Virtual <|special_separator|> -(142.34, 605.03) (161.78, 605.03) (161.78, 612.05) (142.34, 612.05) /T1_2 cells: <|special_separator|> -(163.38, 605.03) (191.82, 605.03) (191.82, 612.05) (163.38, 612.05) /T1_2 predict, <|special_separator|> -(193.42, 605.03) (222.10, 605.03) (222.10, 612.05) (193.42, 612.05) /T1_2 explain, <|special_separator|> -(223.70, 605.03) (256.53, 605.03) (256.53, 612.05) (223.70, 612.05) /T1_2 discover. <|special_separator|> -(258.13, 605.03) (287.17, 605.03) (287.17, 612.05) (258.13, 612.05) /T1_2 Preprint <|special_separator|> -(056.69, 594.28) (063.75, 594.28) (063.75, 601.29) (056.69, 601.29) /T1_2 at <|special_separator|> -(065.35, 594.28) (177.92, 594.28) (177.92, 601.29) (065.35, 601.29) /T1_2 http://arxiv.org/abs/2505.14613 <|special_separator|> -(179.52, 594.28) (205.35, 594.28) (205.35, 601.29) (179.52, 601.29) /T1_2 (2025). <|special_separator|> -(039.68, 583.53) (051.18, 583.53) (051.18, 590.54) (039.68, 590.54) /T1_2 20. <|special_separator|> -(056.69, 583.53) (090.71, 583.53) (090.71, 590.54) (056.69, 590.54) /T1_2 Csendes, <|special_separator|> -(092.31, 583.53) (102.36, 583.53) (102.36, 590.54) (092.31, 590.54) /T1_2 G., <|special_separator|> -(103.96, 583.53) (123.78, 583.53) (123.78, 590.54) (103.96, 590.54) /T1_2 Sanz, <|special_separator|> -(125.38, 583.53) (135.42, 583.53) (135.42, 590.54) (125.38, 590.54) /T1_2 G., <|special_separator|> -(137.02, 583.53) (162.04, 583.53) (162.04, 590.54) (137.02, 590.54) /T1_2 Szalay, <|special_separator|> -(163.64, 583.53) (170.72, 583.53) (170.72, 590.54) (163.64, 590.54) /T1_2 K. <|special_separator|> -(172.32, 583.53) (179.24, 583.53) (179.24, 590.54) (172.32, 590.54) /T1_2 Z. <|special_separator|> -(180.84, 583.53) (211.43, 583.53) (211.43, 590.54) (180.84, 590.54) /T1_2 &Szalai, <|special_separator|> -(213.03, 583.53) (220.15, 583.53) (220.15, 590.54) (213.03, 590.54) /T1_2 B. <|special_separator|> -(221.75, 583.53) (274.83, 583.53) (274.83, 590.54) (221.75, 590.54) /T1_2 Benchmarking <|special_separator|> -(056.69, 572.78) (096.53, 572.78) (096.53, 579.79) (056.69, 579.79) /T1_2 foundation <|special_separator|> -(098.13, 572.78) (111.61, 572.78) (111.61, 579.79) (098.13, 579.79) /T1_2 cell <|special_separator|> -(113.21, 572.78) (140.54, 572.78) (140.54, 579.79) (113.21, 579.79) /T1_2 models <|special_separator|> -(142.14, 572.78) (152.35, 572.78) (152.35, 579.79) (142.14, 579.79) /T1_2 for <|special_separator|> -(153.95, 572.78) (218.53, 572.78) (218.53, 579.79) (153.95, 579.79) /T1_2 post-perturbation <|special_separator|> -(220.13, 572.78) (251.54, 572.78) (251.54, 579.79) (220.13, 579.79) /T1_2 RNA-seq <|special_separator|> -(253.14, 572.78) (292.92, 572.78) (292.92, 579.79) (253.14, 579.79) /T1_2 prediction. <|special_separator|> -(056.69, 562.02) (104.36, 562.02) (104.36, 569.04) (056.69, 569.04) /T1_3 BMCGenom. <|special_separator|> -(105.92, 562.02) (117.08, 562.02) (117.08, 569.04) (105.92, 569.04) /T1_2 26, <|special_separator|> -(118.68, 562.02) (133.02, 562.02) (133.02, 569.04) (118.68, 569.04) /T1_2 393 <|special_separator|> -(134.62, 562.02) (160.45, 562.02) (160.45, 569.04) (134.62, 569.04) /T1_2 (2025). <|special_separator|> -(039.68, 551.27) (049.20, 551.27) (049.20, 558.29) (039.68, 558.29) /T1_2 21. <|special_separator|> -(056.69, 551.27) (112.04, 551.27) (112.04, 558.29) (056.69, 558.29) /T1_2 Ahlmann-Eltze, <|special_separator|> -(113.64, 551.27) (123.55, 551.27) (123.55, 558.29) (113.64, 558.29) /T1_2 C., <|special_separator|> -(125.15, 551.27) (149.57, 551.27) (149.57, 558.29) (125.15, 558.29) /T1_2 Huber, <|special_separator|> -(151.17, 551.27) (160.11, 551.27) (160.11, 558.29) (151.17, 558.29) /T1_2 W. <|special_separator|> -(161.71, 551.27) (196.98, 551.27) (196.98, 558.29) (161.71, 558.29) /T1_2 &Anders, <|special_separator|> -(198.58, 551.27) (205.51, 551.27) (205.51, 558.29) (198.58, 558.29) /T1_2 S. <|special_separator|> -(207.11, 551.27) (283.87, 551.27) (283.87, 558.29) (207.11, 558.29) /T1_2 Deep-learning-based <|special_separator|> -(056.69, 540.52) (075.27, 540.52) (075.27, 547.53) (056.69, 547.53) /T1_2 gene <|special_separator|> -(076.87, 540.52) (122.77, 540.52) (122.77, 547.53) (076.87, 547.53) /T1_2 perturbation <|special_separator|> -(124.37, 540.52) (145.31, 540.52) (145.31, 547.53) (124.37, 547.53) /T1_2 effect <|special_separator|> -(146.91, 540.52) (184.54, 540.52) (184.54, 547.53) (146.91, 547.53) /T1_2 prediction <|special_separator|> -(186.14, 540.52) (204.10, 540.52) (204.10, 547.53) (186.14, 547.53) /T1_2 does <|special_separator|> -(205.70, 540.52) (217.95, 540.52) (217.95, 547.53) (205.70, 547.53) /T1_2 not <|special_separator|> -(219.55, 540.52) (230.84, 540.52) (230.84, 547.53) (219.55, 547.53) /T1_2 yet <|special_separator|> -(232.44, 540.52) (274.27, 540.52) (274.27, 547.53) (232.44, 547.53) /T1_2 outperform <|special_separator|> -(056.69, 529.77) (081.24, 529.77) (081.24, 536.78) (056.69, 536.78) /T1_2 simple <|special_separator|> -(082.84, 529.77) (103.40, 529.77) (103.40, 536.78) (082.84, 536.78) /T1_2 linear <|special_separator|> -(105.00, 529.77) (141.63, 529.77) (141.63, 536.78) (105.00, 536.78) /T1_2 baselines. <|special_separator|> -(143.23, 529.77) (157.82, 529.77) (157.82, 536.78) (143.23, 536.78) /T1_3 Nat. <|special_separator|> -(159.34, 529.77) (190.76, 529.77) (190.76, 536.78) (159.34, 536.78) /T1_3 Methods <|special_separator|> -(192.32, 529.77) (203.15, 529.77) (203.15, 536.78) (192.32, 536.78) /T1_2 22, <|special_separator|> -(204.75, 529.77) (240.59, 529.77) (240.59, 536.78) (204.75, 536.78) /T1_2 1657-1661 <|special_separator|> -(242.19, 529.77) (268.03, 529.77) (268.03, 536.78) (242.19, 536.78) /T1_2 (2025). <|special_separator|> -(039.68, 519.02) (050.52, 519.02) (050.52, 526.03) (039.68, 526.03) /T1_2 22. <|special_separator|> -(056.69, 519.02) (079.84, 519.02) (079.84, 526.03) (056.69, 526.03) /T1_2 Radig, <|special_separator|> -(081.44, 519.02) (086.72, 519.02) (086.72, 526.03) (081.44, 526.03) /T1_2 J. <|special_separator|> -(088.32, 519.02) (095.60, 519.02) (095.60, 526.03) (088.32, 526.03) /T1_2 et <|special_separator|> -(097.20, 519.02) (105.91, 519.02) (105.91, 526.03) (097.20, 526.03) /T1_2 al. <|special_separator|> -(107.51, 519.02) (144.62, 519.02) (144.62, 526.03) (107.51, 526.03) /T1_2 scArchon: <|special_separator|> -(146.22, 519.02) (183.77, 519.02) (183.77, 526.03) (146.22, 526.03) /T1_2 Ascalable <|special_separator|> -(185.37, 519.02) (238.49, 519.02) (238.49, 526.03) (185.37, 526.03) /T1_2 benchmarking <|special_separator|> -(240.09, 519.02) (279.35, 519.02) (279.35, 526.03) (240.09, 526.03) /T1_2 framework <|special_separator|> -(280.94, 519.02) (291.15, 519.02) (291.15, 526.03) (280.94, 526.03) /T1_2 for <|special_separator|> -(056.69, 508.26) (091.97, 508.26) (091.97, 515.28) (056.69, 515.28) /T1_2 assessing <|special_separator|> -(093.57, 508.26) (131.56, 508.26) (131.56, 515.28) (093.57, 515.28) /T1_2 single-cell <|special_separator|> -(133.16, 508.26) (179.06, 508.26) (179.06, 515.28) (133.16, 515.28) /T1_2 perturbation <|special_separator|> -(180.66, 508.26) (210.15, 508.26) (210.15, 515.28) (180.66, 515.28) /T1_2 models. <|special_separator|> -(211.75, 508.26) (260.01, 508.26) (260.01, 515.28) (211.75, 515.28) /T1_3 GenomeBiol. <|special_separator|> -(261.68, 508.26) (288.34, 508.26) (288.34, 515.28) (261.68, 515.28) /T1_2 https:// <|special_separator|> -(056.69, 497.51) (188.56, 497.51) (188.56, 504.53) (056.69, 504.53) /T1_2 doi.org/10.1186/s13059-026-04104-z <|special_separator|> -(190.16, 497.51) (216.18, 497.51) (216.18, 504.53) (190.16, 504.53) /T1_2 (2026). <|special_separator|> -(039.68, 486.76) (050.84, 486.76) (050.84, 493.77) (039.68, 493.77) /T1_2 23. <|special_separator|> -(056.69, 486.76) (082.55, 486.76) (082.55, 493.77) (056.69, 493.77) /T1_2 Maleki, <|special_separator|> -(084.15, 486.76) (091.08, 486.76) (091.08, 493.77) (084.15, 493.77) /T1_2 S. <|special_separator|> -(092.68, 486.76) (099.96, 486.76) (099.96, 493.77) (092.68, 493.77) /T1_2 et <|special_separator|> -(101.56, 486.76) (110.27, 486.76) (110.27, 493.77) (101.56, 493.77) /T1_2 al. <|special_separator|> -(111.87, 486.76) (141.43, 486.76) (141.43, 493.77) (111.87, 493.77) /T1_2 Efficient <|special_separator|> -(143.03, 486.76) (182.90, 486.76) (182.90, 493.77) (143.03, 493.77) /T1_2 fine-tuning <|special_separator|> -(184.50, 486.76) (191.99, 486.76) (191.99, 493.77) (184.50, 493.77) /T1_2 of <|special_separator|> -(193.59, 486.76) (231.58, 486.76) (231.58, 493.77) (193.59, 493.77) /T1_2 single-cell <|special_separator|> -(233.18, 486.76) (273.02, 486.76) (273.02, 493.77) (233.18, 493.77) /T1_2 foundation <|special_separator|> -(056.69, 476.01) (084.03, 476.01) (084.03, 483.02) (056.69, 483.02) /T1_2 models <|special_separator|> -(085.63, 476.01) (114.44, 476.01) (114.44, 483.02) (085.63, 483.02) /T1_2 enables <|special_separator|> -(116.04, 476.01) (150.27, 476.01) (150.27, 483.02) (116.04, 483.02) /T1_2 zero-shot <|special_separator|> -(151.87, 476.01) (189.02, 476.01) (189.02, 483.02) (151.87, 483.02) /T1_2 molecular <|special_separator|> -(190.62, 476.01) (236.52, 476.01) (236.52, 483.02) (190.62, 483.02) /T1_2 perturbation <|special_separator|> -(238.12, 476.01) (277.90, 476.01) (277.90, 483.02) (238.12, 483.02) /T1_2 prediction. <|special_separator|> -(056.69, 465.26) (085.74, 465.26) (085.74, 472.27) (056.69, 472.27) /T1_2 Preprint <|special_separator|> -(087.34, 465.26) (094.39, 465.26) (094.39, 472.27) (087.34, 472.27) /T1_2 at <|special_separator|> -(095.99, 465.26) (206.88, 465.26) (206.88, 472.27) (095.99, 472.27) /T1_2 http://arxiv.org/abs/2412.13478 <|special_separator|> -(208.48, 465.26) (234.32, 465.26) (234.32, 472.27) (208.48, 472.27) /T1_2 (2025). <|special_separator|> -(039.68, 454.50) (050.75, 454.50) (050.75, 461.52) (039.68, 461.52) /T1_2 24. <|special_separator|> -(056.69, 454.50) (077.07, 454.50) (077.07, 461.52) (056.69, 461.52) /T1_2 Klein, <|special_separator|> -(078.67, 454.50) (088.06, 454.50) (088.06, 461.52) (078.67, 461.52) /T1_2 D., <|special_separator|> -(089.66, 454.50) (121.51, 454.50) (121.51, 461.52) (089.66, 461.52) /T1_2 Uscidda, <|special_separator|> -(123.11, 454.50) (130.95, 454.50) (130.95, 461.52) (123.11, 461.52) /T1_2 T., <|special_separator|> -(132.55, 454.50) (153.98, 454.50) (153.98, 461.52) (132.55, 461.52) /T1_2 Theis, <|special_separator|> -(155.58, 454.50) (161.38, 454.50) (161.38, 461.52) (155.58, 461.52) /T1_2 F. <|special_separator|> -(162.98, 454.50) (194.94, 454.50) (194.94, 461.52) (162.98, 461.52) /T1_2 &Cuturi, <|special_separator|> -(196.53, 454.50) (205.53, 454.50) (205.53, 461.52) (196.53, 461.52) /T1_2 M. <|special_separator|> -(207.13, 454.50) (232.08, 454.50) (232.08, 461.52) (207.13, 461.52) /T1_2 Genot: <|special_separator|> -(233.68, 454.50) (264.46, 454.50) (264.46, 461.52) (233.68, 461.52) /T1_2 entropic <|special_separator|> -(056.69, 443.75) (091.24, 443.75) (091.24, 450.77) (056.69, 450.77) /T1_2 (Gromov) <|special_separator|> -(092.84, 443.75) (136.55, 443.75) (136.55, 450.77) (092.84, 450.77) /T1_2 Wasserstein <|special_separator|> -(138.15, 443.75) (153.94, 443.75) (153.94, 450.77) (138.15, 450.77) /T1_2 flow <|special_separator|> -(155.54, 443.75) (190.44, 443.75) (190.44, 450.77) (155.54, 450.77) /T1_2 matching <|special_separator|> -(192.04, 443.75) (207.74, 443.75) (207.74, 450.77) (192.04, 450.77) /T1_2 with <|special_separator|> -(209.34, 443.75) (254.15, 443.75) (254.15, 450.77) (209.34, 450.77) /T1_2 applications <|special_separator|> -(056.69, 433.00) (064.26, 433.00) (064.26, 440.01) (056.69, 440.01) /T1_2 to <|special_separator|> -(065.86, 433.00) (103.85, 433.00) (103.85, 440.01) (065.86, 440.01) /T1_2 single-cell <|special_separator|> -(105.45, 433.00) (143.77, 433.00) (143.77, 440.01) (105.45, 440.01) /T1_2 genomics. <|special_separator|> -(145.37, 433.00) (152.27, 433.00) (152.27, 440.01) (145.37, 440.01) /T1_2 In <|special_separator|> -(153.87, 433.00) (170.53, 433.00) (170.53, 440.01) (153.87, 440.01) /T1_3 38th <|special_separator|> -(172.05, 433.00) (214.23, 433.00) (214.23, 440.01) (172.05, 440.01) /T1_3 Conference <|special_separator|> -(215.75, 433.00) (250.33, 433.00) (250.33, 440.01) (215.75, 440.01) /T1_3 onNeural <|special_separator|> -(056.69, 422.25) (098.21, 422.25) (098.21, 429.26) (056.69, 429.26) /T1_3 Information <|special_separator|> -(099.73, 422.25) (139.01, 422.25) (139.01, 429.26) (099.73, 429.26) /T1_3 Processing <|special_separator|> -(140.53, 422.25) (170.64, 422.25) (170.64, 429.26) (140.53, 429.26) /T1_3 Systems <|special_separator|> -(172.16, 422.25) (203.49, 422.25) (203.49, 429.26) (172.16, 429.26) /T1_3 (NeurIPS <|special_separator|> -(205.01, 422.25) (225.91, 422.25) (225.91, 429.26) (205.01, 429.26) /T1_3 2024) <|special_separator|> -(227.53, 422.25) (254.19, 422.25) (254.19, 429.26) (227.53, 429.26) /T1_2 https:// <|special_separator|> -(056.69, 411.50) (247.18, 411.50) (247.18, 418.51) (056.69, 418.51) /T1_2 proceedings.neurips.cc/paper_files/paper/2024/file/ <|special_separator|> -(056.69, 400.74) (278.26, 400.74) (278.26, 407.76) (056.69, 407.76) /T1_2 bc46e29f91e676747c584ca181cb0ea1-Paper-Conference.pdf <|special_separator|> -(056.69, 389.99) (082.49, 389.99) (082.49, 397.01) (056.69, 397.01) /T1_2 (2024). <|special_separator|> -(039.68, 379.24) (050.54, 379.24) (050.54, 386.25) (039.68, 386.25) /T1_2 25. <|special_separator|> -(056.69, 379.24) (082.23, 379.24) (082.23, 386.25) (056.69, 386.25) /T1_2 Bunne, <|special_separator|> -(083.83, 379.24) (093.74, 379.24) (093.74, 386.25) (083.83, 386.25) /T1_2 C., <|special_separator|> -(095.34, 379.24) (141.32, 379.24) (141.32, 386.25) (095.34, 386.25) /T1_2 Schiebinger, <|special_separator|> -(142.92, 379.24) (152.97, 379.24) (152.97, 386.25) (142.92, 386.25) /T1_2 G., <|special_separator|> -(154.57, 379.24) (181.52, 379.24) (181.52, 386.25) (154.57, 386.25) /T1_2 Krause, <|special_separator|> -(183.12, 379.24) (192.89, 379.24) (192.89, 386.25) (183.12, 386.25) /T1_2 A., <|special_separator|> -(194.49, 379.24) (218.92, 379.24) (218.92, 386.25) (194.49, 386.25) /T1_2 Regev, <|special_separator|> -(220.52, 379.24) (228.14, 379.24) (228.14, 386.25) (220.52, 386.25) /T1_2 A. <|special_separator|> -(229.74, 379.24) (261.70, 379.24) (261.70, 386.25) (229.74, 386.25) /T1_2 &Cuturi, <|special_separator|> -(263.30, 379.24) (272.30, 379.24) (272.30, 386.25) (263.30, 386.25) /T1_2 M. <|special_separator|> -(056.69, 368.49) (086.40, 368.49) (086.40, 375.50) (056.69, 375.50) /T1_2 Optimal <|special_separator|> -(088.00, 368.49) (121.66, 368.49) (121.66, 375.50) (088.00, 375.50) /T1_2 transport <|special_separator|> -(123.26, 368.49) (133.46, 368.49) (133.46, 375.50) (123.26, 375.50) /T1_2 for <|special_separator|> -(135.06, 368.49) (173.05, 368.49) (173.05, 375.50) (135.06, 375.50) /T1_2 single-cell <|special_separator|> -(174.65, 368.49) (188.57, 368.49) (188.57, 375.50) (174.65, 375.50) /T1_2 and <|special_separator|> -(190.17, 368.49) (214.26, 368.49) (214.26, 375.50) (190.17, 375.50) /T1_2 spatial <|special_separator|> -(215.86, 368.49) (240.15, 368.49) (240.15, 375.50) (215.86, 375.50) /T1_2 omics. <|special_separator|> -(241.75, 368.49) (256.34, 368.49) (256.34, 375.50) (241.75, 375.50) /T1_3 Nat. <|special_separator|> -(257.86, 368.49) (272.61, 368.49) (272.61, 375.50) (257.86, 375.50) /T1_3 Rev. <|special_separator|> -(056.69, 357.74) (088.11, 357.74) (088.11, 364.75) (056.69, 364.75) /T1_3 Methods <|special_separator|> -(089.63, 357.74) (116.88, 357.74) (116.88, 364.75) (089.63, 364.75) /T1_3 Primers <|special_separator|> -(118.47, 357.75) (123.72, 357.75) (123.72, 364.75) (118.47, 364.75) /T1_4 4 <|special_separator|> -(123.58, 357.74) (125.84, 357.74) (125.84, 364.75) (123.58, 364.75) /T1_2 , <|special_separator|> -(127.44, 357.74) (136.85, 357.74) (136.85, 364.75) (127.44, 364.75) /T1_2 58 <|special_separator|> -(138.45, 357.74) (164.25, 357.74) (164.25, 364.75) (138.45, 364.75) /T1_2 (2024). <|special_separator|> -(039.68, 346.98) (050.84, 346.98) (050.84, 354.00) (039.68, 354.00) /T1_2 26. <|special_separator|> -(056.69, 346.98) (102.68, 346.98) (102.68, 354.00) (056.69, 354.00) /T1_2 Schiebinger, <|special_separator|> -(104.28, 346.98) (112.17, 346.98) (112.17, 354.00) (104.28, 354.00) /T1_2 G. <|special_separator|> -(113.77, 346.98) (121.05, 346.98) (121.05, 354.00) (113.77, 354.00) /T1_2 et <|special_separator|> -(122.65, 346.98) (131.36, 346.98) (131.36, 354.00) (122.65, 354.00) /T1_2 al. <|special_separator|> -(132.96, 346.98) (198.79, 346.98) (198.79, 354.00) (132.96, 354.00) /T1_2 Optimal-transport <|special_separator|> -(200.39, 346.98) (229.38, 346.98) (229.38, 354.00) (200.39, 354.00) /T1_2 analysis <|special_separator|> -(230.98, 346.98) (238.48, 346.98) (238.48, 354.00) (230.98, 354.00) /T1_2 of <|special_separator|> -(240.08, 346.98) (278.07, 346.98) (278.07, 354.00) (240.08, 354.00) /T1_2 single-cell <|special_separator|> -(056.69, 336.23) (075.27, 336.23) (075.27, 343.25) (056.69, 343.25) /T1_2 gene <|special_separator|> -(076.87, 336.23) (116.41, 336.23) (116.41, 343.25) (076.87, 343.25) /T1_2 expression <|special_separator|> -(118.01, 336.23) (151.46, 336.23) (151.46, 343.25) (118.01, 343.25) /T1_2 identifies <|special_separator|> -(153.06, 336.23) (208.38, 336.23) (208.38, 343.25) (153.06, 343.25) /T1_2 developmental <|special_separator|> -(209.98, 336.23) (251.25, 336.23) (251.25, 343.25) (209.98, 343.25) /T1_2 trajectories <|special_separator|> -(252.85, 336.23) (259.57, 336.23) (259.57, 343.25) (252.85, 343.25) /T1_2 in <|special_separator|> -(056.69, 325.48) (116.48, 325.48) (116.48, 332.49) (056.69, 332.49) /T1_2 reprogramming. <|special_separator|> -(118.10, 325.48) (132.55, 325.48) (132.55, 332.49) (118.10, 332.49) /T1_3 Cell <|special_separator|> -(134.12, 325.50) (146.79, 325.50) (146.79, 332.49) (134.12, 332.49) /T1_4 176 <|special_separator|> -(146.64, 325.48) (148.90, 325.48) (148.90, 332.49) (146.64, 332.49) /T1_2 , <|special_separator|> -(150.50, 325.48) (182.79, 325.48) (182.79, 332.49) (150.50, 332.49) /T1_2 928-943 <|special_separator|> -(184.39, 325.48) (209.17, 325.48) (209.17, 332.49) (184.39, 332.49) /T1_2 (2019). <|special_separator|> -(039.68, 314.73) (049.48, 314.73) (049.48, 321.74) (039.68, 321.74) /T1_2 27. <|special_separator|> -(056.68, 314.73) (077.06, 314.73) (077.06, 321.74) (056.68, 321.74) /T1_2 Klein, <|special_separator|> -(078.66, 314.73) (085.90, 314.73) (085.90, 321.74) (078.66, 321.74) /T1_2 D. <|special_separator|> -(087.50, 314.73) (094.78, 314.73) (094.78, 321.74) (087.50, 321.74) /T1_2 et <|special_separator|> -(096.38, 314.73) (105.10, 314.73) (105.10, 321.74) (096.38, 321.74) /T1_2 al. <|special_separator|> -(106.70, 314.73) (139.11, 314.73) (139.11, 321.74) (106.70, 321.74) /T1_2 Mapping <|special_separator|> -(140.71, 314.73) (157.93, 314.73) (157.93, 321.74) (140.71, 321.74) /T1_2 cells <|special_separator|> -(159.53, 314.73) (188.80, 314.73) (188.80, 321.74) (159.53, 321.74) /T1_2 through <|special_separator|> -(190.40, 314.73) (206.79, 314.73) (206.79, 321.74) (190.40, 321.74) /T1_2 time <|special_separator|> -(208.39, 314.73) (222.31, 314.73) (222.31, 321.74) (208.39, 321.74) /T1_2 and <|special_separator|> -(223.91, 314.73) (245.77, 314.73) (245.77, 321.74) (223.91, 321.74) /T1_2 space <|special_separator|> -(247.37, 314.73) (263.07, 314.73) (263.07, 321.74) (247.37, 321.74) /T1_2 with <|special_separator|> -(264.67, 314.73) (294.49, 314.73) (294.49, 321.74) (264.67, 321.74) /T1_2 moscot. <|special_separator|> -(056.68, 303.98) (080.81, 303.98) (080.81, 310.99) (056.68, 310.99) /T1_3 Nature <|special_separator|> -(082.38, 303.99) (097.53, 303.99) (097.53, 310.99) (082.38, 310.99) /T1_4 638 <|special_separator|> -(097.40, 303.98) (099.65, 303.98) (099.65, 310.99) (097.40, 310.99) /T1_2 , <|special_separator|> -(101.25, 303.98) (139.98, 303.98) (139.98, 310.99) (101.25, 310.99) /T1_2 1065-1075 <|special_separator|> -(141.58, 303.98) (167.42, 303.98) (167.42, 310.99) (141.58, 310.99) /T1_2 (2025). <|special_separator|> -(039.68, 293.22) (050.74, 293.22) (050.74, 300.24) (039.68, 300.24) /T1_2 28. <|special_separator|> -(056.69, 293.22) (079.11, 293.22) (079.11, 300.24) (056.69, 300.24) /T1_2 Gossi, <|special_separator|> -(080.71, 293.22) (086.51, 293.22) (086.51, 300.24) (080.71, 300.24) /T1_2 F. <|special_separator|> -(088.11, 293.22) (095.39, 293.22) (095.39, 300.24) (088.11, 300.24) /T1_2 et <|special_separator|> -(096.99, 293.22) (105.70, 293.22) (105.70, 300.24) (096.99, 300.24) /T1_2 al. <|special_separator|> -(107.30, 293.22) (141.77, 293.22) (141.77, 300.24) (107.30, 300.24) /T1_2 Matching <|special_separator|> -(143.37, 293.22) (165.42, 293.22) (165.42, 300.24) (143.37, 300.24) /T1_2 single <|special_separator|> -(167.02, 293.22) (184.23, 293.22) (184.23, 300.24) (167.02, 300.24) /T1_2 cells <|special_separator|> -(185.83, 293.22) (209.59, 293.22) (209.59, 300.24) (185.83, 300.24) /T1_2 across <|special_separator|> -(211.19, 293.22) (249.39, 293.22) (249.39, 300.24) (211.19, 300.24) /T1_2 modalities <|special_separator|> -(250.99, 293.22) (266.70, 293.22) (266.70, 300.24) (250.99, 300.24) /T1_2 with <|special_separator|> -(056.69, 282.47) (097.37, 282.47) (097.37, 289.49) (056.69, 289.49) /T1_2 contrastive <|special_separator|> -(098.97, 282.47) (129.13, 282.47) (129.13, 289.49) (098.97, 289.49) /T1_2 learning <|special_separator|> -(130.73, 282.47) (144.64, 282.47) (144.64, 289.49) (130.73, 289.49) /T1_2 and <|special_separator|> -(146.24, 282.47) (174.29, 282.47) (174.29, 289.49) (146.24, 289.49) /T1_2 optimal <|special_separator|> -(175.89, 282.47) (211.70, 282.47) (211.70, 289.49) (175.89, 289.49) /T1_2 transport. <|special_separator|> -(213.30, 282.47) (231.52, 282.47) (231.52, 289.49) (213.30, 289.49) /T1_3 Brief. <|special_separator|> -(233.04, 282.47) (270.03, 282.47) (270.03, 289.49) (233.04, 289.49) /T1_3 Bioinform. <|special_separator|> -(271.70, 282.49) (281.40, 282.49) (281.40, 289.49) (271.70, 289.49) /T1_4 24 <|special_separator|> -(281.26, 282.47) (283.52, 282.47) (283.52, 289.49) (281.26, 289.49) /T1_2 , <|special_separator|> -(056.69, 271.72) (088.90, 271.72) (088.90, 278.73) (056.69, 278.73) /T1_2 bbad130 <|special_separator|> -(090.50, 271.72) (116.47, 271.72) (116.47, 278.73) (090.50, 278.73) /T1_2 (2023). <|special_separator|> -(039.68, 260.97) (050.68, 260.97) (050.68, 267.98) (039.68, 267.98) /T1_2 29. <|special_separator|> -(056.69, 260.97) (073.51, 260.97) (073.51, 267.98) (056.69, 267.98) /T1_2 Cao, <|special_separator|> -(075.11, 260.97) (084.34, 260.97) (084.34, 267.98) (075.11, 267.98) /T1_2 K., <|special_separator|> -(085.94, 260.97) (108.55, 260.97) (108.55, 267.98) (085.94, 267.98) /T1_2 Gong, <|special_separator|> -(110.15, 260.97) (120.62, 260.97) (120.62, 267.98) (110.15, 267.98) /T1_2 Q., <|special_separator|> -(122.22, 260.97) (144.53, 260.97) (144.53, 267.98) (122.22, 267.98) /T1_2 Hong, <|special_separator|> -(146.13, 260.97) (152.11, 260.97) (152.11, 267.98) (146.13, 267.98) /T1_2 Y. <|special_separator|> -(153.71, 260.97) (187.27, 260.97) (187.27, 267.98) (153.71, 267.98) /T1_2 &Wan,L. <|special_separator|> -(188.87, 260.97) (221.06, 260.97) (221.06, 267.98) (188.87, 267.98) /T1_2 Aunified <|special_separator|> -(222.66, 260.97) (276.21, 260.97) (276.21, 267.98) (222.66, 267.98) /T1_2 computational <|special_separator|> -(056.69, 250.22) (095.94, 250.22) (095.94, 257.23) (056.69, 257.23) /T1_2 framework <|special_separator|> -(097.54, 250.22) (107.75, 250.22) (107.75, 257.23) (097.54, 257.23) /T1_2 for <|special_separator|> -(109.35, 250.22) (147.34, 250.22) (147.34, 257.23) (109.35, 257.23) /T1_2 single-cell <|special_separator|> -(148.94, 250.22) (165.16, 250.22) (165.16, 257.23) (148.94, 257.23) /T1_2 data <|special_separator|> -(166.76, 250.22) (206.74, 250.22) (206.74, 257.23) (166.76, 257.23) /T1_2 integration <|special_separator|> -(208.34, 250.22) (224.04, 250.22) (224.04, 257.23) (208.34, 257.23) /T1_2 with <|special_separator|> -(225.64, 250.22) (253.69, 250.22) (253.69, 257.23) (225.64, 257.23) /T1_2 optimal <|special_separator|> -(255.29, 250.22) (291.10, 250.22) (291.10, 257.23) (255.29, 257.23) /T1_2 transport. <|special_separator|> -(056.69, 239.46) (108.64, 239.46) (108.64, 246.48) (056.69, 246.48) /T1_3 Nat.Commun. <|special_separator|> -(110.21, 239.48) (118.61, 239.48) (118.61, 246.48) (110.21, 246.48) /T1_4 13 <|special_separator|> -(118.47, 239.46) (120.73, 239.46) (120.73, 246.48) (118.47, 246.48) /T1_2 , <|special_separator|> -(122.33, 239.46) (138.59, 239.46) (138.59, 246.48) (122.33, 246.48) /T1_2 7419 <|special_separator|> -(140.19, 239.46) (165.76, 239.46) (165.76, 246.48) (140.19, 246.48) /T1_2 (2022). <|special_separator|> -(039.68, 228.71) (051.61, 228.71) (051.61, 235.73) (039.68, 235.73) /T1_2 30. <|special_separator|> -(056.69, 228.71) (082.22, 228.71) (082.22, 235.73) (056.69, 235.73) /T1_2 Bunne, <|special_separator|> -(083.82, 228.71) (091.58, 228.71) (091.58, 235.73) (083.82, 235.73) /T1_2 C. <|special_separator|> -(093.18, 228.71) (100.46, 228.71) (100.46, 235.73) (093.18, 235.73) /T1_2 et <|special_separator|> -(102.06, 228.71) (110.77, 228.71) (110.77, 235.73) (102.06, 235.73) /T1_2 al. <|special_separator|> -(112.37, 228.71) (144.38, 228.71) (144.38, 235.73) (112.37, 235.73) /T1_2 Learning <|special_separator|> -(145.98, 228.71) (183.98, 228.71) (183.98, 235.73) (145.98, 235.73) /T1_2 single-cell <|special_separator|> -(185.58, 228.71) (231.47, 228.71) (231.47, 235.73) (185.58, 235.73) /T1_2 perturbation <|special_separator|> -(233.07, 228.71) (270.46, 228.71) (270.46, 235.73) (233.07, 235.73) /T1_2 responses <|special_separator|> -(272.06, 228.71) (292.07, 228.71) (292.07, 235.73) (272.06, 235.73) /T1_2 using <|special_separator|> -(056.69, 217.96) (079.88, 217.96) (079.88, 224.97) (056.69, 224.97) /T1_2 neural <|special_separator|> -(081.48, 217.96) (109.53, 217.96) (109.53, 224.97) (081.48, 224.97) /T1_2 optimal <|special_separator|> -(111.13, 217.96) (146.94, 217.96) (146.94, 224.97) (111.13, 224.97) /T1_2 transport. <|special_separator|> -(148.54, 217.96) (163.13, 217.96) (163.13, 224.97) (148.54, 224.97) /T1_3 Nat. <|special_separator|> -(164.65, 217.96) (196.07, 217.96) (196.07, 224.97) (164.65, 224.97) /T1_3 Methods <|special_separator|> -(197.69, 217.98) (207.96, 217.98) (207.96, 224.97) (197.69, 224.97) /T1_4 20 <|special_separator|> -(207.83, 217.96) (210.08, 217.96) (210.08, 224.97) (207.83, 224.97) /T1_2 , <|special_separator|> -(211.68, 217.96) (248.88, 217.96) (248.88, 224.97) (211.68, 224.97) /T1_2 1759-1768 <|special_separator|> -(250.48, 217.96) (276.45, 217.96) (276.45, 224.97) (250.48, 224.97) /T1_2 (2023). <|special_separator|> -(039.68, 207.21) (049.56, 207.21) (049.56, 214.22) (039.68, 214.22) /T1_2 31. <|special_separator|> -(056.69, 207.21) (077.95, 207.21) (077.95, 214.22) (056.69, 214.22) /T1_2 Jiang, <|special_separator|> -(079.55, 207.21) (090.02, 207.21) (090.02, 214.22) (079.55, 214.22) /T1_2 Q., <|special_separator|> -(091.62, 207.21) (113.42, 207.21) (113.42, 214.22) (091.62, 214.22) /T1_2 Chen, <|special_separator|> -(115.02, 207.21) (124.10, 207.21) (124.10, 214.22) (115.02, 214.22) /T1_2 S., <|special_separator|> -(125.70, 207.21) (147.49, 207.21) (147.49, 214.22) (125.70, 214.22) /T1_2 Chen, <|special_separator|> -(149.09, 207.21) (156.50, 207.21) (156.50, 214.22) (149.09, 214.22) /T1_2 X. <|special_separator|> -(158.10, 207.21) (186.44, 207.21) (186.44, 214.22) (158.10, 214.22) /T1_2 &Jiang, <|special_separator|> -(188.04, 207.21) (195.14, 207.21) (195.14, 214.22) (188.04, 214.22) /T1_2 R. <|special_separator|> -(196.74, 207.21) (266.64, 207.21) (266.64, 214.22) (196.74, 214.22) /T1_2 scPRAMaccurately <|special_separator|> -(056.69, 196.46) (086.71, 196.46) (086.71, 203.47) (056.69, 203.47) /T1_2 predicts <|special_separator|> -(088.31, 196.46) (126.30, 196.46) (126.30, 203.47) (088.31, 203.47) /T1_2 single-cell <|special_separator|> -(127.90, 196.46) (146.48, 196.46) (146.48, 203.47) (127.90, 203.47) /T1_2 gene <|special_separator|> -(148.08, 196.46) (187.62, 196.46) (187.62, 203.47) (148.08, 203.47) /T1_2 expression <|special_separator|> -(189.22, 196.46) (235.12, 196.46) (235.12, 203.47) (189.22, 203.47) /T1_2 perturbation <|special_separator|> -(236.72, 196.46) (270.33, 196.46) (270.33, 203.47) (236.72, 203.47) /T1_2 response <|special_separator|> -(271.93, 196.46) (294.25, 196.46) (294.25, 203.47) (271.93, 203.47) /T1_2 based <|special_separator|> -(056.69, 185.70) (066.20, 185.70) (066.20, 192.72) (056.69, 192.72) /T1_2 on <|special_separator|> -(067.80, 185.70) (100.80, 185.70) (100.80, 192.72) (067.80, 192.72) /T1_2 attention <|special_separator|> -(102.40, 185.70) (147.15, 185.70) (147.15, 192.72) (102.40, 192.72) /T1_2 mechanism. <|special_separator|> -(148.75, 185.70) (200.24, 185.70) (200.24, 192.72) (148.75, 192.72) /T1_3 Bioinformatics <|special_separator|> -(201.87, 185.72) (212.64, 185.72) (212.64, 192.72) (201.87, 192.72) /T1_4 40 <|special_separator|> -(212.51, 185.70) (214.76, 185.70) (214.76, 192.72) (212.51, 192.72) /T1_2 , <|special_separator|> -(216.36, 185.70) (246.51, 185.70) (246.51, 192.72) (216.36, 192.72) /T1_2 btae265 <|special_separator|> -(248.11, 185.70) (273.91, 185.70) (273.91, 192.72) (248.11, 192.72) /T1_2 (2024). <|special_separator|> -(039.68, 174.95) (050.92, 174.95) (050.92, 181.97) (039.68, 181.97) /T1_2 32. <|special_separator|> -(056.69, 174.95) (088.54, 174.95) (088.54, 181.97) (056.69, 181.97) /T1_2 Uscidda, <|special_separator|> -(090.14, 174.95) (095.83, 174.95) (095.83, 181.97) (090.14, 181.97) /T1_2 T. <|special_separator|> -(097.43, 174.95) (129.38, 174.95) (129.38, 181.97) (097.43, 181.97) /T1_2 &Cuturi, <|special_separator|> -(130.98, 174.95) (139.98, 174.95) (139.98, 181.97) (130.98, 181.97) /T1_2 M. <|special_separator|> -(141.58, 174.95) (155.19, 174.95) (155.19, 181.97) (141.58, 181.97) /T1_2 The <|special_separator|> -(156.79, 174.95) (248.26, 174.95) (248.26, 181.97) (156.79, 181.97) /T1_2 MongeGap:aregularizer <|special_separator|> -(249.86, 174.95) (257.43, 174.95) (257.43, 181.97) (249.86, 181.97) /T1_2 to <|special_separator|> -(259.03, 174.95) (277.65, 174.95) (277.65, 181.97) (259.03, 181.97) /T1_2 learn <|special_separator|> -(056.69, 164.20) (065.49, 164.20) (065.49, 171.21) (056.69, 171.21) /T1_2 all <|special_separator|> -(067.09, 164.20) (100.75, 164.20) (100.75, 171.21) (067.09, 171.21) /T1_2 transport <|special_separator|> -(102.35, 164.20) (124.65, 164.20) (124.65, 171.21) (102.35, 171.21) /T1_2 maps. <|special_separator|> -(126.25, 164.20) (133.15, 164.20) (133.15, 171.21) (126.25, 171.21) /T1_2 In <|special_separator|> -(134.75, 164.20) (153.30, 164.20) (153.30, 171.21) (134.75, 171.21) /T1_3 Proc. <|special_separator|> -(154.82, 164.20) (172.21, 164.20) (172.21, 171.21) (154.82, 171.21) /T1_3 40th <|special_separator|> -(173.73, 164.20) (218.85, 164.20) (218.85, 171.21) (173.73, 171.21) /T1_3 International <|special_separator|> -(220.37, 164.20) (273.38, 164.20) (273.38, 171.21) (220.37, 171.21) /T1_3 Conferenceon <|special_separator|> -(056.69, 153.45) (087.35, 153.45) (087.35, 160.46) (056.69, 160.46) /T1_3 Machine <|special_separator|> -(088.87, 153.45) (120.28, 153.45) (120.28, 160.46) (088.87, 160.46) /T1_3 Learning <|special_separator|> -(121.84, 153.45) (137.43, 153.45) (137.43, 160.46) (121.84, 160.46) /T1_2 (eds <|special_separator|> -(139.03, 153.45) (165.98, 153.45) (165.98, 160.46) (139.03, 160.46) /T1_2 Krause, <|special_separator|> -(167.58, 153.45) (175.21, 153.45) (175.21, 160.46) (167.58, 160.46) /T1_2 A. <|special_separator|> -(176.81, 153.45) (184.09, 153.45) (184.09, 160.46) (176.81, 160.46) /T1_2 et <|special_separator|> -(185.69, 153.45) (196.87, 153.45) (196.87, 160.46) (185.69, 160.46) /T1_2 al.) <|special_separator|> -(198.47, 153.45) (248.75, 153.45) (248.75, 160.46) (198.47, 160.46) /T1_2 34709-34733 <|special_separator|> -(250.35, 153.45) (275.39, 153.45) (275.39, 160.46) (250.35, 160.46) /T1_2 (PMLR, <|special_separator|> -(056.69, 142.70) (080.11, 142.70) (080.11, 149.71) (056.69, 149.71) /T1_2 2023). <|special_separator|> -(039.68, 131.94) (051.20, 131.94) (051.20, 138.96) (039.68, 138.96) /T1_2 33. <|special_separator|> -(056.69, 131.94) (082.23, 131.94) (082.23, 138.96) (056.69, 138.96) /T1_2 Bunne, <|special_separator|> -(083.83, 131.94) (093.74, 131.94) (093.74, 138.96) (083.83, 138.96) /T1_2 C., <|special_separator|> -(095.34, 131.94) (122.29, 131.94) (122.29, 138.96) (095.34, 138.96) /T1_2 Krause, <|special_separator|> -(123.89, 131.94) (131.51, 131.94) (131.51, 138.96) (123.89, 138.96) /T1_2 A. <|special_separator|> -(133.11, 131.94) (165.06, 131.94) (165.06, 138.96) (133.11, 138.96) /T1_2 &Cuturi, <|special_separator|> -(166.66, 131.94) (175.66, 131.94) (175.66, 138.96) (166.66, 138.96) /T1_2 M. <|special_separator|> -(177.26, 131.94) (218.38, 131.94) (218.38, 138.96) (177.26, 138.96) /T1_2 Supervised <|special_separator|> -(219.98, 131.94) (248.05, 131.94) (248.05, 138.96) (219.98, 138.96) /T1_2 training <|special_separator|> -(249.65, 131.94) (257.14, 131.94) (257.14, 138.96) (249.65, 138.96) /T1_2 of <|special_separator|> -(056.69, 121.19) (098.04, 121.19) (098.04, 128.21) (056.69, 128.21) /T1_2 conditional <|special_separator|> -(099.64, 121.19) (157.64, 121.19) (157.64, 128.21) (099.64, 128.21) /T1_2 Mongemaps.In <|special_separator|> -(159.24, 121.19) (164.08, 121.19) (164.08, 128.21) (159.24, 128.21) /T1_2 3 <|special_separator|> -(163.97, 121.19) (176.12, 121.19) (176.12, 128.21) (163.97, 128.21) /T1_3 6th <|special_separator|> -(177.64, 121.19) (219.82, 121.19) (219.82, 128.21) (177.64, 128.21) /T1_3 Conference <|special_separator|> -(221.34, 121.19) (255.92, 121.19) (255.92, 128.21) (221.34, 128.21) /T1_3 onNeural <|special_separator|> -(056.69, 110.44) (098.21, 110.44) (098.21, 117.45) (056.69, 117.45) /T1_3 Information <|special_separator|> -(099.73, 110.44) (139.01, 110.44) (139.01, 117.45) (099.73, 117.45) /T1_3 Processing <|special_separator|> -(140.53, 110.44) (170.64, 110.44) (170.64, 117.45) (140.53, 117.45) /T1_3 Systems <|special_separator|> -(172.16, 110.44) (203.49, 110.44) (203.49, 117.45) (172.16, 117.45) /T1_3 (NeurIPS <|special_separator|> -(205.01, 110.44) (225.84, 110.44) (225.84, 117.45) (205.01, 117.45) /T1_3 2022) <|special_separator|> -(227.47, 110.44) (254.13, 110.44) (254.13, 117.45) (227.47, 117.45) /T1_2 https:// <|special_separator|> -(056.69, 099.69) (192.28, 099.69) (192.28, 106.70) (056.69, 106.70) /T1_2 openreview.net/pdf?id=sPNtVVUq7wi <|special_separator|> -(193.88, 099.69) (219.46, 099.69) (219.46, 106.70) (193.88, 106.70) /T1_2 (2022). <|special_separator|> -(039.68, 088.94) (051.35, 088.94) (051.35, 095.95) (039.68, 095.95) /T1_2 34. <|special_separator|> -(056.69, 088.94) (084.06, 088.94) (084.06, 095.95) (056.69, 095.95) /T1_2 Vedula, <|special_separator|> -(085.66, 088.94) (092.59, 088.94) (092.59, 095.95) (085.66, 095.95) /T1_2 S. <|special_separator|> -(094.19, 088.94) (101.47, 088.94) (101.47, 095.95) (094.19, 095.95) /T1_2 et <|special_separator|> -(103.07, 088.94) (111.78, 088.94) (111.78, 095.95) (103.07, 095.95) /T1_2 al. <|special_separator|> -(113.38, 088.94) (155.74, 088.94) (155.74, 095.95) (113.38, 095.95) /T1_2 Continuous <|special_separator|> -(157.34, 088.94) (180.76, 088.94) (180.76, 095.95) (157.34, 095.95) /T1_2 vector <|special_separator|> -(182.36, 088.94) (212.27, 088.94) (212.27, 095.95) (182.36, 095.95) /T1_2 quantile <|special_separator|> -(213.87, 088.94) (254.35, 088.94) (254.35, 095.95) (213.87, 095.95) /T1_2 regression. <|special_separator|> -(255.95, 088.94) (262.85, 088.94) (262.85, 095.95) (255.95, 095.95) /T1_2 In <|special_separator|> -(264.45, 088.94) (283.20, 088.94) (283.20, 095.95) (264.45, 095.95) /T1_3 ICML <|special_separator|> -(056.69, 078.18) (092.63, 078.18) (092.63, 085.20) (056.69, 085.20) /T1_3 Workshop <|special_separator|> -(094.15, 078.18) (153.94, 078.18) (153.94, 085.20) (094.15, 085.20) /T1_3 onNewFrontiers <|special_separator|> -(155.46, 078.18) (162.09, 078.18) (162.09, 085.20) (155.46, 085.20) /T1_3 in <|special_separator|> -(163.61, 078.18) (197.14, 078.18) (197.14, 085.20) (163.61, 085.20) /T1_3 Learning, <|special_separator|> -(198.66, 078.18) (228.13, 078.18) (228.13, 085.20) (198.66, 085.20) /T1_3 Control, <|special_separator|> -(229.65, 078.18) (243.38, 078.18) (243.38, 085.20) (229.65, 085.20) /T1_3 and <|special_separator|> -(244.90, 078.18) (282.70, 078.18) (282.70, 085.20) (244.90, 085.20) /T1_3 Dynamical <|special_separator|> -(056.69, 067.43) (086.80, 067.43) (086.80, 074.45) (056.69, 074.45) /T1_3 Systems <|special_separator|> -(088.36, 067.43) (249.32, 067.43) (249.32, 074.45) (088.36, 074.45) /T1_2 https://openreview.net/pdf?id=DUZbGAXcyL <|special_separator|> -(250.92, 067.43) (276.89, 067.43) (276.89, 074.45) (250.92, 074.45) /T1_2 (2023). <|special_separator|> -(039.68, 056.68) (050.94, 056.68) (050.94, 063.69) (039.68, 063.69) /T1_2 35. <|special_separator|> -(056.69, 056.68) (077.00, 056.68) (077.00, 063.69) (056.69, 063.69) /T1_2 Yang, <|special_separator|> -(078.60, 056.68) (085.68, 056.68) (085.68, 063.69) (078.60, 063.69) /T1_2 K. <|special_separator|> -(087.28, 056.68) (094.56, 056.68) (094.56, 063.69) (087.28, 063.69) /T1_2 et <|special_separator|> -(096.16, 056.68) (104.87, 056.68) (104.87, 063.69) (096.16, 063.69) /T1_2 al. <|special_separator|> -(106.47, 056.68) (142.42, 056.68) (142.42, 063.69) (106.47, 063.69) /T1_2 Analyzing <|special_separator|> -(144.02, 056.68) (171.99, 056.68) (171.99, 063.69) (144.02, 063.69) /T1_2 learned <|special_separator|> -(173.59, 056.68) (210.75, 056.68) (210.75, 063.69) (173.59, 063.69) /T1_2 molecular <|special_separator|> -(212.35, 056.68) (269.52, 056.68) (269.52, 063.69) (212.35, 063.69) /T1_2 representations <|special_separator|> -(271.12, 056.68) (281.33, 056.68) (281.33, 063.69) (271.12, 063.69) /T1_2 for <|special_separator|> -(056.69, 045.93) (088.39, 045.93) (088.39, 052.94) (056.69, 052.94) /T1_2 property <|special_separator|> -(089.99, 045.93) (129.77, 045.93) (129.77, 052.94) (089.99, 052.94) /T1_2 prediction. <|special_separator|> -(131.37, 045.93) (136.60, 045.93) (136.60, 052.94) (131.37, 052.94) /T1_3 J. <|special_separator|> -(138.12, 045.93) (174.57, 045.93) (174.57, 052.94) (138.12, 052.94) /T1_3 Chem.Inf. <|special_separator|> -(176.09, 045.93) (200.96, 045.93) (200.96, 052.94) (176.09, 052.94) /T1_3 Model. <|special_separator|> -(202.58, 045.94) (212.52, 045.94) (212.52, 052.94) (202.58, 052.94) /T1_4 59 <|special_separator|> -(212.39, 045.93) (214.64, 045.93) (214.64, 052.94) (212.39, 052.94) /T1_2 , <|special_separator|> -(216.24, 045.93) (257.65, 045.93) (257.65, 052.94) (216.24, 052.94) /T1_2 3370-3388 <|special_separator|> -(259.25, 045.93) (284.03, 045.93) (284.03, 052.94) (259.25, 052.94) /T1_2 (2019). <|special_separator|> -(306.15, 733.93) (317.78, 733.93) (317.78, 740.94) (306.15, 740.94) /T1_2 36. <|special_separator|> -(323.16, 733.93) (338.83, 733.93) (338.83, 740.94) (323.16, 740.94) /T1_2 Gut, <|special_separator|> -(340.43, 733.93) (350.47, 733.93) (350.47, 740.94) (340.43, 740.94) /T1_2 G., <|special_separator|> -(352.07, 733.93) (391.06, 733.93) (391.06, 740.94) (352.07, 740.94) /T1_2 Herrmann, <|special_separator|> -(392.66, 733.93) (401.66, 733.93) (401.66, 740.94) (392.66, 740.94) /T1_2 M. <|special_separator|> -(403.26, 733.93) (410.51, 733.93) (410.51, 740.94) (403.26, 740.94) /T1_2 D. <|special_separator|> -(412.11, 733.93) (456.39, 733.93) (456.39, 740.94) (412.11, 740.94) /T1_2 &Pelkmans, <|special_separator|> -(457.99, 733.93) (464.62, 733.93) (464.62, 740.94) (457.99, 740.94) /T1_2 L. <|special_separator|> -(466.22, 733.93) (509.38, 733.93) (509.38, 740.94) (466.22, 740.94) /T1_2 Multiplexed <|special_separator|> -(510.98, 733.93) (537.30, 733.93) (537.30, 740.94) (510.98, 740.94) /T1_2 protein <|special_separator|> -(323.16, 723.18) (357.88, 723.18) (357.88, 730.19) (323.16, 730.19) /T1_2 mapslink <|special_separator|> -(359.48, 723.18) (400.27, 723.18) (400.27, 730.19) (359.48, 730.19) /T1_2 subcellular <|special_separator|> -(401.87, 723.18) (447.33, 723.18) (447.33, 730.19) (401.87, 730.19) /T1_2 organization <|special_separator|> -(448.93, 723.18) (456.50, 723.18) (456.50, 730.19) (448.93, 730.19) /T1_2 to <|special_separator|> -(458.10, 723.18) (485.59, 723.18) (485.59, 730.19) (458.10, 730.19) /T1_2 cellular <|special_separator|> -(487.19, 723.18) (511.13, 723.18) (511.13, 730.19) (487.19, 730.19) /T1_2 states. <|special_separator|> -(512.73, 723.18) (541.37, 723.18) (541.37, 730.19) (512.73, 730.19) /T1_3 Science <|special_separator|> -(543.04, 723.19) (556.49, 723.19) (556.49, 730.19) (543.04, 730.19) /T1_4 361 <|special_separator|> -(556.35, 723.18) (558.60, 723.18) (558.60, 730.19) (556.35, 730.19) /T1_2 , <|special_separator|> -(323.16, 712.42) (357.36, 712.42) (357.36, 719.44) (323.16, 719.44) /T1_2 eaar7042 <|special_separator|> -(358.96, 712.42) (383.65, 712.42) (383.65, 719.44) (358.96, 719.44) /T1_2 (2018). <|special_separator|> -(306.15, 701.67) (316.23, 701.67) (316.23, 708.69) (306.15, 708.69) /T1_2 37. <|special_separator|> -(323.16, 701.67) (339.90, 701.67) (339.90, 708.69) (323.16, 708.69) /T1_2 Hao, <|special_separator|> -(341.50, 701.67) (350.50, 701.67) (350.50, 708.69) (341.50, 708.69) /T1_2 M. <|special_separator|> -(352.10, 701.67) (359.38, 701.67) (359.38, 708.69) (352.10, 708.69) /T1_2 et <|special_separator|> -(360.98, 701.67) (369.69, 701.67) (369.69, 708.69) (360.98, 708.69) /T1_2 al. <|special_separator|> -(371.29, 701.67) (413.58, 701.67) (413.58, 708.69) (371.29, 708.69) /T1_2 Large-scale <|special_separator|> -(415.18, 701.67) (455.01, 701.67) (455.01, 708.69) (415.18, 708.69) /T1_2 foundation <|special_separator|> -(456.61, 701.67) (480.21, 701.67) (480.21, 708.69) (456.61, 708.69) /T1_2 model <|special_separator|> -(481.81, 701.67) (491.32, 701.67) (491.32, 708.69) (481.81, 708.69) /T1_2 on <|special_separator|> -(492.92, 701.67) (530.92, 701.67) (530.92, 708.69) (492.92, 708.69) /T1_2 single-cell <|special_separator|> -(323.16, 690.92) (382.62, 690.92) (382.62, 697.93) (323.16, 697.93) /T1_2 transcriptomics. <|special_separator|> -(384.26, 690.92) (398.85, 690.92) (398.85, 697.93) (384.26, 697.93) /T1_3 Nat. <|special_separator|> -(400.37, 690.92) (431.78, 690.92) (431.78, 697.93) (400.37, 697.93) /T1_3 Methods <|special_separator|> -(433.36, 690.94) (441.42, 690.94) (441.42, 697.93) (433.36, 697.93) /T1_4 21 <|special_separator|> -(441.28, 690.92) (443.54, 690.92) (443.54, 697.93) (441.28, 697.93) /T1_2 , <|special_separator|> -(445.14, 690.92) (480.50, 690.92) (480.50, 697.93) (445.14, 697.93) /T1_2 1481-1491 <|special_separator|> -(482.10, 690.92) (507.90, 690.92) (507.90, 697.93) (482.10, 697.93) /T1_2 (2024). <|special_separator|> -(306.15, 680.17) (317.53, 680.17) (317.53, 687.18) (306.15, 687.18) /T1_2 38. <|special_separator|> -(323.16, 680.17) (342.56, 680.17) (342.56, 687.18) (323.16, 687.18) /T1_2 Viñas <|special_separator|> -(344.00, 680.17) (366.36, 680.17) (366.36, 687.18) (344.00, 687.18) /T1_2 Torné, <|special_separator|> -(367.80, 680.17) (374.82, 680.17) (374.82, 687.18) (367.80, 687.18) /T1_2 R. <|special_separator|> -(376.26, 680.17) (383.46, 680.17) (383.46, 687.18) (376.26, 687.18) /T1_2 et <|special_separator|> -(384.90, 680.17) (393.45, 680.17) (393.45, 687.18) (384.90, 687.18) /T1_2 al. <|special_separator|> -(394.89, 680.17) (427.93, 680.17) (427.93, 687.18) (394.89, 687.18) /T1_2 Systema: <|special_separator|> -(429.37, 680.17) (433.69, 680.17) (433.69, 687.18) (429.37, 687.18) /T1_2 a <|special_separator|> -(435.13, 680.17) (473.75, 680.17) (473.75, 687.18) (435.13, 687.18) /T1_2 framework <|special_separator|> -(475.19, 680.17) (485.23, 680.17) (485.23, 687.18) (475.19, 687.18) /T1_2 for <|special_separator|> -(486.67, 680.17) (524.17, 680.17) (524.17, 687.18) (486.67, 687.18) /T1_2 evaluating <|special_separator|> -(525.61, 680.17) (552.84, 680.17) (552.84, 687.18) (525.61, 687.18) /T1_2 genetic <|special_separator|> -(323.16, 669.42) (368.18, 669.42) (368.18, 676.43) (323.16, 676.43) /T1_2 perturbation <|special_separator|> -(369.62, 669.42) (402.66, 669.42) (402.66, 676.43) (369.62, 676.43) /T1_2 response <|special_separator|> -(404.10, 669.42) (441.00, 669.42) (441.00, 676.43) (404.10, 676.43) /T1_2 prediction <|special_separator|> -(442.44, 669.42) (509.83, 669.42) (509.83, 676.43) (442.44, 676.43) /T1_2 beyondsystematic <|special_separator|> -(511.27, 669.42) (544.29, 669.42) (544.29, 676.43) (511.27, 676.43) /T1_2 variation. <|special_separator|> -(545.73, 669.42) (560.08, 669.42) (560.08, 676.43) (545.73, 676.43) /T1_3 Nat. <|special_separator|> -(323.16, 658.66) (360.87, 658.66) (360.87, 665.68) (323.16, 665.68) /T1_3 Biotechnol <|special_separator|> -(360.65, 658.66) (362.90, 658.66) (362.90, 665.68) (360.65, 665.68) /T1_2 . <|special_separator|> -(364.36, 658.66) (518.97, 658.66) (518.97, 665.68) (364.36, 665.68) /T1_2 https://doi.org/10.1038/s41587-025-02777-8 <|special_separator|> -(520.41, 658.66) (545.77, 658.66) (545.77, 665.68) (520.41, 665.68) /T1_2 (2025). <|special_separator|> -(306.15, 647.91) (317.31, 647.91) (317.31, 654.93) (306.15, 654.93) /T1_2 39. <|special_separator|> -(323.16, 647.91) (355.83, 647.91) (355.83, 654.93) (323.16, 654.93) /T1_2 Driessen, <|special_separator|> -(357.19, 647.91) (366.73, 647.91) (366.73, 654.93) (357.19, 654.93) /T1_2 A., <|special_separator|> -(368.09, 647.91) (387.05, 647.91) (387.05, 654.93) (368.09, 654.93) /T1_2 Born, <|special_separator|> -(388.41, 647.91) (547.65, 647.91) (547.65, 654.93) (388.41, 654.93) /T1_2 J.,Rueda,R.C.,Reddy,S.T.&Rapsomaniki,M. <|special_separator|> -(323.16, 637.16) (449.99, 637.16) (449.99, 644.17) (323.16, 644.17) /T1_2 Modelingchimericantigenreceptor <|special_separator|> -(451.35, 637.16) (484.12, 637.16) (484.12, 644.17) (451.35, 644.17) /T1_2 response <|special_separator|> -(485.48, 637.16) (492.41, 637.16) (492.41, 644.17) (485.48, 644.17) /T1_2 at <|special_separator|> -(493.77, 637.16) (505.49, 637.16) (505.49, 644.17) (493.77, 644.17) /T1_2 the <|special_separator|> -(506.85, 637.16) (543.64, 637.16) (543.64, 644.17) (506.85, 644.17) /T1_2 single-cell <|special_separator|> -(323.16, 626.41) (340.15, 626.41) (340.15, 633.42) (323.16, 633.42) /T1_2 level <|special_separator|> -(341.51, 626.41) (356.86, 626.41) (356.86, 633.42) (341.51, 633.42) /T1_2 with <|special_separator|> -(358.22, 626.41) (398.37, 626.41) (398.37, 633.42) (358.22, 633.42) /T1_2 conditional <|special_separator|> -(399.73, 626.41) (427.06, 626.41) (427.06, 633.42) (399.73, 633.42) /T1_2 optimal <|special_separator|> -(428.42, 626.41) (463.15, 626.41) (463.15, 633.42) (428.42, 633.42) /T1_2 transport. <|special_separator|> -(464.51, 626.41) (509.27, 626.41) (509.27, 633.42) (464.51, 633.42) /T1_3 CellSystems <|special_separator|> -(510.59, 626.41) (534.07, 626.41) (534.07, 633.42) (510.59, 633.42) /T1_2 101591 <|special_separator|> -(535.43, 626.41) (560.73, 626.41) (560.73, 633.42) (535.43, 633.42) /T1_2 (2026). <|special_separator|> -(306.15, 615.66) (318.15, 615.66) (318.15, 622.67) (306.15, 622.67) /T1_2 40. <|special_separator|> -(323.16, 615.66) (351.99, 615.66) (351.99, 622.67) (323.16, 622.67) /T1_2 Lübeck, <|special_separator|> -(353.59, 615.66) (359.39, 615.66) (359.39, 622.67) (353.59, 622.67) /T1_2 F. <|special_separator|> -(360.99, 615.66) (368.27, 615.66) (368.27, 622.67) (360.99, 622.67) /T1_2 et <|special_separator|> -(369.87, 615.66) (378.59, 615.66) (378.59, 622.67) (369.87, 622.67) /T1_2 al. <|special_separator|> -(380.19, 615.66) (404.39, 615.66) (404.39, 622.67) (380.19, 622.67) /T1_2 Neural <|special_separator|> -(405.99, 615.66) (449.49, 615.66) (449.49, 622.67) (405.99, 622.67) /T1_2 unbalanced <|special_separator|> -(451.09, 615.66) (479.14, 615.66) (479.14, 622.67) (451.09, 622.67) /T1_2 optimal <|special_separator|> -(480.74, 615.66) (514.40, 615.66) (514.40, 622.67) (480.74, 622.67) /T1_2 transport <|special_separator|> -(516.00, 615.66) (526.43, 615.66) (526.43, 622.67) (516.00, 622.67) /T1_2 via <|special_separator|> -(323.16, 604.90) (383.26, 604.90) (383.26, 611.92) (323.16, 611.92) /T1_2 cycle-consistent <|special_separator|> -(384.86, 604.90) (443.22, 604.90) (443.22, 611.92) (384.86, 611.92) /T1_2 semi-couplings. <|special_separator|> -(444.82, 604.90) (473.86, 604.90) (473.86, 611.92) (444.82, 611.92) /T1_2 Preprint <|special_separator|> -(475.46, 604.90) (482.52, 604.90) (482.52, 611.92) (475.46, 611.92) /T1_2 at <|special_separator|> -(484.12, 604.90) (540.93, 604.90) (540.93, 611.92) (484.12, 611.92) /T1_2 http://arxiv.org/ <|special_separator|> -(323.16, 594.15) (378.70, 594.15) (378.70, 601.17) (323.16, 601.17) /T1_2 abs/2209.15621 <|special_separator|> -(380.30, 594.15) (405.87, 594.15) (405.87, 601.17) (380.30, 601.17) /T1_2 (2022). <|special_separator|> -(306.15, 583.40) (316.09, 583.40) (316.09, 590.41) (306.15, 590.41) /T1_2 41. <|special_separator|> -(323.16, 583.40) (348.32, 583.40) (348.32, 590.41) (323.16, 590.41) /T1_2 Eyring, <|special_separator|> -(349.92, 583.40) (356.55, 583.40) (356.55, 590.41) (349.92, 590.41) /T1_2 L. <|special_separator|> -(358.15, 583.40) (365.43, 583.40) (365.43, 590.41) (358.15, 590.41) /T1_2 et <|special_separator|> -(367.03, 583.40) (375.75, 583.40) (375.75, 590.41) (367.03, 590.41) /T1_2 al. <|special_separator|> -(377.35, 583.40) (438.26, 583.40) (438.26, 590.41) (377.35, 590.41) /T1_2 Unbalancedness <|special_separator|> -(439.86, 583.40) (446.58, 583.40) (446.58, 590.41) (439.86, 590.41) /T1_2 in <|special_separator|> -(448.18, 583.40) (520.33, 583.40) (520.33, 590.41) (448.18, 590.41) /T1_2 neuralMongemaps <|special_separator|> -(323.16, 572.65) (356.95, 572.65) (356.95, 579.66) (323.16, 579.66) /T1_2 improves <|special_separator|> -(358.55, 572.65) (391.19, 572.65) (391.19, 579.66) (358.55, 579.66) /T1_2 unpaired <|special_separator|> -(392.79, 572.65) (420.54, 572.65) (420.54, 579.66) (392.79, 579.66) /T1_2 domain <|special_separator|> -(422.14, 572.65) (463.17, 572.65) (463.17, 579.66) (422.14, 579.66) /T1_2 translation. <|special_separator|> -(464.77, 572.65) (471.66, 572.65) (471.66, 579.66) (464.77, 579.66) /T1_2 In <|special_separator|> -(473.26, 572.65) (491.81, 572.65) (491.81, 579.66) (473.26, 579.66) /T1_3 Proc. <|special_separator|> -(493.33, 572.65) (508.00, 572.65) (508.00, 579.66) (493.33, 579.66) /T1_3 12th <|special_separator|> -(509.52, 572.65) (554.64, 572.65) (554.64, 579.66) (509.52, 579.66) /T1_3 International <|special_separator|> -(323.16, 561.90) (365.34, 561.90) (365.34, 568.91) (323.16, 568.91) /T1_3 Conference <|special_separator|> -(366.86, 561.90) (409.10, 561.90) (409.10, 568.91) (366.86, 568.91) /T1_3 onLearning <|special_separator|> -(410.62, 561.90) (468.46, 561.90) (468.46, 568.91) (410.62, 568.91) /T1_3 Representations <|special_separator|> -(470.07, 561.90) (555.45, 561.90) (555.45, 568.91) (470.07, 568.91) /T1_2 https://openreview.net/ <|special_separator|> -(323.16, 551.14) (392.53, 551.14) (392.53, 558.16) (323.16, 558.16) /T1_2 pdf?id=2UnCj3jeao <|special_separator|> -(394.13, 551.14) (419.93, 551.14) (419.93, 558.16) (394.13, 558.16) /T1_2 (2024). <|special_separator|> -(306.15, 540.39) (317.49, 540.39) (317.49, 547.41) (306.15, 547.41) /T1_2 42. <|special_separator|> -(323.16, 540.39) (348.56, 540.39) (348.56, 547.41) (323.16, 547.41) /T1_2 Zhang, <|special_separator|> -(350.16, 540.39) (355.43, 540.39) (355.43, 547.41) (350.16, 547.41) /T1_2 J. <|special_separator|> -(357.03, 540.39) (364.31, 540.39) (364.31, 547.41) (357.03, 547.41) /T1_2 et <|special_separator|> -(365.91, 540.39) (374.62, 540.39) (374.62, 547.41) (365.91, 547.41) /T1_2 al. <|special_separator|> -(376.22, 540.39) (423.42, 540.39) (423.42, 547.41) (376.22, 547.41) /T1_2 Tahoe-100M: <|special_separator|> -(425.02, 540.39) (429.34, 540.39) (429.34, 547.41) (425.02, 547.41) /T1_2 a <|special_separator|> -(430.94, 540.39) (468.68, 540.39) (468.68, 547.41) (430.94, 547.41) /T1_2 giga-scale <|special_separator|> -(470.28, 540.39) (508.27, 540.39) (508.27, 547.41) (470.28, 547.41) /T1_2 single-cell <|special_separator|> -(509.87, 540.39) (555.77, 540.39) (555.77, 547.41) (509.87, 547.41) /T1_2 perturbation <|special_separator|> -(323.16, 529.64) (340.40, 529.64) (340.40, 536.65) (323.16, 536.65) /T1_2 atlas <|special_separator|> -(342.00, 529.64) (352.21, 529.64) (352.21, 536.65) (342.00, 536.65) /T1_2 for <|special_separator|> -(353.81, 529.64) (424.41, 529.64) (424.41, 536.65) (353.81, 536.65) /T1_2 context-dependent <|special_separator|> -(426.01, 529.64) (444.59, 529.64) (444.59, 536.65) (426.01, 536.65) /T1_2 gene <|special_separator|> -(446.19, 529.64) (476.82, 529.64) (476.82, 536.65) (446.19, 536.65) /T1_2 function <|special_separator|> -(478.42, 529.64) (492.33, 529.64) (492.33, 536.65) (478.42, 536.65) /T1_2 and <|special_separator|> -(493.93, 529.64) (521.42, 529.64) (521.42, 536.65) (493.93, 536.65) /T1_2 cellular <|special_separator|> -(523.02, 529.64) (560.30, 529.64) (560.30, 536.65) (523.02, 536.65) /T1_2 modeling. <|special_separator|> -(323.16, 518.89) (352.20, 518.89) (352.20, 525.90) (323.16, 525.90) /T1_2 Preprint <|special_separator|> -(353.80, 518.89) (360.86, 518.89) (360.86, 525.90) (353.80, 525.90) /T1_2 at <|special_separator|> -(362.46, 518.89) (388.47, 518.89) (388.47, 525.90) (362.46, 525.90) /T1_3 bioRxiv <|special_separator|> -(390.07, 518.89) (543.60, 518.89) (543.60, 525.90) (390.07, 525.90) /T1_2 https://doi.org/10.1101/2025.02.20.639398 <|special_separator|> -(323.16, 508.14) (348.99, 508.14) (348.99, 515.15) (323.16, 515.15) /T1_2 (2025). <|special_separator|> -(306.15, 497.38) (317.81, 497.38) (317.81, 504.40) (306.15, 504.40) /T1_2 43. <|special_separator|> -(323.16, 497.38) (363.53, 497.38) (363.53, 504.40) (323.16, 504.40) /T1_2 Borgwardt, <|special_separator|> -(365.13, 497.38) (372.21, 497.38) (372.21, 504.40) (365.13, 504.40) /T1_2 K. <|special_separator|> -(373.81, 497.38) (382.81, 497.38) (382.81, 504.40) (373.81, 504.40) /T1_2 M. <|special_separator|> -(384.41, 497.38) (391.69, 497.38) (391.69, 504.40) (384.41, 504.40) /T1_2 et <|special_separator|> -(393.29, 497.38) (402.00, 497.38) (402.00, 504.40) (393.29, 504.40) /T1_2 al. <|special_separator|> -(403.60, 497.38) (443.93, 497.38) (443.93, 504.40) (403.60, 504.40) /T1_2 Integrating <|special_separator|> -(445.53, 497.38) (483.64, 497.38) (483.64, 504.40) (445.53, 504.40) /T1_2 structured <|special_separator|> -(485.24, 497.38) (521.65, 497.38) (521.65, 504.40) (485.24, 504.40) /T1_2 biological <|special_separator|> -(523.25, 497.38) (539.48, 497.38) (539.48, 504.40) (523.25, 504.40) /T1_2 data <|special_separator|> -(541.08, 497.38) (550.14, 497.38) (550.14, 504.40) (541.08, 504.40) /T1_2 by <|special_separator|> -(323.16, 486.63) (345.83, 486.63) (345.83, 493.65) (323.16, 493.65) /T1_2 kernel <|special_separator|> -(347.43, 486.63) (454.16, 486.63) (454.16, 493.65) (347.43, 493.65) /T1_2 maximummeandiscrepancy. <|special_separator|> -(455.76, 486.63) (507.25, 486.63) (507.25, 493.65) (455.76, 493.65) /T1_3 Bioinformatics <|special_separator|> -(508.89, 486.65) (518.28, 486.65) (518.28, 493.65) (508.89, 493.65) /T1_4 22 <|special_separator|> -(518.15, 486.63) (520.40, 486.63) (520.40, 493.65) (518.15, 493.65) /T1_2 , <|special_separator|> -(522.00, 486.63) (553.19, 486.63) (553.19, 493.65) (522.00, 493.65) /T1_2 e49-e57 <|special_separator|> -(323.16, 475.88) (350.27, 475.88) (350.27, 482.89) (323.16, 482.89) /T1_2 (2006). <|special_separator|> -(306.16, 465.13) (318.00, 465.13) (318.00, 472.14) (306.16, 472.14) /T1_2 44. <|special_separator|> -(323.16, 465.13) (348.03, 465.13) (348.03, 472.14) (323.16, 472.14) /T1_2 Cuturi, <|special_separator|> -(349.63, 465.13) (358.63, 465.13) (358.63, 472.14) (349.63, 472.14) /T1_2 M. <|special_separator|> -(360.23, 465.13) (392.88, 465.13) (392.88, 472.14) (360.23, 472.14) /T1_2 Sinkhorn <|special_separator|> -(394.48, 465.13) (431.75, 465.13) (431.75, 472.14) (394.48, 472.14) /T1_2 distances: <|special_separator|> -(433.35, 465.13) (472.39, 465.13) (472.39, 472.14) (433.35, 472.14) /T1_2 lightspeed <|special_separator|> -(473.99, 465.13) (521.09, 465.13) (521.09, 472.14) (473.99, 472.14) /T1_2 computation <|special_separator|> -(522.69, 465.13) (530.19, 465.13) (530.19, 472.14) (522.69, 472.14) /T1_2 of <|special_separator|> -(323.16, 454.38) (351.21, 454.38) (351.21, 461.39) (323.16, 461.39) /T1_2 optimal <|special_separator|> -(352.81, 454.38) (388.63, 454.38) (388.63, 461.39) (352.81, 461.39) /T1_2 transport. <|special_separator|> -(390.23, 454.38) (397.12, 454.38) (397.12, 461.39) (390.23, 461.39) /T1_2 In <|special_separator|> -(398.72, 454.38) (417.27, 454.38) (417.27, 461.39) (398.72, 461.39) /T1_3 Proc. <|special_separator|> -(418.79, 454.38) (434.31, 454.38) (434.31, 461.39) (418.79, 461.39) /T1_3 27th <|special_separator|> -(435.83, 454.38) (480.95, 454.38) (480.95, 461.39) (435.83, 461.39) /T1_3 International <|special_separator|> -(482.47, 454.38) (524.65, 454.38) (524.65, 461.39) (482.47, 461.39) /T1_3 Conference <|special_separator|> -(323.16, 443.62) (357.74, 443.62) (357.74, 450.64) (323.16, 450.64) /T1_3 onNeural <|special_separator|> -(359.26, 443.62) (400.78, 443.62) (400.78, 450.64) (359.26, 450.64) /T1_3 Information <|special_separator|> -(402.30, 443.62) (441.58, 443.62) (441.58, 450.64) (402.30, 450.64) /T1_3 Processing <|special_separator|> -(443.10, 443.62) (473.20, 443.62) (473.20, 450.64) (443.10, 450.64) /T1_3 Systems <|special_separator|> -(474.82, 443.62) (501.48, 443.62) (501.48, 450.64) (474.82, 450.64) /T1_2 https:// <|special_separator|> -(323.16, 432.87) (512.35, 432.87) (512.35, 439.89) (323.16, 439.89) /T1_2 proceedings.neurips.cc/paper_files/paper/2013/file/ <|special_separator|> -(323.16, 422.12) (500.47, 422.12) (500.47, 429.13) (323.16, 429.13) /T1_2 af21d0c97db2e27e13572cbf59eb343d-Paper.pdf <|special_separator|> -(502.07, 422.12) (526.81, 422.12) (526.81, 429.13) (502.07, 429.13) /T1_2 (2013). <|special_separator|> -(306.16, 411.37) (317.52, 411.37) (317.52, 418.38) (306.16, 418.38) /T1_2 45. <|special_separator|> -(323.16, 411.37) (356.78, 411.37) (356.78, 418.38) (323.16, 418.38) /T1_2 Genevay, <|special_separator|> -(358.38, 411.37) (368.15, 411.37) (368.15, 418.38) (358.38, 418.38) /T1_2 A., <|special_separator|> -(369.75, 411.37) (392.16, 411.37) (392.16, 418.38) (369.75, 418.38) /T1_2 Peyre, <|special_separator|> -(393.76, 411.37) (401.65, 411.37) (401.65, 418.38) (393.76, 418.38) /T1_2 G. <|special_separator|> -(403.25, 411.37) (435.20, 411.37) (435.20, 418.38) (403.25, 418.38) /T1_2 &Cuturi, <|special_separator|> -(436.80, 411.37) (445.80, 411.37) (445.80, 418.38) (436.80, 418.38) /T1_2 M. <|special_separator|> -(447.40, 411.37) (479.41, 411.37) (479.41, 418.38) (447.40, 418.38) /T1_2 Learning <|special_separator|> -(481.01, 411.37) (519.83, 411.37) (519.83, 418.38) (481.01, 418.38) /T1_2 generative <|special_separator|> -(521.43, 411.37) (548.77, 411.37) (548.77, 418.38) (521.43, 418.38) /T1_2 models <|special_separator|> -(323.16, 400.62) (338.87, 400.62) (338.87, 407.63) (323.16, 407.63) /T1_2 with <|special_separator|> -(340.47, 400.62) (373.12, 400.62) (373.12, 407.63) (340.47, 407.63) /T1_2 Sinkhorn <|special_separator|> -(374.72, 400.62) (421.80, 400.62) (421.80, 407.63) (374.72, 407.63) /T1_2 divergences. <|special_separator|> -(423.40, 400.62) (430.30, 400.62) (430.30, 407.63) (423.40, 407.63) /T1_2 In <|special_separator|> -(431.90, 400.62) (450.45, 400.62) (450.45, 407.63) (431.90, 407.63) /T1_3 Proc. <|special_separator|> -(451.97, 400.62) (465.75, 400.62) (465.75, 407.63) (451.97, 407.63) /T1_3 21st <|special_separator|> -(467.27, 400.62) (512.38, 400.62) (512.38, 407.63) (467.27, 407.63) /T1_3 International <|special_separator|> -(513.90, 400.62) (556.09, 400.62) (556.09, 407.63) (513.90, 407.63) /T1_3 Conference <|special_separator|> -(323.16, 389.86) (363.55, 389.86) (363.55, 396.88) (323.16, 396.88) /T1_3 onArtificial <|special_separator|> -(365.07, 389.86) (407.10, 389.86) (407.10, 396.88) (365.07, 396.88) /T1_3 Intelligence <|special_separator|> -(408.62, 389.86) (422.35, 389.86) (422.35, 396.88) (408.62, 396.88) /T1_3 and <|special_separator|> -(423.87, 389.86) (456.32, 389.86) (456.32, 396.88) (423.87, 396.88) /T1_3 Statistics <|special_separator|> -(457.88, 389.86) (473.47, 389.86) (473.47, 396.88) (457.88, 396.88) /T1_2 (eds <|special_separator|> -(475.07, 389.86) (520.76, 389.86) (520.76, 396.88) (475.07, 396.88) /T1_2 Storkey,A.& <|special_separator|> -(323.16, 379.11) (364.82, 379.11) (364.82, 386.13) (323.16, 386.13) /T1_2 Perez-Cruz, <|special_separator|> -(366.42, 379.11) (374.69, 379.11) (374.69, 386.13) (366.42, 386.13) /T1_2 F.) <|special_separator|> -(376.29, 379.11) (413.38, 379.11) (413.38, 386.13) (376.29, 386.13) /T1_2 1608-1617 <|special_separator|> -(414.98, 379.11) (440.01, 379.11) (440.01, 386.13) (414.98, 386.13) /T1_2 (PMLR, <|special_separator|> -(441.61, 379.11) (463.74, 379.11) (463.74, 386.13) (441.61, 386.13) /T1_2 2018). <|special_separator|> -(306.16, 368.36) (317.90, 368.36) (317.90, 375.37) (306.16, 375.37) /T1_2 46. <|special_separator|> -(323.16, 368.36) (346.38, 368.36) (346.38, 375.37) (323.16, 375.37) /T1_2 Feydy, <|special_separator|> -(347.98, 368.36) (353.26, 368.36) (353.26, 375.37) (347.98, 375.37) /T1_2 J. <|special_separator|> -(354.86, 368.36) (362.14, 368.36) (362.14, 375.37) (354.86, 375.37) /T1_2 et <|special_separator|> -(363.74, 368.36) (372.45, 368.36) (372.45, 375.37) (363.74, 375.37) /T1_2 al. <|special_separator|> -(374.05, 368.36) (421.39, 368.36) (421.39, 375.37) (374.05, 375.37) /T1_2 Interpolating <|special_separator|> -(422.99, 368.36) (454.91, 368.36) (454.91, 375.37) (422.99, 375.37) /T1_2 between <|special_separator|> -(456.51, 368.36) (484.56, 368.36) (484.56, 375.37) (456.51, 375.37) /T1_2 optimal <|special_separator|> -(486.16, 368.36) (519.82, 368.36) (519.82, 375.37) (486.16, 375.37) /T1_2 transport <|special_separator|> -(521.42, 368.36) (535.33, 368.36) (535.33, 375.37) (521.42, 375.37) /T1_2 and <|special_separator|> -(323.16, 357.61) (398.08, 357.61) (398.08, 364.62) (323.16, 364.62) /T1_2 MMDusingSinkhorn <|special_separator|> -(399.68, 357.61) (446.77, 357.61) (446.77, 364.62) (399.68, 364.62) /T1_2 divergences. <|special_separator|> -(448.37, 357.61) (455.27, 357.61) (455.27, 364.62) (448.37, 364.62) /T1_2 In <|special_separator|> -(456.87, 357.61) (475.42, 357.61) (475.42, 364.62) (456.87, 364.62) /T1_3 Proc. <|special_separator|> -(476.94, 357.61) (495.20, 357.61) (495.20, 364.62) (476.94, 364.62) /T1_3 22nd <|special_separator|> -(496.72, 357.61) (541.84, 357.61) (541.84, 364.62) (496.72, 364.62) /T1_3 International <|special_separator|> -(323.16, 346.86) (365.35, 346.86) (365.35, 353.87) (323.16, 353.87) /T1_3 Conference <|special_separator|> -(366.87, 346.86) (407.26, 346.86) (407.26, 353.87) (366.87, 353.87) /T1_3 onArtificial <|special_separator|> -(408.78, 346.86) (450.80, 346.86) (450.80, 353.87) (408.78, 353.87) /T1_3 Intelligence <|special_separator|> -(452.32, 346.86) (466.05, 346.86) (466.05, 353.87) (452.32, 353.87) /T1_3 and <|special_separator|> -(467.57, 346.86) (500.03, 346.86) (500.03, 353.87) (467.57, 353.87) /T1_3 Statistics <|special_separator|> -(501.59, 346.86) (517.18, 346.86) (517.18, 353.87) (501.59, 353.87) /T1_2 (eds <|special_separator|> -(518.78, 346.86) (559.39, 346.86) (559.39, 353.87) (518.78, 353.87) /T1_2 Chaudhuri, <|special_separator|> -(323.16, 336.10) (330.24, 336.10) (330.24, 343.12) (323.16, 343.12) /T1_2 K. <|special_separator|> -(331.84, 336.10) (377.26, 336.10) (377.26, 343.12) (331.84, 343.12) /T1_2 &Sugiyama, <|special_separator|> -(378.86, 336.10) (390.33, 336.10) (390.33, 343.12) (378.86, 343.12) /T1_2 M.) <|special_separator|> -(391.93, 336.10) (432.10, 336.10) (432.10, 343.12) (391.93, 343.12) /T1_2 2681-2690 <|special_separator|> -(433.70, 336.10) (458.74, 336.10) (458.74, 343.12) (433.70, 343.12) /T1_2 (PMLR, <|special_separator|> -(460.34, 336.10) (482.57, 336.10) (482.57, 343.12) (460.34, 343.12) /T1_2 2019). <|special_separator|> -(306.16, 325.35) (316.39, 325.35) (316.39, 332.37) (306.16, 332.37) /T1_2 47. <|special_separator|> -(323.16, 325.35) (351.00, 325.35) (351.00, 332.37) (323.16, 332.37) /T1_2 Brenier, <|special_separator|> -(352.60, 325.35) (358.58, 325.35) (358.58, 332.37) (352.60, 332.37) /T1_2 Y. <|special_separator|> -(360.18, 325.35) (415.94, 325.35) (415.94, 332.37) (360.18, 332.37) /T1_2 Décomposition <|special_separator|> -(417.54, 325.35) (442.91, 325.35) (442.91, 332.37) (417.54, 332.37) /T1_2 polaire <|special_separator|> -(444.51, 325.35) (451.79, 325.35) (451.79, 332.37) (444.51, 332.37) /T1_2 et <|special_separator|> -(453.39, 325.35) (508.03, 325.35) (508.03, 332.37) (453.39, 332.37) /T1_2 réarrangement <|special_separator|> -(509.63, 325.35) (547.63, 325.35) (547.63, 332.37) (509.63, 332.37) /T1_2 monotone <|special_separator|> -(323.16, 314.60) (336.40, 314.60) (336.40, 321.61) (323.16, 321.61) /T1_2 des <|special_separator|> -(337.98, 314.60) (413.52, 314.60) (413.52, 321.61) (337.98, 321.61) /T1_2 champsdevecteurs. <|special_separator|> -(415.10, 314.60) (462.72, 314.60) (462.72, 321.61) (415.10, 321.61) /T1_3 CRAcad.Sci. <|special_separator|> -(464.23, 314.60) (478.01, 314.60) (478.01, 321.61) (464.23, 321.61) /T1_3 Sér. <|special_separator|> -(479.51, 314.60) (481.74, 314.60) (481.74, 321.61) (479.51, 321.61) /T1_3 I <|special_separator|> -(483.35, 314.62) (498.78, 314.62) (498.78, 321.61) (483.35, 321.61) /T1_4 305 <|special_separator|> -(498.64, 314.60) (500.90, 314.60) (500.90, 321.61) (498.64, 321.61) /T1_2 , <|special_separator|> -(502.48, 314.60) (536.00, 314.60) (536.00, 321.61) (502.48, 321.61) /T1_2 805-808 <|special_separator|> -(537.58, 314.60) (561.26, 314.60) (561.26, 321.61) (537.58, 321.61) /T1_2 (1987). <|special_separator|> -(306.16, 303.85) (317.80, 303.85) (317.80, 310.86) (306.16, 310.86) /T1_2 48. <|special_separator|> -(323.16, 303.85) (346.35, 303.85) (346.35, 310.86) (323.16, 310.86) /T1_2 Amos, <|special_separator|> -(347.95, 303.85) (357.23, 303.85) (357.23, 310.86) (347.95, 310.86) /T1_2 B., <|special_separator|> -(358.83, 303.85) (370.73, 303.85) (370.73, 310.86) (358.83, 310.86) /T1_2 Xu, <|special_separator|> -(372.33, 303.85) (378.96, 303.85) (378.96, 310.86) (372.33, 310.86) /T1_2 L. <|special_separator|> -(380.56, 303.85) (411.06, 303.85) (411.06, 310.86) (380.56, 310.86) /T1_2 &Kolter, <|special_separator|> -(412.66, 303.85) (417.94, 303.85) (417.94, 310.86) (412.66, 310.86) /T1_2 J. <|special_separator|> -(419.54, 303.85) (426.45, 303.85) (426.45, 310.86) (419.54, 310.86) /T1_2 Z. <|special_separator|> -(428.05, 303.85) (447.24, 303.85) (447.24, 310.86) (428.05, 310.86) /T1_2 Input <|special_separator|> -(448.84, 303.85) (475.23, 303.85) (475.23, 310.86) (448.84, 310.86) /T1_2 convex <|special_separator|> -(476.82, 303.85) (500.02, 303.85) (500.02, 310.86) (476.82, 310.86) /T1_2 neural <|special_separator|> -(501.62, 303.85) (537.21, 303.85) (537.21, 310.86) (501.62, 310.86) /T1_2 networks. <|special_separator|> -(538.81, 303.85) (545.71, 303.85) (545.71, 310.86) (538.81, 310.86) /T1_2 In <|special_separator|> -(323.16, 293.10) (341.71, 293.10) (341.71, 300.11) (323.16, 300.11) /T1_3 Proc. <|special_separator|> -(343.23, 293.10) (359.89, 293.10) (359.89, 300.11) (343.23, 300.11) /T1_3 34th <|special_separator|> -(361.41, 293.10) (406.53, 293.10) (406.53, 300.11) (361.41, 300.11) /T1_3 International <|special_separator|> -(408.05, 293.10) (450.23, 293.10) (450.23, 300.11) (408.05, 300.11) /T1_3 Conference <|special_separator|> -(451.75, 293.10) (526.16, 293.10) (526.16, 300.11) (451.75, 300.11) /T1_3 onMachineLearning <|special_separator|> -(527.72, 293.10) (543.31, 293.10) (543.31, 300.11) (527.72, 300.11) /T1_2 (eds <|special_separator|> -(323.16, 282.34) (351.05, 282.34) (351.05, 289.36) (323.16, 289.36) /T1_2 Precup, <|special_separator|> -(352.65, 282.34) (359.89, 282.34) (359.89, 289.36) (352.65, 289.36) /T1_2 D. <|special_separator|> -(361.49, 282.34) (383.74, 282.34) (383.74, 289.36) (361.49, 289.36) /T1_2 &Teh, <|special_separator|> -(385.34, 282.34) (391.32, 282.34) (391.32, 289.36) (385.34, 289.36) /T1_2 Y. <|special_separator|> -(392.92, 282.34) (404.34, 282.34) (404.34, 289.36) (392.92, 289.36) /T1_2 W.) <|special_separator|> -(405.94, 282.34) (434.75, 282.34) (434.75, 289.36) (405.94, 289.36) /T1_2 146-155 <|special_separator|> -(436.35, 282.34) (461.39, 282.34) (461.39, 289.36) (436.35, 289.36) /T1_2 (PMLR, <|special_separator|> -(462.99, 282.34) (484.62, 282.34) (484.62, 289.36) (462.99, 289.36) /T1_2 2017). <|special_separator|> -(306.16, 271.59) (317.66, 271.59) (317.66, 278.61) (306.16, 278.61) /T1_2 49. <|special_separator|> -(323.16, 271.59) (357.31, 271.59) (357.31, 278.61) (323.16, 278.61) /T1_2 Makkuva, <|special_separator|> -(358.91, 271.59) (368.69, 271.59) (368.69, 278.61) (358.91, 278.61) /T1_2 A., <|special_separator|> -(370.29, 271.59) (404.86, 271.59) (404.86, 278.61) (370.29, 278.61) /T1_2 Taghvaei, <|special_separator|> -(406.46, 271.59) (416.24, 271.59) (416.24, 278.61) (406.46, 278.61) /T1_2 A., <|special_separator|> -(417.84, 271.59) (431.16, 271.59) (431.16, 278.61) (417.84, 278.61) /T1_2 Oh, <|special_separator|> -(432.76, 271.59) (439.69, 271.59) (439.69, 278.61) (432.76, 278.61) /T1_2 S. <|special_separator|> -(441.29, 271.59) (463.58, 271.59) (463.58, 278.61) (441.29, 278.61) /T1_2 &Lee, <|special_separator|> -(465.18, 271.59) (470.45, 271.59) (470.45, 278.61) (465.18, 278.61) /T1_2 J. <|special_separator|> -(472.05, 271.59) (501.76, 271.59) (501.76, 278.61) (472.05, 278.61) /T1_2 Optimal <|special_separator|> -(503.36, 271.59) (537.02, 271.59) (537.02, 278.61) (503.36, 278.61) /T1_2 transport <|special_separator|> -(323.16, 260.84) (356.01, 260.84) (356.01, 267.85) (323.16, 267.85) /T1_2 mapping <|special_separator|> -(357.61, 260.84) (368.04, 260.84) (368.04, 267.85) (357.61, 267.85) /T1_2 via <|special_separator|> -(369.64, 260.84) (388.65, 260.84) (388.65, 267.85) (369.64, 267.85) /T1_2 input <|special_separator|> -(390.25, 260.84) (416.64, 260.84) (416.64, 267.85) (390.25, 267.85) /T1_2 convex <|special_separator|> -(418.24, 260.84) (441.43, 260.84) (441.43, 267.85) (418.24, 267.85) /T1_2 neural <|special_separator|> -(443.03, 260.84) (478.63, 260.84) (478.63, 267.85) (443.03, 267.85) /T1_2 networks. <|special_separator|> -(480.23, 260.84) (487.12, 260.84) (487.12, 267.85) (480.23, 267.85) /T1_2 In <|special_separator|> -(488.72, 260.84) (507.27, 260.84) (507.27, 267.85) (488.72, 267.85) /T1_3 Proc. <|special_separator|> -(508.79, 260.84) (524.60, 260.84) (524.60, 267.85) (508.79, 267.85) /T1_3 37th <|special_separator|> -(323.16, 250.09) (368.28, 250.09) (368.28, 257.10) (323.16, 257.10) /T1_3 International <|special_separator|> -(369.80, 250.09) (411.99, 250.09) (411.99, 257.10) (369.80, 257.10) /T1_3 Conference <|special_separator|> -(413.51, 250.09) (487.92, 250.09) (487.92, 257.10) (413.51, 257.10) /T1_3 onMachineLearning <|special_separator|> -(489.48, 250.09) (505.07, 250.09) (505.07, 257.10) (489.48, 257.10) /T1_2 (eds <|special_separator|> -(506.67, 250.09) (534.83, 250.09) (534.83, 257.10) (506.67, 257.10) /T1_2 Daumé, <|special_separator|> -(536.43, 250.09) (544.42, 250.09) (544.42, 257.10) (536.43, 257.10) /T1_2 H. <|special_separator|> -(546.02, 250.09) (559.55, 250.09) (559.55, 257.10) (546.02, 257.10) /T1_2 III& <|special_separator|> -(323.16, 239.34) (346.54, 239.34) (346.54, 246.35) (323.16, 246.35) /T1_2 Singh, <|special_separator|> -(348.14, 239.34) (358.24, 239.34) (358.24, 246.35) (348.14, 246.35) /T1_2 A.) <|special_separator|> -(359.84, 239.34) (399.16, 239.34) (399.16, 246.35) (359.84, 246.35) /T1_2 6672-6681 <|special_separator|> -(400.76, 239.34) (425.79, 239.34) (425.79, 246.35) (400.76, 246.35) /T1_2 (PMLR, <|special_separator|> -(427.39, 239.34) (451.39, 239.34) (451.39, 246.35) (427.39, 246.35) /T1_2 2020). <|special_separator|> -(306.16, 228.58) (317.91, 228.58) (317.91, 235.60) (306.16, 235.60) /T1_2 50. <|special_separator|> -(323.16, 228.58) (344.96, 228.58) (344.96, 235.60) (323.16, 235.60) /T1_2 Chen, <|special_separator|> -(346.56, 228.58) (354.69, 228.58) (354.69, 235.60) (346.56, 235.60) /T1_2 Y., <|special_separator|> -(356.29, 228.58) (368.93, 228.58) (368.93, 235.60) (356.29, 235.60) /T1_2 Hu, <|special_separator|> -(370.53, 228.58) (379.59, 228.58) (379.59, 235.60) (370.53, 235.60) /T1_2 Z., <|special_separator|> -(381.19, 228.58) (402.98, 228.58) (402.98, 235.60) (381.19, 235.60) /T1_2 Chen, <|special_separator|> -(404.58, 228.58) (413.52, 228.58) (413.52, 235.60) (404.58, 235.60) /T1_2 W. <|special_separator|> -(415.12, 228.58) (448.66, 228.58) (448.66, 235.60) (415.12, 235.60) /T1_2 &Huang, <|special_separator|> -(450.26, 228.58) (458.25, 228.58) (458.25, 235.60) (450.26, 235.60) /T1_2 H. <|special_separator|> -(459.85, 228.58) (474.75, 228.58) (474.75, 235.60) (459.85, 235.60) /T1_2 Fast <|special_separator|> -(476.35, 228.58) (490.26, 228.58) (490.26, 235.60) (476.35, 235.60) /T1_2 and <|special_separator|> -(491.86, 228.58) (522.42, 228.58) (522.42, 235.60) (491.86, 235.60) /T1_2 scalable <|special_separator|> -(323.16, 217.83) (372.37, 217.83) (372.37, 224.85) (323.16, 224.85) /T1_2 Wasserstein-1 <|special_separator|> -(373.97, 217.83) (397.17, 217.83) (397.17, 224.85) (373.97, 224.85) /T1_2 neural <|special_separator|> -(398.77, 217.83) (426.82, 217.83) (426.82, 224.85) (398.77, 224.85) /T1_2 optimal <|special_separator|> -(428.42, 217.83) (462.08, 217.83) (462.08, 224.85) (428.42, 224.85) /T1_2 transport <|special_separator|> -(463.68, 217.83) (485.88, 217.83) (485.88, 224.85) (463.68, 224.85) /T1_2 solver <|special_separator|> -(487.48, 217.83) (497.68, 217.83) (497.68, 224.85) (487.48, 224.85) /T1_2 for <|special_separator|> -(499.28, 217.83) (537.27, 217.83) (537.27, 224.85) (499.28, 224.85) /T1_2 single-cell <|special_separator|> -(323.16, 207.08) (369.06, 207.08) (369.06, 214.09) (323.16, 214.09) /T1_2 perturbation <|special_separator|> -(370.66, 207.08) (410.44, 207.08) (410.44, 214.09) (370.66, 214.09) /T1_2 prediction. <|special_separator|> -(412.04, 207.08) (463.53, 207.08) (463.53, 214.09) (412.04, 214.09) /T1_3 Bioinformatics <|special_separator|> -(465.16, 207.10) (473.60, 207.10) (473.60, 214.09) (465.16, 214.09) /T1_4 41 <|special_separator|> -(473.47, 207.08) (475.72, 207.08) (475.72, 214.09) (473.47, 214.09) /T1_2 , <|special_separator|> -(477.32, 207.08) (511.06, 207.08) (511.06, 214.09) (477.32, 214.09) /T1_2 i513-i522 <|special_separator|> -(512.66, 207.08) (538.50, 207.08) (538.50, 214.09) (512.66, 214.09) /T1_2 (2025). <|special_separator|> -(306.16, 196.33) (315.94, 196.33) (315.94, 203.34) (306.16, 203.34) /T1_2 51. <|special_separator|> -(323.16, 196.33) (364.86, 196.33) (364.86, 203.34) (323.16, 203.34) /T1_2 Rosenberg, <|special_separator|> -(366.46, 196.33) (374.08, 196.33) (374.08, 203.34) (366.46, 203.34) /T1_2 A. <|special_separator|> -(375.68, 196.33) (385.46, 196.33) (385.46, 203.34) (375.68, 203.34) /T1_2 A., <|special_separator|> -(387.06, 196.33) (414.43, 196.33) (414.43, 203.34) (387.06, 203.34) /T1_2 Vedula, <|special_separator|> -(416.03, 196.33) (425.11, 196.33) (425.11, 203.34) (416.03, 203.34) /T1_2 S., <|special_separator|> -(426.71, 196.33) (459.10, 196.33) (459.10, 203.34) (426.71, 203.34) /T1_2 Romano, <|special_separator|> -(460.70, 196.33) (466.68, 196.33) (466.68, 203.34) (460.70, 203.34) /T1_2 Y. <|special_separator|> -(468.27, 196.33) (512.21, 196.33) (512.21, 203.34) (468.27, 203.34) /T1_2 &Bronstein, <|special_separator|> -(513.81, 196.33) (521.44, 196.33) (521.44, 203.34) (513.81, 203.34) /T1_2 A. <|special_separator|> -(523.04, 196.33) (537.94, 196.33) (537.94, 203.34) (523.04, 203.34) /T1_2 Fast <|special_separator|> -(323.16, 185.58) (357.82, 185.58) (357.82, 192.59) (323.16, 192.59) /T1_2 nonlinear <|special_separator|> -(359.42, 185.58) (382.83, 185.58) (382.83, 192.59) (359.42, 192.59) /T1_2 vector <|special_separator|> -(384.43, 185.58) (414.35, 185.58) (414.35, 192.59) (384.43, 192.59) /T1_2 quantile <|special_separator|> -(415.95, 185.58) (456.43, 185.58) (456.43, 192.59) (415.95, 192.59) /T1_2 regression. <|special_separator|> -(458.03, 185.58) (464.92, 185.58) (464.92, 192.59) (458.03, 192.59) /T1_2 In <|special_separator|> -(466.52, 185.58) (485.07, 185.58) (485.07, 192.59) (466.52, 192.59) /T1_3 Proc. <|special_separator|> -(486.59, 185.58) (500.02, 185.58) (500.02, 192.59) (486.59, 192.59) /T1_3 11th <|special_separator|> -(501.54, 185.58) (546.66, 185.58) (546.66, 192.59) (501.54, 192.59) /T1_3 International <|special_separator|> -(323.16, 174.82) (365.35, 174.82) (365.35, 181.84) (323.16, 181.84) /T1_3 Conference <|special_separator|> -(366.87, 174.82) (409.11, 174.82) (409.11, 181.84) (366.87, 181.84) /T1_3 onLearning <|special_separator|> -(410.63, 174.82) (468.46, 174.82) (468.46, 181.84) (410.63, 181.84) /T1_3 Representations <|special_separator|> -(470.02, 174.82) (496.00, 174.82) (496.00, 181.84) (470.02, 181.84) /T1_2 (2023). <|special_separator|> -(306.16, 164.07) (317.26, 164.07) (317.26, 171.09) (306.16, 171.09) /T1_2 52. <|special_separator|> -(323.16, 164.07) (358.33, 164.07) (358.33, 171.09) (323.16, 171.09) /T1_2 Pegoraro, <|special_separator|> -(359.93, 164.07) (368.93, 164.07) (368.93, 171.09) (359.93, 171.09) /T1_2 M. <|special_separator|> -(370.53, 164.07) (377.81, 164.07) (377.81, 171.09) (370.53, 171.09) /T1_2 et <|special_separator|> -(379.41, 164.07) (388.12, 164.07) (388.12, 171.09) (379.41, 171.09) /T1_2 al. <|special_separator|> -(389.72, 164.07) (413.74, 164.07) (413.74, 171.09) (389.72, 171.09) /T1_2 Vector <|special_separator|> -(415.34, 164.07) (445.25, 164.07) (445.25, 171.09) (415.34, 171.09) /T1_2 quantile <|special_separator|> -(446.85, 164.07) (485.18, 164.07) (485.18, 171.09) (446.85, 171.09) /T1_2 regression <|special_separator|> -(486.78, 164.07) (496.29, 164.07) (496.29, 171.09) (486.78, 171.09) /T1_2 on <|special_separator|> -(497.89, 164.07) (536.24, 164.07) (536.24, 171.09) (497.89, 171.09) /T1_2 manifolds. <|special_separator|> -(537.84, 164.07) (544.73, 164.07) (544.73, 171.09) (537.84, 171.09) /T1_2 In <|special_separator|> -(323.16, 153.32) (341.71, 153.32) (341.71, 160.33) (323.16, 160.33) /T1_3 Proc. <|special_separator|> -(343.23, 153.32) (358.76, 153.32) (358.76, 160.33) (343.23, 160.33) /T1_3 27th <|special_separator|> -(360.28, 153.32) (405.39, 153.32) (405.39, 160.33) (360.28, 160.33) /T1_3 International <|special_separator|> -(406.91, 153.32) (449.10, 153.32) (449.10, 160.33) (406.91, 160.33) /T1_3 Conference <|special_separator|> -(450.62, 153.32) (491.01, 153.32) (491.01, 160.33) (450.62, 160.33) /T1_3 onArtificial <|special_separator|> -(492.53, 153.32) (534.56, 153.32) (534.56, 160.33) (492.53, 160.33) /T1_3 Intelligence <|special_separator|> -(536.08, 153.32) (549.80, 153.32) (549.80, 160.33) (536.08, 160.33) /T1_3 and <|special_separator|> -(323.16, 142.57) (355.62, 142.57) (355.62, 149.58) (323.16, 149.58) /T1_3 Statistics <|special_separator|> -(357.18, 142.57) (372.77, 142.57) (372.77, 149.58) (357.18, 149.58) /T1_2 (eds <|special_separator|> -(374.37, 142.57) (411.50, 142.57) (411.50, 149.58) (374.37, 149.58) /T1_2 Dasgupta, <|special_separator|> -(413.10, 142.57) (422.19, 142.57) (422.19, 149.58) (413.10, 149.58) /T1_2 S., <|special_separator|> -(423.79, 142.57) (449.33, 142.57) (449.33, 149.58) (423.79, 149.58) /T1_2 Mandt, <|special_separator|> -(450.93, 142.57) (457.86, 142.57) (457.86, 149.58) (450.93, 149.58) /T1_2 S. <|special_separator|> -(459.46, 142.57) (474.92, 142.57) (474.92, 149.58) (459.46, 149.58) /T1_2 &Li, <|special_separator|> -(476.52, 142.57) (484.97, 142.57) (484.97, 149.58) (476.52, 149.58) /T1_2 Y.) <|special_separator|> -(486.57, 142.57) (526.84, 142.57) (526.84, 149.58) (486.57, 149.58) /T1_2 1999-2007 <|special_separator|> -(528.44, 142.57) (553.48, 142.57) (553.48, 149.58) (528.44, 149.58) /T1_2 (PMLR, <|special_separator|> -(323.16, 131.82) (346.41, 131.82) (346.41, 138.83) (323.16, 138.83) /T1_2 2024). <|special_separator|> -(306.16, 121.06) (317.58, 121.06) (317.58, 128.08) (306.16, 128.08) /T1_2 53. <|special_separator|> -(323.16, 121.06) (343.53, 121.06) (343.53, 128.08) (323.16, 128.08) /T1_2 Tong, <|special_separator|> -(345.13, 121.06) (352.75, 121.06) (352.75, 128.08) (345.13, 128.08) /T1_2 A. <|special_separator|> -(354.35, 121.06) (361.63, 121.06) (361.63, 128.08) (354.35, 128.08) /T1_2 et <|special_separator|> -(363.23, 121.06) (371.95, 121.06) (371.95, 128.08) (363.23, 128.08) /T1_2 al. <|special_separator|> -(373.55, 121.06) (410.94, 121.06) (410.94, 128.08) (373.55, 128.08) /T1_2 Improving <|special_separator|> -(412.54, 121.06) (426.46, 121.06) (426.46, 128.08) (412.54, 128.08) /T1_2 and <|special_separator|> -(428.06, 121.06) (473.18, 121.06) (473.18, 128.08) (428.06, 128.08) /T1_2 generalizing <|special_separator|> -(474.78, 121.06) (515.19, 121.06) (515.19, 128.08) (474.78, 128.08) /T1_2 flow-based <|special_separator|> -(516.79, 121.06) (555.61, 121.06) (555.61, 128.08) (516.79, 128.08) /T1_2 generative <|special_separator|> -(323.16, 110.31) (350.50, 110.31) (350.50, 117.33) (323.16, 117.33) /T1_2 models <|special_separator|> -(352.10, 110.31) (367.80, 110.31) (367.80, 117.33) (352.10, 117.33) /T1_2 with <|special_separator|> -(369.40, 110.31) (406.24, 110.31) (406.24, 117.33) (369.40, 117.33) /T1_2 minibatch <|special_separator|> -(407.84, 110.31) (435.88, 110.31) (435.88, 117.33) (407.84, 117.33) /T1_2 optimal <|special_separator|> -(437.48, 110.31) (473.30, 110.31) (473.30, 117.33) (437.48, 117.33) /T1_2 transport. <|special_separator|> -(474.90, 110.31) (481.79, 110.31) (481.79, 117.33) (474.90, 117.33) /T1_2 In <|special_separator|> -(483.39, 110.31) (501.94, 110.31) (501.94, 117.33) (483.39, 117.33) /T1_3 Proc. <|special_separator|> -(503.59, 110.31) (559.80, 110.31) (559.80, 117.33) (503.59, 117.33) /T1_3 ICMLWorkshop <|special_separator|> -(323.16, 099.56) (382.96, 099.56) (382.96, 106.57) (323.16, 106.57) /T1_3 onNewFrontiers <|special_separator|> -(384.48, 099.56) (391.11, 099.56) (391.11, 106.57) (384.48, 106.57) /T1_3 in <|special_separator|> -(392.63, 099.56) (426.16, 099.56) (426.16, 106.57) (392.63, 106.57) /T1_3 Learning, <|special_separator|> -(427.68, 099.56) (457.15, 099.56) (457.15, 106.57) (427.68, 106.57) /T1_3 Control, <|special_separator|> -(458.67, 099.56) (472.40, 099.56) (472.40, 106.57) (458.67, 106.57) /T1_3 and <|special_separator|> -(473.92, 099.56) (511.72, 099.56) (511.72, 106.57) (473.92, 106.57) /T1_3 Dynamical <|special_separator|> -(513.24, 099.56) (543.34, 099.56) (543.34, 106.57) (513.24, 106.57) /T1_3 Systems <|special_separator|> -(323.16, 088.81) (486.28, 088.81) (486.28, 095.82) (323.16, 095.82) /T1_2 https://openreview.net/pdf?id=CD9Snc73AW <|special_separator|> -(487.88, 088.81) (513.85, 088.81) (513.85, 095.82) (487.88, 095.82) /T1_2 (2023). <|special_separator|> -(306.16, 078.06) (317.80, 078.06) (317.80, 085.07) (306.16, 085.07) /T1_2 54. <|special_separator|> -(323.16, 078.06) (362.22, 078.06) (362.22, 085.07) (323.16, 085.07) /T1_2 Pooladian, <|special_separator|> -(363.98, 078.06) (381.78, 078.06) (381.78, 085.07) (363.98, 085.07) /T1_2 A.-A. <|special_separator|> -(383.54, 078.06) (390.90, 078.06) (390.90, 085.07) (383.54, 085.07) /T1_2 et <|special_separator|> -(392.66, 078.06) (401.54, 078.06) (401.54, 085.07) (392.66, 085.07) /T1_2 al. <|special_separator|> -(403.30, 078.06) (449.22, 078.06) (449.22, 085.07) (403.30, 085.07) /T1_2 Multisample <|special_separator|> -(450.98, 078.06) (467.01, 078.06) (467.01, 085.07) (450.98, 085.07) /T1_2 flow <|special_separator|> -(468.77, 078.06) (506.54, 078.06) (506.54, 085.07) (468.77, 085.07) /T1_2 matching: <|special_separator|> -(508.30, 078.06) (557.83, 078.06) (557.83, 085.07) (508.30, 085.07) /T1_2 straightening <|special_separator|> -(323.16, 067.30) (343.01, 067.30) (343.01, 074.32) (323.16, 074.32) /T1_2 flows <|special_separator|> -(344.77, 067.30) (360.71, 067.30) (360.71, 074.32) (344.77, 074.32) /T1_2 with <|special_separator|> -(362.47, 067.30) (399.94, 067.30) (399.94, 074.32) (362.47, 074.32) /T1_2 minibatch <|special_separator|> -(401.70, 067.30) (440.93, 067.30) (440.93, 074.32) (401.70, 074.32) /T1_2 couplings. <|special_separator|> -(442.69, 067.30) (449.66, 067.30) (449.66, 074.32) (442.69, 074.32) /T1_2 In <|special_separator|> -(451.42, 067.30) (470.45, 067.30) (470.45, 074.32) (451.42, 074.32) /T1_3 Proc. <|special_separator|> -(472.21, 067.30) (489.97, 067.30) (489.97, 074.32) (472.21, 074.32) /T1_3 40th <|special_separator|> -(491.73, 067.30) (538.28, 067.30) (538.28, 074.32) (491.73, 074.32) /T1_3 International <|special_separator|> -(323.16, 056.55) (366.43, 056.55) (366.43, 063.57) (323.16, 063.57) /T1_3 Conference <|special_separator|> -(368.19, 056.55) (377.61, 056.55) (377.61, 063.57) (368.19, 063.57) /T1_3 on <|special_separator|> -(379.37, 056.55) (410.75, 056.55) (410.75, 063.57) (379.37, 063.57) /T1_3 Machine <|special_separator|> -(412.51, 056.55) (444.76, 056.55) (444.76, 063.57) (412.51, 063.57) /T1_3 Learning <|special_separator|> -(446.52, 056.55) (462.35, 056.55) (462.35, 063.57) (446.52, 063.57) /T1_2 (eds <|special_separator|> -(464.11, 056.55) (491.54, 056.55) (491.54, 063.57) (464.11, 063.57) /T1_2 Krause, <|special_separator|> -(493.30, 056.55) (501.01, 056.55) (501.01, 063.57) (493.30, 063.57) /T1_2 A. <|special_separator|> -(502.77, 056.55) (510.13, 056.55) (510.13, 063.57) (502.77, 063.57) /T1_2 et <|special_separator|> -(511.89, 056.55) (521.08, 056.55) (521.08, 063.57) (511.89, 063.57) /T1_2 al) <|special_separator|> -(323.16, 045.80) (371.27, 045.80) (371.27, 052.81) (323.16, 052.81) /T1_2 28100-28127 <|special_separator|> -(373.03, 045.80) (398.47, 045.80) (398.47, 052.81) (373.03, 052.81) /T1_2 (PMLR, <|special_separator|> -(400.23, 045.80) (424.05, 045.80) (424.05, 052.81) (400.23, 052.81) /T1_2 2023). \ No newline at end of file