Thu, Jun 29 09:00 - 18:25 |
(1) |
09:00-09:25 |
|
(2) |
09:25-09:50 |
|
(3) |
09:50-10:15 |
|
(4) |
10:15-10:40 |
|
(5) |
10:40-11:05 |
|
(6) |
11:05-11:20 |
|
|
11:20-13:30 |
Lunch Break ( 130 min. ) |
(7) |
13:30-13:55 |
|
(8) |
13:55-14:20 |
|
(9) |
14:20-14:45 |
|
(10) |
14:45-15:10 |
|
(11) |
15:10-15:35 |
|
(12) |
15:35-15:50 |
|
(13) |
15:50-16:05 |
|
|
16:05-16:15 |
Break ( 10 min. ) |
(14) |
16:15-16:40 |
|
(15) |
16:40-17:05 |
|
(16) |
17:05-17:30 |
|
(17) |
17:30-17:55 |
|
(18) |
17:55-18:10 |
|
(19) |
18:10-18:25 |
|
Fri, Jun 30 AM 09:00 - 11:20 |
(20) |
09:00-09:25 |
|
(21) |
09:25-09:50 |
|
(22) |
09:50-10:15 |
|
(23) |
10:15-10:40 |
|
(24) |
10:40-11:05 |
|
(25) |
11:05-11:20 |
|
Fri, Jun 30 PM 13:30 - 15:40 |
(26) |
13:30-13:55 |
|
(27) |
13:55-14:20 |
|
(28) |
14:20-14:45 |
|
(29) |
14:45-15:10 |
|
(30) |
15:10-15:25 |
|
(31) |
15:25-15:40 |
|
Fri, Jun 30 PM 16:00 - 17:00 |
|
- |
|
Fri, Jun 30 PM 17:00 - 18:00 |
|
- |
Speaker:Han Zhao
Title:Fair and Optimal Prediction via Post-Processing
Abstract:To mitigate the bias exhibited by machine learning models, fairness criteria can be integrated into the training process to ensure fair treatment across all demographics, but it often comes at the expense of model performance. Understanding such tradeoffs, therefore, underlies the design of fair algorithms. In this talk, I will first discuss our recent work on proving the tradeoff between fairness and accuracy in both classification and regression problems, where we show that the cost of fairness could be characterized by the optimal value of a Wasserstein-barycenter problem. Then I will show that the complexity of learning the optimal fair predictor is the same as learning the Bayes predictor, and present a post-processing algorithm based on the solution to the Wasserstein-barycenter problem that derives the optimal fair predictors from Bayes score functions. I will also present the empirical results of our fair algorithm and conclude the talk with some discussion on the close interplay between algorithmic fairness and domain generalization. |
Sat, Jul 1 AM 09:00 - 16:05 |
(32) |
09:00-09:15 |
|
(33) |
09:15-09:30 |
|
(34) |
09:30-09:45 |
|
(35) |
09:45-10:10 |
|
(36) |
10:10-10:35 |
|
(37) |
10:35-11:00 |
|
(38) |
11:00-11:25 |
|
|
11:25-13:30 |
Lunch Break ( 125 min. ) |
(39) |
13:30-13:55 |
|
(40) |
13:55-14:20 |
|
(41) |
14:20-14:45 |
|
(42) |
14:45-15:00 |
|
(43) |
15:00-15:25 |
|
(44) |
15:25-15:40 |
|
(45) |
15:40-16:05 |
|
Thu, Jun 29 13:30 - 18:05 |
(46) IBISML |
13:30-13:55 |
Selective Inference for a Combination of Feature Selection Algorithms |
Tatsuya Matsukawa (Nagoya Univ.), Daiki Miwa (NITech), Vo Nguyen Le Duy (RIKEN), Koichi Taji (Nagoya Univ.), Ichiro Takeuchi (Nagoya Univ./RIKEN) |
(47) IBISML |
13:55-14:20 |
|
|
(48) IBISML |
14:20-14:45 |
Crystal structure X-ray absorption spectrum prediction and valence ratio estimation based on Gaussian process regression |
Takumi Iwashita, Haruki Hirai, Ryo Kobayashi, Tomoyuki Tamura, Masayuki Karasuyama (NIT) |
(49) IBISML |
14:45-15:10 |
Multiple Instance Learning with major class |
Kaito Shiku, Shinnosuke Matsuo (Kyushu Univ), Daiki Suehiro (YCU), Bise Ryoma (Kyushu Univ) |
(50) IBISML |
15:10-15:35 |
Selective Inference for DNN-driven Saliency Map |
Daiki Miwa (NITech), Vo Nguyen Le Duy (RIKEN), Tomohiro Shiraishi (Nagoya Univ.), Ichiro Takeuchi (Nagoya Univ./RIKEN) |
|
15:35-16:00 |
Break ( 25 min. ) |
(51) IBISML |
16:00-16:25 |
|
Kotaro Nagata, Kazuhiro Hotta (Meijo Univ) |
(52) IBISML |
16:25-16:50 |
Minorization-Maximization for Determinantal Point Processes |
Takahiro Kawashima (SOKENDAI), Hideitsu Hino (ISM/RIKEN) |
(53) IBISML |
16:50-17:15 |
Fast Regularized Discrete Optimal Transport with Group-Sparse Regularizers |
Yasutoshi Ida, Sekitoshi Kanai, Kazuki Adachi, Atsutoshi Kumagai, Yasuhiro Fujiwara (NTT) |
(54) IBISML |
17:15-17:40 |
|
Satoshi Kamiya (Meijo Univ.), Taka-aki Tsunoyama, Akihiro Kusumi (OIST), Kazuhiro Hotta (Meijo Univ.) |
(55) IBISML |
17:40-18:05 |
Potential of Domain-agnostic Encoder for Long-range DNA Sequences |
Naoki Kobayashi, Rintaro Saito, Satoru Morimoto, Tadashi Okoshi, Jin Nakazawa (Keio) |
Fri, Jun 30 09:30 - 15:35 |
(56) IBISML |
09:30-09:55 |
Lipschitz bandits in unbounded metric spaces and their applications |
Takanobu Hara (Hokkaido Univ.) |
(57) IBISML |
09:55-10:20 |
|
|
(58) IBISML |
10:20-10:45 |
Hierarchical Classification Transformer for Facial Expression Recognition Utilizing Basic Emotion Relationships |
Ryo Miyoshi, Shuichi Akizuki (Chukyo Univ.), Kensuke Tobitani (University of Nagasaki), Noriko Nagata (Kwansei Gakuin Univ.), Manabu Hashimoto (Chukyo Univ.) |
(59) IBISML |
10:45-11:10 |
Beyond Exponential Graph: Communication-Efficient Topologies for Decentralized Learning via Finite-time Convergence |
Yuki Takezawa, Ryoma Sato, Han Bao (Kyoto Univ./OIST), Kenta Niwa (NTT CS Lab.), Makoto Yamada (OIST) |
(60) IBISML |
11:10-11:35 |
On performance degradation of a method by minimizing the conditional mutual information for the out-of-distribution generalization |
Genki Takahashi, Toshiyuki Tanaka (Kyoto University) |
|
11:35-13:30 |
Lunch Break ( 115 min. ) |
(61) IBISML |
13:30-13:55 |
Predictive graph mining for attributed graph data through proximal gradient pruning |
Ren Sugihara, Shinji Tajima, Ryota Kitahara, Masayuki Karasuyama (NIT) |
(62) IBISML |
13:55-14:20 |
Exploring Regioselective Catalysts with Hierarchical Bandits |
Hongyuan Guo, Koji Tabata, Yoshihiro Matsumura, Tamiki Komatsuzaki (Hokkaido Univ.) |
(63) IBISML |
14:20-14:45 |
Analogy Tasks in BioConceptVec using Biological Pathways |
Hiroaki Yamagiwa, Ryoma Hashimoto (Kyoto Univ.), Kiwamu Arakane, Ken Murakami (IPR), Momose Oyama, Hidetoshi Shimodaira (Kyoto Univ.), Mariko Okada (IPR) |
(64) IBISML |
14:45-15:10 |
Diffusion model with MASKed input for generating gestures during dyadic conversation |
Yuya Okadome (TUS), Yutaka Nakamura (Riken) |
(65) IBISML |
15:10-15:35 |
Analysis of Mode Connectivity Between Models with Different Hidden Layer Widths |
Yusuke Takase, Hidetoshi Shimodaira (Kyoto Univ.) |
Sat, Jul 1 09:30 - 11:10 |
(66) NC |
09:30-09:55 |
Input interactions in hippocampal dentate gyrus granule cell dendrites |
Tadanobu Kamijo (Ryukyudai), Naoki Nakajima (Kyushujohodai), Takeshi Aihara (Tamagawa Univ.) |
(67) NC |
09:55-10:20 |
Analysis of nonequilibrium neural spiking activity using a state-space kinetic Ising model |
Ken Ishihara (Hokkaido Univ.), Hideaki Shimazaki (Kyoto Univ.) |
(68) NC |
10:20-10:45 |
Solving the Traveling Salesman Problem Using Oscillator Interaction |
Tomoaki Kinugasa (Tohoku Univ), Futo Ono, Kazuhiro Sakamoto (TMPU) |
(69) NC |
10:45-11:10 |
A Study on generating First-Person Video in moving daily space using variational auto encoders |
Xu Chenfei (Osaka Univ.), Okadome Yuya (TUS), Ishiguro Hiroshi (Osaka Univ.), Nakamura Yutaka (RIKEN) |