The paper proposes CT, a three-sample nonparametric Mann–Whitney test for detecting generative-model data-copying.

Casey Meehan, Kamalika Chaudhuri, Sanjoy Dasgupta

Proposes FLOPs-regularized high-dimensional sparse embeddings, achieving 10x speedup with comparable accuracy.

Biswajit Paria, Chih-Kuan Yeh, Ian E. H. Yen et al.

CURL integrates contrastive learning with off-policy RL, boosting sample efficiency by nearly 2x in DM control and 1.2x in Atari.

Aravind Srinivas, Michael Laskin, Pieter Abbeel

Using symmetry and bifurcation theory, the paper derives power series expansions of critical points in shallow ReLU networks, revealing different loss decay behaviors of spurious minima.

Yossi Arjevani, Michael Field

Proposes a reinforcement learning curriculum framework using DAGs, classifies methods, emphasizes task sequencing and transfer mechanisms.

Sanmit Narvekar, Bei Peng, Matteo Leonetti et al.

AutoML-Zero uses evolutionary search to discover complete ML algorithms from basic math operations, surpassing traditional AutoML limitations.

Esteban Real, Chen Liang, David R. So et al.

AutoAttack combines Auto-PGD, FAB, Square Attack for parameter-free, automated robustness evaluation, revealing vulnerabilities missed by prior methods.

Francesco Croce, Matthias Hein

Proposed CASTER method enhances meta-reinforcement learning efficiency, reducing sample requirements.

Haozhe Wang, Jiale Zhou, Xuming He

Proposes a decoder-free predictive coding approach for locally-linear control, improving high-dimensional observation control by maximizing latent mutual information.

Rui Shu, Tung Nguyen, Yinlam Chow et al.

Proposed a modular benchmarking framework for GNNs with diverse datasets and Laplace positional encoding, advancing model evaluation.

Vijay Prakash Dwivedi, Chaitanya K. Joshi, Anh Tuan Luu et al.

Training only BatchNorm γ/β yields 82% on CIFAR-10 and 32% top-5 on ImageNet in deep ResNets.

Jonathan Frankle, David J. Schwab, Ari S. Morcos

Proposes GP-MRO algorithm combining Gaussian processes and online learning to optimize robust mixed strategies, enhancing autonomous vehicle trajectory planning.

Pier Giuseppe Sessa, Ilija Bogunovic, Maryam Kamgarpour et al.

Proposes a knowledge distillation framework with outlier rejection and multi-task learning to improve regression accuracy under noisy labels.

Makoto Takamoto, Yusuke Morishita, Hitoshi Imaoka

Sparse Sinkhorn Attention employs differentiable sorting for efficient attention, significantly reducing memory usage.

Yi Tay, Dara Bahri, Liu Yang et al.

Path-based message passing GNN incorporating path features (angles, dihedral angles) significantly improves molecular property prediction, achieving MAE of 8.70×10^-3 eV on QM8.

Daniel Flam-Shepherd, Tony Wu, Pascal Friederich et al.

Introduced Predictive Sampling, reducing ARM inference calls by 96.7% and achieving 27.6x speedup on MNIST using fixed-point iteration.

Auke Wiggers, Emiel Hoogeboom

Proposes Bayesian deep learning with marginalization, using deep ensembles and MultiSWAG to improve accuracy, calibration, and mitigate double descent.

Andrew Gordon Wilson, Pavel Izmailov

TracIn estimates training sample influence via gradient inner products and checkpoints, applicable to any SGD-trained model.

Garima Pruthi, Frederick Liu, Mukund Sundararajan et al.

GraSP prunes 80% weights at initialization, with only 1.6% accuracy drop, by preserving gradient flow.

Chaoqi Wang, Guodong Zhang, Roger Grosse

Conditioned normalizing flow-based multivariate probabilistic forecasting, outperforming state-of-the-art on thousands of interacting time series.

Kashif Rasul, Abdul-Saboor Sheikh, Ingmar Schuster et al.