Doubly-stochastic Gaussian kernel normalization automatically corrects heteroskedastic noise, ensuring robust high-dimensional affinity matrices.

Boris Landa, Ronald R. Coifman, Yuval Kluger

Proposes sub-band knowledge distillation, using expert sub-models to improve speech enhancement, outperforming full-band models with fewer parameters.

Xiang Hao, Shixue Wen, Xiangdong Su et al.

Scaling up to 175B parameters, GPT-3 achieves remarkable few-shot and zero-shot NLP performance, surpassing many fine-tuned models.

Tom B. Brown, Benjamin Mann, Nick Ryder et al.

The study shows that by considering user interaction time, RecentPop and DecayPop methods can improve recommendation accuracy by over 70% on the MovieLens dataset.

Yitong Ji, Aixin Sun, Jie Zhang et al.

This paper advances Orthogonal Monte Carlo (OMC) theory by applying negative dependence, deriving exponential error bounds, and introduces Near-Orthogonal Monte Carlo (NOMC) for superior high-dimensional sampling.

Han Lin, Haoxian Chen, Tianyi Zhang et al.

MOPO optimizes offline policies using dynamics uncertainty penalties, excelling on D4RL benchmarks.

Tianhe Yu, Garrett Thomas, Lantao Yu et al.

Propose a kernel-based method to approximate the Koopman generator and Schrödinger operator, applied to molecular dynamics and quantum chemistry.

Stefan Klus, Feliks Nüske, Boumediene Hamzi

Combining learned data abstraction, nonlinear filtering, and time-optimal trajectory planning, the autonomous drone racing system achieved second place at the 2019 AlphaPilot Challenge, reaching speeds up to 8 m/s.

Philipp Foehn, Dario Brescianini, Elia Kaufmann et al.

This study analyzes implementation details of PPO and TRPO, revealing that code optimizations significantly boost performance.

Logan Engstrom, Andrew Ilyas, Shibani Santurkar et al.

Optimizing alignment and uniformity in contrastive loss enhances representation learning performance.

Tongzhou Wang, Phillip Isola

Proposes a gradient ascent-based optimal sampling strategy for spectral bandit best-arm identification under graph smoothness constraints, achieving asymptotic optimality.

Tomáš Kocák, Aurélien Garivier

This study compares neural collaborative filtering (MLP) with matrix factorization (dot product), showing that after hyperparameter tuning, dot product outperforms MLP in recommendation tasks with lower complexity.

Steffen Rendle, Walid Krichene, Li Zhang et al.

U²-Net employs a two-level nested U-structure with residual U-blocks, enabling effective multi-scale feature extraction without pre-trained backbones, achieving state-of-the-art SOD results.

Xuebin Qin, Zichen Zhang, Chenyang Huang et al.

DeepBidirectional RNN classifier automates identification of operating segments in financial reports, achieving over 97% F1 score.

Zhiqiang Ma, Steven Pomerville, Mingyang Di et al.

This paper provides a comprehensive review of Conformal Prediction (CP), emphasizing its theoretical foundations and recent innovations, ensuring distribution-free error control.

Matteo Fontana, Gianluca Zeni, Simone Vantini

Introduced a spike-triggered non-autoregressive Transformer achieving a real-time factor of 0.0056 on AISHELL-1 dataset.

Zhengkun Tian, Jiangyan Yi, Jianhua Tao et al.

Embedding a Variational Auto-Encoder (VAE) into Bayesian Optimization (BO) enables efficient high-dimensional parameter estimation in personalized cardiac models, achieving over 10x speedup.

Jwala Dhamala, Sandesh Ghimire, John L. Sapp et al.

Movement pruning leverages first-order importance scores during fine-tuning, achieving high sparsity with minimal accuracy loss, outperforming magnitude pruning.

Victor Sanh, Thomas Wolf, Alexander M. Rush

Introduced INFOTABS dataset for semi-structured table inference; models underperform on complex reasoning, highlighting challenges.

Vivek Gupta, Maitrey Mehta, Pegah Nokhiz et al.

MOReL is a model-based offline reinforcement learning algorithm that achieves state-of-the-art performance on the D4RL benchmark.

Rahul Kidambi, Aravind Rajeswaran, Praneeth Netrapalli et al.