VGAS achieves training-free reward alignment by reducing variance and adaptive selection, enhancing biological sequence generation.

Kwanyoung Kim

Using an autoresearch protocol, an AI agent achieved 99.5% power control optimization in multicell networks, reducing costs by 600x.

Ahmad Khan, Akram Bin Sediq, Sara Azadegi Naeini et al.

Proposes SVD-based channel group-shared low-rank approximation to compress pointwise convolutions in large-kernel CNNs, reducing parameters by over 81%.

Hao Luo, Yiting Yang, Wenyi Zhao et al.

Proposes Bellman calibration using monotone transformations to reduce occupancy ratio bias in offline RL.

Lars van der Laan, Nathan Kallus

Proposes a geometric framework analyzing robustness of neighborhood fairness audits, validated by experiments on benchmark datasets.

Binita Maity

ELR (effective learning rate) governs loss dynamics; matching ELR schedules collapses loss trajectories across configurations.

Zihan Liu, Ruiheng Zheng, Shaobo Zhang et al.

OPDVR combines OPD and RLVR, achieving superior performance on six reasoning benchmarks.

Wenze Lin, Jiale Zhao, Xitai Jiang et al.

PhysicsBench provides a unified benchmark for generative and predictive models in engineering design, evaluating geometric and physical accuracy across multiple tasks and scales.

Sang Won Lee, Hyogu Jeong, Namwoo Kang

This study systematically analyzes how base model priors, reward density, and prompt diversity influence RL post-training success, validated via controlled experiments.

Donovan Clay, Saket Gollapudi, Sankar Harilal et al.

Introduces an adaptive sampling method using leave-one-out denoiser, reducing sampling complexity.

Daniil Dmitriev, Zhihan Huang, Yuting Wei

ProxyFormer uses a proxy architecture for ultra-long context and high-resolution generation, extending sequence length and retrieval accuracy.

Zhongpan Tang

Introduces causal leave-one-out measurement and closed-loop scheduling to improve evidence utilization and context allocation, scalable up to 32B parameters.

Peiyang Liu, Xi Wang, Di Liang et al.

Proposed a lightweight audit method that successfully localized 192 causal break layers.

Taebong Kim, Youngsik Hong, Minsik Kim et al.

DSSM-CRF achieves 75.81% UA on IEMOCAP, addressing cross-speaker influence in conversational emotion recognition.

Guan-Hua Wen, Hou-Chiang Tseng, Kuan-Yu Chen

ARCHER employs Fourier-space features for zero-shot protein pose estimation, achieving median error of 5.0° across unseen structures.

Nhan D. Nguyen, Bao Pham

Proposed a stochastic alternating least-squares algorithm for sparse data analysis, applied to Drosophila connectome data.

Lawrence K. Saul, Ningyuan Huang, Dennis Bollweg et al.

First systematic analysis of model capacity distribution in self-refinement; larger generator and refiner models improve performance, critic size less impactful.

Zhuoyi Yang, Ian G. Harris, Salar Hashemitaheri et al.

Proposes across-design uncertainty analysis using simulated price trajectories, decomposing estimation error into within-design and across-design components.

Pedro Cadahia Delgado

SPARCL uses spectral partitioning to freeze high-energy directions, mitigating spectral interference in analytic continual learning.

James Hartley, Zeropy Surio, Daniel Whitmore et al.

TRACE-C employs rank calibration and multi-channel residuals to detect joint anomalies in multistream telemetry data.

Matthew Faucher