Projection

Follow this topic Ask about this topic

Momentum

12 papers in the last four weeks, up 100% on the four weeks before. 0.1% of all new papers.

Jul 6Week of Sep 21

Latest papers 89

All topics

Directional Alignment Mitigates Reward Hacking in Reinforcement Learning for Language Models
May 24, 2026Wenlong Deng, Jiaji Huang, Kaan Ozkara +4Reward Gradients Exploitation
University of British Columbia · Vector Institute · Amazon
Lifted Schrödinger Bridges for Gaussian Mixture Endpoints: Projection Gaps and Path-Space Obstructions
May 24, 2026Siddhartha Ganguly, George Rapakoulias, Panagiotis TsiotrasSchrödinger Bridges Gaussian Mixture Models
S. Ganguly, G. Rapakoulias, and P. Tsiotras are with Daniel Guggenheim School of Aerospace, Georgia Institute of Technology, ½ Atlanta, USA.
Soft Tuy-Completeness for Robust Projection Selection in Cone-Beam CT
May 20, 2026Linda-Sophie Schneider, Andreas MaierCone-Beam Computed Tomography Projection
Pattern Recognition Lab, Friedrich-Alexander-Universität Erlangen-Nürnberg, Germany
On Stability and Decomposition of Sample Quantiles under Heavy-Tailed Distributions
May 18, 2026Choudur LakshminarayanQuantile Regression Long-Tailed Distribution
School of Business, Stevens Institute of Technology
On RGB-TIR Stereo Calibration under Extreme Resolution Asymmetry
May 15, 2026Michał Król, Michał Salamonowicz, Władysław Skarbek +1Camera Intrinsics Projection
Independent Researcher, Poland · Warsaw University of Technology, Poland · Polish Japanese Academy of Information Technology, Poland
A Novel Schur-Decomposition-Based Weight Projection Method for Stable State-Space Neural-Network Architectures
May 14, 2026Sergio Vanegas, Lasse Lensu, Fredy RuizSystem Identification State Space Models
Computational Engineering LUT University Lappeenranta, Finland · DEIB Politecnico di Milano Milan, Italy
Efficient and provably convergent end-to-end training of deep neural networks with linear constraints
May 12, 2026Zonglin Yang, Zhexuan Gu, Yancheng YuanBackpropagation Inequality Constraints
Department of Applied Mathematics, The Hong Kong Polytechnic University, Hung Hom, Hong Kong
Metric-Gradient Projection for Stable Multi-Agent Policy Learning
May 12, 2026Zuyuan Zhang, Sizhe Tang, Mahdi Imani +1Multi-Agent Reinforcement Learning Wasserstein Gradient Flows
The George Washington University · Northeastern University
Towards Compact Sign Language Translation: Frame Rate and Model Size Trade-offs
May 10, 2026Kuanwei Chen, Mengfeng TsaiGloss-Free Sign Language Translation Sign Language Translation
Computer Science and Information Engineering, National Central University, Zhongli, Taiwan
Optimal Regret for Single Index Bandits
May 10, 2026Devdan Dey, Sujoy Bhore, Avishek GhoshStochastic Multi-Armed Bandits Optimal Regret
Department of Computer Science and Engineering, Indian Institute of Technology Bombay.
Adversary-Robust Learning from Fully Asynchronous Directional Derivative Estimates
May 10, 2026Anik Kumar Paul, Nibedita Roy, Nagesh Talagani +3Learning-Augmented Algorithms Zeroth-Order
Computer Science and Automation, Indian Institute of Science, Bengaluru · Edwardson School of Industrial Engineering, Purdue University · Department of Computer Science, IMT Atlantique
A Flow Matching Algorithm for Many-Shot Adaptation to Unseen Distributions
May 7, 2026Tyler Ingebrand, Ruihan Zhao, Kushagra Gupta +3Generative Flow Networks Generative Models
University of Texas at Austin
Listwise Policy Optimization: Group-based RLVR as Target-Projection on the LLM Response Simplex
May 7, 2026Yun Qu, Qi Wang, Yixiu Mao +11Reinforcement Learning With Verifiable Reward Proximal Policy Optimization
Department of Automation, Tsinghua University · 2LLM Department, Tencent
LLM-Augmented Semantic Steering of Text Embedding Projection Spaces
May 3, 2026Wei Liu, Eric Krokos, Kirsten Whitley +2Text Embeddings Projection
Virginia Tech Blacksburg, Virginia, USA · Department of Defense Washington, DC, USA · Tulane University New Orleans, Louisiana, USA
SplAttN: Bridging 2D and 3D with Gaussian Soft Splatting and Attention for Point Cloud Completion
May 2, 2026Zhaoyang Li, Zhichao You, Tianrui LiPoint Clouds Gaussian Splatting
School of Computing and Artificial Intelligence, Southwest Jiaotong University, Chengdu, China
CheXmix: Unified Generative Pretraining for Vision Language Models in Medical Imaging
Apr 24, 2026Ashwin Kumar, Robbie Holland, Corey Barrett +8Chest X-Ray Medical Vision-Language Models
Stanford AIMI, Stanford University · Oracle Health AI · Department of Radiology, Stanford University
Contrastive Semantic Projection: Faithful Neuron Labeling with Contrastive Examples
Apr 24, 2026Oussama Bouanani, Jim Berend, Wojciech Samek +2Contrastive Learning Neurons
Fraunhofer Heinrich Hertz Institute, Berlin, Germany · Technische Universität Berlin, Berlin, Germany · BIFOLD – Berlin Institute for the Foundations of Learning and Data, Berlin, Germany +1
Local Neighborhood Instability in Parametric Projections: Quantitative and Visual Analysis
Apr 23, 2026Frederik L. Dennig, Daniel A. KeimProjection Projection Heads
University of Konstanz, Germany
ONOTE: Benchmarking Omnimodal Notation Processing for Expert-level Music Intelligence
Apr 22, 2026Menghe Ma, Siqing Wei, Yuecheng Xing +5Music Structure Analysis Notes
Beijing University of Posts and Telecommunications Beijing, China · China Conservatory of Music Beijing, China · Nanyang Technological University Singapore, Singapore
Schema-Key Wording as an Instruction Channel in Structured Generation under Constrained Decoding
Apr 16, 2026Yifan LeConstrained Decoding Data Generation
Independent Researcher
High-Resolution Climate Projections Using Diffusion-Based Downscaling of a Lightweight Climate Emulator
Feb 13, 2026Haiwen Guan, Dibyajyoti Chakraborty, Moein Darman +3Dynamical Downscaling Wind
1Information Sciences and Technology, The Pennsylvania State University, University Park, Pennsylvania · Department of Applied Mathematics, University of California, Santa Cruz, California · 3Allen Institute for Artificial Intelligence (AI2), Seattle, Washington +1
BEV-ODOM2: Enhanced BEV-based Monocular Visual Odometry with PV-BEV Fusion and Dense Flow Supervision for Ground Robots
Sep 18, 2025Yufei Wei, Chenxiao Hu, Wangtao Lu +5Visual Inertial Odometry Bird'S-Eye View
Deep Unrolling of Sparsity-Induced RDO for 3D Point Cloud Attribute Coding
Sep 10, 2025Tam Thuc Do, Philip A. Chou, Gene CheungPoint Clouds Deep Unfolding Networks
department of EECS, York University, 4700 Keele Street, Toronto, M3J 1P3, Canada
DE-VAE: Revealing Uncertainty in Parametric and Inverse Projections with Variational Autoencoders using Differential Entropy
Aug 16, 2025Frederik L. Dennig, Daniel A. KeimConditional Variational Autoencoder Star-Vae
University of Konstanz
Out-of-Sample Embedding with Proximity Data: Projection versus Restricted Reconstruction
May 10, 2025Michael W. Trosset, Kaiyi Tan, Minh Tang +1Principal Component Analysis Projection
Department of Statistics, Indiana University. · Department of Statistics, North Carolina State University. · Department of Applied Mathematics & Statistics, Johns Hopkins University.
Anisotropic View Distance Metric for High-Dimensional Data: Theory, Geometry, and Fast Computation
Jun 10, 2022Yiqun Zhang, Hou-biao LiK-Means Distance
School of Mathematical Sciences, University of Electronic Science and Technology of China, Chengdu 611731, China
EPMF: Efficient Perception-aware Multi-sensor Fusion for 3D Semantic Segmentation
Jun 21, 2021Mingkui Tan, Zhuangwei Zhuang, Sitao Chen +4Multi-Sensor Fusion 3D Perception
School of Software Engineering, South China University of Technology · Pazhou Laboratory, Guangzhou, China · School of Electronic and Information Engineering, South China University of Technology +2
Differentiable Jitter Correction using Deep Learning-based Image Quality Metric for Phase-Contrast Micro-CT
Date pendingJunan Chen, Yiting Jia, Joscha Maier +6Cone-Beam Computed Tomography Image Registration
ImFusion GmbH, Ridlerstr. 57, 80339 Munich, Germany. · Research Group Biomedical Imaging Physics, Department of Physics, TUM School of Natural Sciences and Munich Institute of Biomedical Engineering, Technical University of Munich, James-Franck-Str. 1, 85748 Garching, Germany. · Centre for X-Ray and Nano Science CXNS, Deutsches Elektronen-Synchrotron (DESY), Notkestr. 85, 22067 Hamburg, Germany. +1
Denoising as Projection: Constrained Optimization with Gradient-Guided Diffusion
Date pendingRunyu Zhang, Jiawei Zhang, Gioele Zardini +2Wasserstein Gradient Flows Diffusion Models
MIT · University of Wisconsin–Madison

May 24, 2026, cs.LGJ/K move · Enter open · S save

Directional Alignment Mitigates Reward Hacking in Reinforcement Learning for Language Models

Wenlong Deng, Jiaji Huang, Kaan Ozkara, Yushu Li+3

University of British Columbia · Vector Institute · Amazon

Reward Gradients Exploitation Projection

Open paper AskPDF arXiv

Reward hacking arises when a model improves a proxy reward by exploiting shortcuts rather than solving the intended task. We study this failure mode through the geometry of reinforcement learning updates in language models and argue that hacking emerges when optimization drifts away from a stable low-dimensional learning trajectory. We analyze this drift through dominant singular directions of parameter updates and show that reward-hacking runs exhibit substantially larger directional change than clean runs. Motivated by this observation, we introduce trusted-direction projection, which constrains gradients to remain within a clean reference subspace. Across reward-hacking experiments on mathematical reasoning, the proposed approach delays shortcut exploitation and better preserves task performance.

Momentum

Related topics

Latest papers 89