Research radar

arXiv preprints, filtered for developer impact.

Concise summaries of new AI papers, ranked for how likely they are to change how production systems are built. arXiv collection remains rate-limited.

Papers tracked

3,367

matching records

Shown

current page

Top radar

http://arxiv.org/abs/2609.31589v1

Code links

on this page

Paper	Authors	arXiv ID	Categories	Published	Code	Radar	Summary
Common-Mode Collapse and Recovery in Direct Feedback Alignment	Varun Reddy, Bernardo L. Sabatini, Houman Safaai	http://arxiv.org/abs/2609.31589v1	cs.LG, cs.NE	Sep 25, 2026	None	Score62	Direct feedback alignment (DFA) trains hidden layers through fixed random projections of outp...
Counterfactual Video Generation Enables Scalable Humanoid Loco-Manipulation	Zihan Wang, Zhen Wu, Pieter Abbeel	http://arxiv.org/abs/2609.38172v1	cs.RO, cs.CV	Sep 29, 2026	None	Score68	Teaching humanoids loco-manipulation skills, such as carrying diverse objects, via visual imi...
PowerSim: Differentiable Physics Simulation and Rendering with Power Diagrams	Trong-Tung Nguyen, Anand Bhattad	http://arxiv.org/abs/2609.38153v1	cs.CV	Sep 29, 2026	None	Score71	We introduce PowerSim, a method to bring physically grounded, differentiable dynamics to Powe...
LIFT: Layout-In-Future Video Generation under Large Viewpoint Change via On-Policy Self-Distillation	Shengxiang Ji, Boyang Wang, Haiyang Xu	http://arxiv.org/abs/2609.38146v1	cs.CV	Sep 29, 2026	None	Score71	We introduce LIFT, a unified image-to-video generation framework that complements camera cont...
The Statistical Cost of Causal Discovery with Feedback	Sunmin Oh, Seungsu Han, Gunwoong Park	http://arxiv.org/abs/2609.34050v1	stat.ML, cs.LG	Sep 28, 2026	None	Score64	What determines the unavoidable sample cost of learning cyclic causal structure? For cyclic l...
Estimate, Don't Imitate: Reusing Differentiable State-Based Policies for Visuomotor Control	Denis Shcherba, Adrian Abel, Eckart Cobo-Briesewitz	http://arxiv.org/abs/2609.34018v1	cs.RO, cs.LG	Sep 27, 2026	None	Score64	Simulation-trained manipulation policies can exploit privileged state information to learn ef...
First-Order Stationarity of Reverse Diffusions	Zhifeng Chen, Chenyang Jiang, Yazhen Wang	http://arxiv.org/abs/2609.31612v1	stat.ML, cs.LG	Sep 25, 2026	None	Score62	Recent literature has shown a strong connection between optimization and sampling. We develop...
Lagrangian--Hamiltonian Flows for Video Prediction and Image Generation: A Symplectic Perspective	Jiawei Hu	http://arxiv.org/abs/2609.35710v1	cs.CV	Sep 28, 2026	None	Score60	We introduce LHFM, a geometric framework for learning image dynamics. Drawing on structures c...
Position Aware Layer Queries for Test Time Training in Vision Language Models	Rajat Modi, Priyank Pathak, Xin Liang	http://arxiv.org/abs/2609.34021v1	cs.CV	Sep 27, 2026	None	Score60	Test-Time Training (TTT) adapts models to incoming test samples (e.g. out-of-distribution, (O...
Gap-free Differentially Private PCA for Gaussian Data	Alina Ene, Huy L. Nguyen	http://arxiv.org/abs/2609.31614v1	cs.DS, cs.LG	Sep 25, 2026	None	Score61	We give a gap-free differentially private algorithm for the principal component analysis (PCA...