University of California, Davis

7 related articles

Subscribe to updates

arXiv Nov 5, 2026
Influence scores for traffic forecasts must pass through conservation constraints: classic rankings shift, and removing the top ten stations raises constrained-model MAE by 44.2%
When a traffic forecasting model enforces junction flow conservation at its output, a training sample's influence on predictions no longer travels only through the loss but also through the constraint projection. Classical influence functions assume unconstrained learning and cannot separate perturbations the physics keeps from those the projection removes. This work computes influence inside the feasible set, ranks station sensitivity on hourly PeMS District 3 data covering 318 freeway links, and shows that withholding the ten highest-ranked stations raises the constrained model's MAE by 44.2%.
When a traffic forecasting model enforces junction flow conservation at its output, a training sample's influence on predictions no longer travels only through the loss but also through the constraint projection. Classical influence functions assume unconstrained learning and cannot separate perturbations the physics keeps from those the projection removes. This work computes influence inside the feasible set, ranks station sensitivity on hourly PeMS District 3 data covering 318 freeway links, and shows that withholding the ten highest-ranked stations raises the constrained model's MAE by 44.2%.
University of Southern California University of California, Davis
Hamidreza Attari（First author）
Influence Functions Physics-Informed Machine Learning HardNet-Aff HardNetAff-STGNN Traffic Flow Forecasting
arXiv Oct 2, 2026
OmniSeek lets an Omni-LLM decide when to look or listen across turns, consistently improving audio-visual reasoning on multimodal benchmarks
The authors present OmniSeek, an agentic framework that makes evidence acquisition part of reasoning: the model dynamically decides whether to look or listen and over which temporal window within a multi-turn protocol, appends retrieved raw audio or visual segments back into context, cold-starts this behavior by supervising on the synthesized OmniTraj-170K corpus of multi-hop Chain-of-Thought trajectories, then optimizes the policy via two-stage reinforcement learning with verifiable rewards, and adds an Audio-Visual Necessity objective that rewards successful trajectories whose reasoning depends on both modalities, with experiments reporting adaptive cross-modal evidence seeking and consistent gains in audio-visual reasoning.
The authors present OmniSeek, an agentic framework that makes evidence acquisition part of reasoning: the model dynamically decides whether to look or listen and over which temporal window within a multi-turn protocol, appends retrieved raw audio or visual segments back into context, cold-starts this behavior by supervising on the synthesized OmniTraj-170K corpus of multi-hop Chain-of-Thought trajectories, then optimizes the policy via two-stage reinforcement learning with verifiable rewards, and adds an Audio-Visual Necessity objective that rewards successful trajectories whose reasoning depends on both modalities, with experiments reporting adaptive cross-modal evidence seeking and consistent gains in audio-visual reasoning.
Adobe Systems (United States)University of California, Davis
Haibo Wang（First author）
OmniSeek Native Tool Integration OmniTraj-170K Reinforcement Learning with Verifiable Rewards Audio-Visual Reasoning
arXiv Oct 2, 2026
LinkedIn and university collaborators find that 7B-to-72B students trained on rejects from smaller frozen models do better on code and math while using less compute
Holding prompts, chosen responses, SeqKD initialization, DPO objective, and training budget fixed while varying only the reject source, the study finds that across Qwen2.5 students from 7B to 72B, rejects generated by smaller frozen Base models with less inference compute train stronger students than the students' own rejects on code generation and mathematical reasoning, and it derives a finite-horizon utility bound in a linearized feature model that motivates three interventions: source mixing, prompt reassignment with lexical permutation, and reference-likelihood reselection.
Holding prompts, chosen responses, SeqKD initialization, DPO objective, and training budget fixed while varying only the reject source, the study finds that across Qwen2.5 students from 7B to 72B, rejects generated by smaller frozen Base models with less inference compute train stronger students than the students' own rejects on code generation and mathematical reasoning, and it derives a finite-horizon utility bound in a linearized feature model that motivates three interventions: source mixing, prompt reassignment with lexical permutation, and reference-likelihood reselection.
University of Wisconsin–Madison Clemson University Arizona State University Pennsylvania State University University of California, Davis
Rui Cai（First author）
Preference Distillation Reject Source Selection Direct Preference Optimization Code Generation Mathematical Reasoning
arXiv Oct 2, 2026
OmniSeek lets a 30B omni-modal model decide whether to look or listen: leading same-scale open models across 10 audio-visual benchmarks, 74.6% on VideoHolmes
The authors present OmniSeek, which turns an Omni-LLM into an active multi-turn tool-using agent (get_audio_clip, get_video_clip), backed by a data engine that synthesizes OmniTraj-170K with 169,725 interleaved audio-visual chain-of-thought trajectories, plus a three-phase training pipeline and an Audio-Visual Necessity reward, achieving leading or competitive results on 10 omni-modal and 4 general video benchmarks.
The authors present OmniSeek, which turns an Omni-LLM into an active multi-turn tool-using agent (get_audio_clip, get_video_clip), backed by a data engine that synthesizes OmniTraj-170K with 169,725 interleaved audio-visual chain-of-thought trajectories, plus a three-phase training pipeline and an Audio-Visual Necessity reward, achieving leading or competitive results on 10 omni-modal and 4 general video benchmarks.
Adobe Systems (United States)University of California, Davis
Haibo Wang（First author）
OmniSeek OmniTraj-170K Audio-Visual Necessity Reward Native Tool Integration Audio-Visual Reasoning
ACM Internet Measurement Conference (IMC) Sep 29, 2026
Researchers reverse-engineer Meta Pixel configurations, finding 98.4% default-driven tracking on health sites and Core Setup covering only 34.3% while being bypassable via hashed URLs
The study introduces PixelConfig, a differential-analysis framework that reverse-engineers Meta Pixel configurations through code-patching replays and developer-account controlled experiments, and uses Internet Archive's Wayback Machine to longitudinally compare configurations on 18K health websites against a top-10K control group from 2017 to 2024, finding that default-enabled tracking features such as automatic events and first-party cookies reached adoption rates up to 98.4%, that health websites show tracking of potentially sensitive information tied to booking medical appointments and button clicks associated with specific conditions such as erectile dysfunction, and that restriction features like Core Setup were configured on 34.3% of health websites versus 8.
The study introduces PixelConfig, a differential-analysis framework that reverse-engineers Meta Pixel configurations through code-patching replays and developer-account controlled experiments, and uses Internet Archive's Wayback Machine to longitudinally compare configurations on 18K health websites against a top-10K control group from 2017 to 2024, finding that default-enabled tracking features such as automatic events and first-party cookies reached adoption rates up to 98.4%, that health websites show tracking of potentially sensitive information tied to booking medical appointments and button clicks associated with specific conditions such as erectile dysfunction, and that restriction features like Core Setup were configured on 34.3% of health websites versus 8.
Lahore University of Management Sciences University of California, Davis
Abdullah Ghani（First author）
PixelConfig Meta Pixel reverse engineering Internet measurement
Letters in Mathematical Physics Sep 25, 2026
Scalet proves a finite entanglement length for one-dimensional spin-chain Gibbs states at any finite temperature, so left and right half-chains become exactly separable once a long enough interval is traced out
Scalet proves that for any finite-range, bounded-strength local Hamiltonian on a one-dimensional spin chain at any fixed finite temperature, there exists an entanglement length ℓ depending only on temperature and locality, not on system size, such that whenever the middle interval B satisfies |B| ≥ ℓ, tracing out B from the tripartite Gibbs state ρABC yields a state ρAC that is separable between A and C; consequently entanglement of formation, distillable entanglement, entanglement cost, and entanglement relative entropy are exactly zero on that state, the statement holds uniformly in system size and extends to the KMS state of the infinite system, and the author calls this a spatial sudden death of entanglement.
Scalet proves that for any finite-range, bounded-strength local Hamiltonian on a one-dimensional spin chain at any fixed finite temperature, there exists an entanglement length ℓ depending only on temperature and locality, not on system size, such that whenever the middle interval B satisfies |B| ≥ ℓ, tracing out B from the tripartite Gibbs state ρABC yields a state ρAC that is separable between A and C; consequently entanglement of formation, distillable entanglement, entanglement cost, and entanglement relative entropy are exactly zero on that state, the statement holds uniformly in system size and extends to the KMS state of the infinite system, and the author calls this a spatial sudden death of entanglement.
University of California, Davis
Samuel O. Scalet（First author, Corresponding author）
Quantum Information and Cryptography Quantum Computing Algorithms and Architecture Quantum many-body systems
arXiv Sep 27, 2026
VideoGen-Agent uses multitask reinforcement learning to teach a video-generation agent to call retrieval, simulation, and verification tools, lifting its base generator from 56.5 to 75.6 on VABench and to 86.1 with upgraded generation tools
The work introduces VideoGen-Agent, a multimodal agent trained with supervised fine-tuning followed by multitask agentic reinforcement learning to coordinate augmentation, generation, and verification tools across six video-generation tasks (Procedural Knowledge, Single-Entity Identity, Multi-Entity Identity, Physics Simulation, Compositional Scene, and Multi-Shot), and builds VABench, a held-out benchmark of 600 prompts; on VABench it improves the base T2V generator Seedance 1.0 from 56.5 to 75.6, and further to 86.1 when compatible upgraded generation tools unseen during training are used, with human raters preferring that configuration in 100 pairwise comparisons.
The work introduces VideoGen-Agent, a multimodal agent trained with supervised fine-tuning followed by multitask agentic reinforcement learning to coordinate augmentation, generation, and verification tools across six video-generation tasks (Procedural Knowledge, Single-Entity Identity, Multi-Entity Identity, Physics Simulation, Compositional Scene, and Multi-Shot), and builds VABench, a held-out benchmark of 600 prompts; on VABench it improves the base T2V generator Seedance 1.0 from 56.5 to 75.6, and further to 86.1 when compatible upgraded generation tools unseen during training are used, with human raters preferring that configuration in 100 pairwise comparisons.
Stanford University Princeton University Chinese University of Hong Kong George Washington University University of California, Davis
Binxu Li（First author）
VideoGen-Agent Agentic Reinforcement Learning VABench Text-to-Video Generation Supervised Fine-tuning

You’re all caught up

Influence scores for traffic forecasts must pass through conservation constraints: classic rankings shift, and removing the top ten stations raises constrained-model MAE by 44.2%

OmniSeek lets an Omni-LLM decide when to look or listen across turns, consistently improving audio-visual reasoning on multimodal benchmarks

LinkedIn and university collaborators find that 7B-to-72B students trained on rejects from smaller frozen models do better on code and math while using less compute

OmniSeek lets a 30B omni-modal model decide whether to look or listen: leading same-scale open models across 10 audio-visual benchmarks, 74.6% on VideoHolmes

Researchers reverse-engineer Meta Pixel configurations, finding 98.4% default-driven tracking on health sites and Core Setup covering only 34.3% while being bypassable via hashed URLs

Scalet proves a finite entanglement length for one-dimensional spin-chain Gibbs states at any finite temperature, so left and right half-chains become exactly separable once a long enough interval is traced out

VideoGen-Agent uses multitask reinforcement learning to teach a video-generation agent to call retrieval, simulation, and verification tools, lifting its base generator from 56.5 to 75.6 on VABench and to 86.1 with upgraded generation tools