AI Research news and analysis on Web Pulse: 8724 curated articles tracking the latest AI Research developments, tools, and research, updated continuously from vetted sources.
Researchers propose a reinforcement learning with verifiable rewards (RLVR) method for data-efficient adaptation of audio-language models to code-switched automatic speech recognition (ASR). Using Qwen2-Audio across 10 l…
Google DeepMind released Gemma 4, a new generation of open-weight multimodal language models featuring dense and Mixture-of-Experts architectures from 2.3B to 31B parameters. The models introduce a thinking mode for reas…
Researchers introduced CoC-Seduce, a benchmark for testing LLM rule adherence in semi-open textual environments, revealing that frontier models like GPT-5.4, Claude Sonnet 4.6, and Gemini 3.5 Flash are vulnerable to Rhet…
Researchers propose TR-RAG, a teacher-regularized reinforcement learning method for cross-lingual retrieval-augmented generation that addresses language drift and unreliable evidence use. The method combines reward optim…
A study on GPT-5.2's Spanish-Chinese journalistic translations found that translation theory-driven prompts improved quality under expert evaluation, with the brief-oriented prompt scoring highest (MQM: 8.66 vs. 7.84), w…
Researchers propose KARMA, a method that uses knowledge graphs to generate slot-aligned contrastive candidates for preference optimization in large language models, addressing the Resolution Mismatch Problem. KARMA outpe…
Researchers introduced MedCalc-Pro, a benchmark for evaluating LLMs on complex medical calculations involving multiple calculators and nested scales, covering 2,268 cases across 14 departments. They also proposed a gener…
Researchers formalized medical diagnosis as an iterative evidence-seeking task and used reinforcement learning with verifiable rewards to train large language models to autonomously gather clinical evidence. Their framew…
Researchers introduced APeB, a benchmark for evaluating personalization in LLM-powered agents, finding that models struggle with raw, underspecified queries due to ineffective use of interaction histories. A proposed pip…
Researchers propose an organizational memory for LLM-based agents to automate business process execution, addressing the challenge of fragmented organization-specific knowledge. The shared, governed reference layer aims …
A study in a graduate-level mobile robotics course found that a Socratic-Guidance (SG) tutor, which uses dialogic questioning, led to higher learning gains and more understanding-driven prompting strategies compared to a…
A new study on agentic AI policy repair in a hotel-pricing simulator shows that LLM-based editors can nearly match benchmark performance using only region-level diagnostic feedback, achieving RevPAR 108.47 vs. benchmark …
Researchers used interpretable machine learning to predict Parkinson's disease motor severity from QSM and fMRI features, achieving 45.4% variance explained and 75% of predictions within 5 points of clinical scores. The …
Researchers introduced a visual analytics framework for exploring attention dynamics in diffusion models, enabling structured analysis of token-level cross-attention maps across generation steps. The framework integrates…
Researchers propose Dual-Adaptive SAM3 (DA-SAM3), a framework for parameter-efficient medical image segmentation that uses a Dynamic Expert Router and Decomposed Parameterized Experts to reduce MoE overhead by over 80% w…
Researchers propose an additive causal construction (ACC) framework for multi-source image fusion that addresses cross-system discrepancy and entanglement, improving out-of-distribution generalization. The method, valida…
Researchers identified five failure modes in perturbation-based construct-validity audits of AI safety benchmarks, demonstrating that implementation details can silently invalidate audit conclusions. In a case study of t…
Researchers introduced QuantFlow, a federated Mamba-based foundation model for time-series forecasting that combines inverted sequence embedding, bidirectional state-space decoders, quantile regression, and federated lea…
Researchers introduced the Lifting Neural Operator (LiNO), a multiresolution neural operator built on the second-generation wavelet lifting scheme, to learn solution operators of differential equations. LiNO adaptively d…
Researchers compared process and outcome reward structures for reinforcement learning with verifiable rewards (RLVR) in small language models, finding that process-only supervision achieved 63.73% accuracy on GSM8K versu…