❌

Normal view

Received — 9 January 2026 ⏭ Favourites

Beyond Monolithic Architectures: A Multi-Agent Search and Knowledge Optimization Framework for Agentic Search

✇cs.AI, q-bio.NC updates on arXiv.org
By: Yiqun Chen · Lingyong Yan · Zixuan Yang · Erhan Zhang · Jiashu Zhao · Shuaiqiang Wang · Dawei Yin · Jiaxin Mao
9 January 2026 at 13:00
arXiv:2601.04703v1 Announce Type: new Abstract: Agentic search has emerged as a promising paradigm for complex information seeking by enabling Large Language Models (LLMs) to interleave reasoning with tool use. However, prevailing systems rely on monolithic agents that suffer from structural bottlenecks, including unconstrained reasoning outputs that inflate trajectories, sparse outcome-level rewards that complicate credit assignment, and stochastic search noise that destabilizes learning. To address these challenges, we propose \textbf{M-ASK} (Multi-Agent Search and Knowledge), a framework that explicitly decouples agentic search into two complementary roles: Search Behavior Agents, which plan and execute search actions, and Knowledge Management Agents, which aggregate, filter, and maintain a compact internal context. This decomposition allows each agent to focus on a well-defined subtask and reduces interference between search and context construction. Furthermore, to enable stable coordination, M-ASK employs turn-level rewards to provide granular supervision for both search decisions and knowledge updates. Experiments on multi-hop QA benchmarks demonstrate that M-ASK outperforms strong baselines, achieving not only superior answer accuracy but also significantly more stable training dynamics.\footnote{The source code for M-ASK is available at https://github.com/chenyiqun/M-ASK.}
Received — 12 December 2025 ⏭ Favourites

AI-driven virtual cell models in preclinical research: technical pathways, validation mechanisms, and clinical translation potential

✇npj Digital Medicine
By: Chunyu Ma · Han Zhang · Yiwei Rao · Xinyu Jiang · Boheng Liu · Zhikang Sun · Zhenyu Song · Yuan Gao · Yuhao Cui · Xinyu Liu · Zedong Li
11 December 2025 at 08:00

npj Digital Medicine, Published online: 11 December 2025; doi:10.1038/s41746-025-02198-6

AI-driven virtual cell models in preclinical research: technical pathways, validation mechanisms, and clinical translation potential
Received — 31 October 2025 ⏭ Favourites

Multi-Agent Evolve: LLM Self-Improve through Co-evolution

✇cs.AI, q-bio.NC updates on arXiv.org
By: Yixing Chen · Yiding Wang · Siqi Zhu · Haofei Yu · Tao Feng · Muhan Zhang · Mostofa Patwary · Jiaxuan You
31 October 2025 at 12:00
arXiv:2510.23595v3 Announce Type: replace Abstract: Reinforcement Learning (RL) has demonstrated significant potential in enhancing the reasoning capabilities of large language models (LLMs). However, the success of RL for LLMs heavily relies on human-curated datasets and verifiable rewards, which limit their scalability and generality. Recent Self-Play RL methods, inspired by the success of the paradigm in games and Go, aim to enhance LLM reasoning capabilities without human-annotated data. However, their methods primarily depend on a grounded environment for feedback (e.g., a Python interpreter or a game engine); extending them to general domains remains challenging. To address these challenges, we propose Multi-Agent Evolve (MAE), a framework that enables LLMs to self-evolve in solving diverse tasks, including mathematics, reasoning, and general knowledge Q&A. The core design of MAE is based on a triplet of interacting agents (Proposer, Solver, Judge) that are instantiated from a single LLM, and applies reinforcement learning to optimize their behaviors. The Proposer generates questions, the Solver attempts solutions, and the Judge evaluates both while co-evolving. Experiments on Qwen2.5-3B-Instruct demonstrate that MAE achieves an average improvement of 4.54% on multiple benchmarks. These results highlight MAE as a scalable, data-efficient method for enhancing the general reasoning abilities of LLMs with minimal reliance on human-curated supervision.
Received — 29 October 2025 ⏭ Favourites

Multi-Agent Evolve: LLM Self-Improve through Co-evolution

✇cs.AI, q-bio.NC updates on arXiv.org
By: Yixing Chen · Yiding Wang · Siqi Zhu · Haofei Yu · Tao Feng · Muhan Zhang · Mostofa Patwary · Jiaxuan You
29 October 2025 at 12:00
arXiv:2510.23595v2 Announce Type: replace Abstract: Reinforcement Learning (RL) has demonstrated significant potential in enhancing the reasoning capabilities of large language models (LLMs). However, the success of RL for LLMs heavily relies on human-curated datasets and verifiable rewards, which limit their scalability and generality. Recent Self-Play RL methods, inspired by the success of the paradigm in games and Go, aim to enhance LLM reasoning capabilities without human-annotated data. However, their methods primarily depend on a grounded environment for feedback (e.g., a Python interpreter or a game engine); extending them to general domains remains challenging. To address these challenges, we propose Multi-Agent Evolve (MAE), a framework that enables LLMs to self-evolve in solving diverse tasks, including mathematics, reasoning, and general knowledge Q&A. The core design of MAE is based on a triplet of interacting agents (Proposer, Solver, Judge) that are instantiated from a single LLM, and applies reinforcement learning to optimize their behaviors. The Proposer generates questions, the Solver attempts solutions, and the Judge evaluates both while co-evolving. Experiments on Qwen2.5-3B-Instruct demonstrate that MAE achieves an average improvement of 4.54% on multiple benchmarks. These results highlight MAE as a scalable, data-efficient method for enhancing the general reasoning abilities of LLMs with minimal reliance on human-curated supervision.
Received — 26 September 2025 ⏭ Favourites

Quality safety and disparity of an AI chatbot in managing chronic diseases: simulated patient experiments

✇npj Digital Medicine
By: Yafei Si · Yurun Meng · Xi Chen · Ruopeng An · Limin Mao · Bingqin Li · Hazel Bateman · Han Zhang · Hongbin Fan · Jiaqi Zu · Shaoqing Gong · Zhongliang Zhou · Yudong Miao · Xiaojing Fan · Gang Chen
25 September 2025 at 08:00

npj Digital Medicine, Published online: 25 September 2025; doi:10.1038/s41746-025-01956-w

Quality safety and disparity of an AI chatbot in managing chronic diseases: simulated patient experiments
Received — 31 July 2025 ⏭ Favourites

Integrative single-cell multi-omics profiling of human pancreatic islets identifies T1D-associated genes and regulatory signals

✇(Multiomics OR Omics) AND (Pancreatic)
By: Ricardo D'Oliveira Albanus · Xiaoshan Zhang · Zeping Zhao · Henry J Taylor · Xuming Tang · Yuling Han · Peter Orchard · Arushi Varshney · Tuo Zhang · Nandini Manickam · Michael R Erdos · Narisu Narisu · Leland Taylor · Xiaxia Saavedra · Xinyi Liu · Aaron Zhong · Bo Li · Ting Zhou · Ali Naji · Chengyang Liu · Francis S Collins · Stephen C J Parker · Shuibing Chen
30 July 2025 at 18:00

Cell Rep. 2025 Jul 29;44(8):116065. doi: 10.1016/j.celrep.2025.116065. Online ahead of print.

ABSTRACT

Genome-wide association studies (GWASs) have identified over 100 signals associated with type 1 diabetes (T1D). However, it has been challenging to translate any given T1D GWAS signal into mechanistic insights, such as causal variants, their target genes, and the specific cell types involved. Here, we present a comprehensive multi-omic integrative analysis of single-cell/nucleus resolution profiles of gene expression and chromatin accessibility in human pancreatic islets under baseline and T1D-stimulating conditions. We nominate effector cell types for all T1D GWAS signals and the regulatory elements and genes for three independent T1D signals acting through β cells at the DLK1/MEG3, RASGRP1, and TOX loci. Subsequently, we validated the functional impact of these genes and regulatory regions using isogenic human embryonic stem cells (hESCs). We found that loss of RASGRP1 or DLK1, as well as disruption of their corresponding regulatory regions, led to increased β cell apoptosis. Furthermore, β cells derived from isogenic hESCs carrying the T1D risk allele of rs3783355 associated with DLK1 showed elevated β cell death. Through additional RNA sequencing (RNA-seq) and assay for transposase-accessible chromatin using sequencing (ATAC-seq) analyses, we identified five genes upregulated in both RASGRP1-/- and DLK1-/- β-like cells, four of which are near T1D GWAS signals. This integrative approach combining single-cell multi-omics, GWASs, and isogenic human pluripotent stem cell (hPSC)-derived β-like cells illuminates cell type context, genes, single nucleotide polymorphisms (SNPs), and regulatory elements underlying T1D-associated signals, providing insights into the biological functions and molecular mechanisms involved.

PMID:40737125 | DOI:10.1016/j.celrep.2025.116065

Received — 24 December 2024 ⏭ Favourites

Novel DNA methylation biomarkers in stool and blood for early detection of colorectal cancer and precancerous lesions

✇Most Recent Articles: Clinical Epigenetics
By: Yuguang Shen， Dongyang Wang， Tianli Yuan， Hongsheng Fang， Chen Zhu， Juan Qin， Xiaojing Xu， Cheng Zhang， Jiahua Liu， Yuanruohan Zhang， Zhoujin Wen， Jian Tang and Zheng Wang
17 February 2023 at 08:00
Early detection and prevention of precancerous lesions can significantly reduce the morbidity and mortality of colorectal cancer (CRC). Here, we developed new candidate CpG site biomarkers for CRC and evaluate...
There are no more articles
❌