❌

Normal view

Received — 27 May 2026 ⏭

From Accuracy to Auditability: A Survey of Determinism in Financial AI Systems

✇cs.AI, q-bio.NC updates on arXiv.org
By: Ruizhe Zhou · Xiaoyang Liu · Gaoyuan Du · Yi Zheng · Shouxi Ren · Deepayan Chakrabarti · Dengdu Jiang
26 May 2026 at 12:00
arXiv:2605.23955v1 Announce Type: new Abstract: Deploying machine learning in regulated financial environments -- credit risk, fraud detection, and anti-money laundering -- exposes critical vulnerabilities in algorithmic reproducibility. While early financial ML addressed statistical challenges such as backtest overfitting, deep neural networks and Generative AI have introduced mechanical nondeterminism rooted in hardware and architecture. This survey provides a systems perspective on reproducibility failures across three modalities now dominant in financial AI: tabular models (post-hoc explanation variance), graph networks (stochastic sampling and temporal asynchrony), and LLM-based agentic workflows (batch-dependent divergence and trajectory drift). We supplement the literature analysis with first-party experiments on public financial datasets -- quantifying explanation rank instability in credit scoring, prediction flip rates in GNN-based fraud detection, and tensor-parallel-induced output divergence in LLM entity extraction. We propose a layered evaluation framework linking modality-specific metrics (RBO, D_cos, TDI, PSD) to audit readiness, and empirically validate the complementarity of logit-level and semantic-level determinism measures.

Reasoning as an Attack Surface: Adaptive Evolutionary CoT Jailbreaks for LLMs

✇cs.AI, q-bio.NC updates on arXiv.org
By: Jianan Li · Simeng Qin · Xiaojun Jia · Lionel Z. Wang · Tianhang Zheng · Xiaoshuang Jia · Yang Liu · Xiaochun Cao
26 May 2026 at 12:00
arXiv:2605.24497v1 Announce Type: new Abstract: Large Reasoning Models (LRMs) have demonstrated remarkable capabilities in reasoning and generation tasks and are increasingly deployed in real-world applications. However, their explicit chain-of-thought (CoT) mechanism introduces new security risks, making them particularly vulnerable to jailbreak attacks. Existing approaches often rely on static CoT templates to elicit harmful outputs, but such fixed designs suffer from limited diversity, adaptability, and effectiveness. To overcome these limitations, we propose an adaptive evolutionary CoT jailbreak framework, called AE-CoT. Specifically, the method first rewrites harmful goals into mild prompts with teacher role-play and decomposes them into semantically coherent reasoning fragments to construct a pool of CoT jailbreak candidates. Then, within a structured representation space, we perform multi-generation evolutionary search, where candidate diversity is expanded through fragment-level crossover and a mutation strategy with an adaptive mutation-rate control mechanism. An independent scoring model provides graded harmfulness evaluations, and high-scoring candidates are further enhanced with a harmful CoT template to induce more destructive generations. Extensive experiments across multiple models and datasets demonstrate the effectiveness of the proposed AE-CoT, consistently outperforming state-of-the-art jailbreak methods.

ProActor: Timing-Aware Reinforcement Learning for Proactive Task Scheduling Agents

✇cs.AI, q-bio.NC updates on arXiv.org
By: Lei Ding · Bin He · Chenguang Wang · Yang Liu
26 May 2026 at 12:00
arXiv:2605.24900v1 Announce Type: new Abstract: Proactive task-oriented agents must autonomously anticipate user needs, identify actionable opportunities, and trigger software actions at appropriate moments - fundamentally shifting from reactive systems that await explicit instructions. However, existing approaches lack generalizable end-to-end solutions for measuring and optimizing such anticipatory behaviors. This paper introduces ProActor, a unified framework for conversational task scheduling that integrates: (1) a domain-agnostic automated annotation methodology that enables scalable proactiveness reinforcement learning (RL) by generating full opportunity time windows instead of rigid point labels, (2) systematic proactiveness metrics capturing both timing quality and reference action alignment, and (3) RL optimization using GRPO with various reward designs. Our insight is that RULER-based rewards with proactiveness rubrics are crucial for improving timing quality, and that proactiveness optimization enabled by stage-aware composite rewards is key to balancing timing quality and reference action alignment. Timing-aware RL requires extensive exploration, demanding efficient infrastructure. We develop ART-F, an adaptive framework combining request-adaptive inference clusters with DDP-based training on single-node multi-GPU systems, enabling LoRA training of 4-bit Qwen2.5-14B-ProActor-Q4 with 4-8x speedups. Experiments on two newly auto-annotated datasets demonstrate significant improvements in proactive timing while maintaining action consistency comparable to state-of-the-art (SOTA) baselines. Ablations validate the effectiveness of distinct composite reward variations.

CODESKILL: Learning Self-Evolving Skills for Coding Agents

✇cs.AI, q-bio.NC updates on arXiv.org
By: Yanzhou Li · Yiran Zhang · Xiaoyu Zhang · Xiaoxia Liu · Yang Liu
26 May 2026 at 12:00
arXiv:2605.25430v1 Announce Type: new Abstract: Coding agents produce rich trajectories while solving software-engineering tasks. To enable agent self-evolution, these trajectories can be distilled into reusable procedural skills that compactly encode experience to guide future behavior. However, existing skill construction and maintenance methods often rely on fixed prompts and heuristic update rules, leaving it unclear how knowledge should be selected, abstracted, and maintained to best serve downstream agents. We propose CODESKILL, an LLM-based framework that reformulates skill extraction and skill-bank maintenance as a learnable management policy. CODESKILL extracts multi-granularity procedural skills from coding-agent trajectories, evolves skills with new experience, and maintains a compact skill bank for future task solving. We train CODESKILL with reinforcement learning, using a hybrid reward that combines dense rubric-based skill-quality feedback with sparse verifiable execution feedback from the frozen downstream agent. Experiments on EnvBench, SWE-Bench Verified, and Terminal-Bench 2 show that CODESKILL improves average pass rate by 9.69 over the no-skill baseline and by 4.01 over the strongest prompt-based or memory baseline, while maintaining the skill bank at a stable size during iterative construction.

STREAM: A Data-Centric Framework for Mining High-Value Task-Oriented Dialogues from Streaming Media

✇cs.AI, q-bio.NC updates on arXiv.org
By: Liang Xue · Haoyu Liu · Cheng Wang · Pengyu Chen · Haozhuo Zheng · Yang Liu
26 May 2026 at 12:00
arXiv:2605.25162v1 Announce Type: cross Abstract: Large language models for vertical domains are bottlenecked by the scarcity of complex, domain-specific task-oriented dialogues. Existing data acquisition pipelines face a persistent trilemma: expert annotation is expensive, real-world service conversations are constrained by privacy and commercial restrictions, and static corpora quickly become temporally stale. We propose Stream, a data-centric framework that leverages publicly available streaming media (live streams and short videos) to synthesize high-value service dialogues at scale. Stream mines authentic interaction signals from noisy streams and synthesizes conversations by integrating role-grounded persona construction with Conversational Blueprint construction; it further adopts retrieval-augmented generation (RAG) to support knowledge-aware responses. Based on Stream, we release StreamDial, a large-scale multi-domain dataset covering Automotive, Restaurant, and Hotel. StreamDial contains 87,498 dialogue sessions and 1,497,320 turns in total, with an average of 17.11 turns per session and a comparable scale across domains. Each session is organized as a structured quadruplet $\langle P_u, P_a, B, H \rangle$ that pairs dialogue history with explicit user/agent personas and a Conversational Blueprint, capturing realistic service behaviors such as requirement mining, constraint conflicts, negotiation, and recovery. Evaluations with automatic judges and downstream tasks show that StreamDial improves intrinsic dialogue quality over strong baselines, and models trained with StreamDial improve Dialogue State Tracking across backbones; we further report a completed human-evaluation set and encouraging multilingual transfer on Qwen3-8B under a controlled training budget. The data is released in https://github.com/hitxueliang/DialogDataSetBySTREAM.

NPSolver: Neural Poisson Solver with Iterative Physics Supervision

✇cs.AI, q-bio.NC updates on arXiv.org
By: Bocheng Zeng · Rui Zhang · Runze Mao · Mengtao Yan · Xuan Bai · Yang Liu · Zhi X. Chen · Hao Sun
26 May 2026 at 12:00
arXiv:2605.25786v1 Announce Type: cross Abstract: Efficiently solving Poisson equations on complex, irregular domains remains a fundamental challenge in scientific computing, as classical iterative solvers often suffer from prohibitive runtime due to ill-conditioned systems. While neural operators offer a fast alternative, they typically rely on large-scale labeled datasets or struggle with unstable training dynamics when using physics-informed residual losses. We propose \textsc{NPSolver}, a neural Poisson solver trained without solution labels via iterative physics supervision. Instead of relying on fully converged numerical solutions or raw PDE residuals, \textsc{NPSolver} utilizes a small number of preconditioned conjugate gradient (PCG) steps to refine its own predictions, providing a more stable and well-scaled training signal. Theoretical analysis confirms that this iterative supervision serves as a well-conditioned error proxy and that a stop-gradient design is essential for optimization stability. To better capture boundary-driven features under mixed boundary conditions, we further introduce the Boundary-Aware Transolver (\textsc{BA-Transolver}) architecture that explicitly separates interior and boundary tokenization. Extensive evaluations on 2D and 3D irregular geometries demonstrate that \textsc{NPSolver} outperforms both physics-informed and data-driven baselines. Furthermore, a downstream thermal control task highlights the model's capability for conducting efficient and reliable gradient-based boundary control. We will release our codes and data at https://github.com/intell-sci-comput/NPSolver.

OASIS: Observation-Action Space Alignment via SE(3) Trajectory Prediction for Robotic Manipulation

✇cs.AI, q-bio.NC updates on arXiv.org
By: Xinzhe Chen · Sihua Ren · Liqi Huang · Haowen Sun · Mingyang Li · Xingyu Chen · Zeyang Liu · Xuguang Lan
26 May 2026 at 12:00
arXiv:2605.25829v1 Announce Type: cross Abstract: Recent vision-language-action (VLA) models and world action models (WAMs) advance robotic manipulation by enriching intermediate representations with auxiliary spatial features or future visual-state prediction. However, these representations largely remain within the observation space and do not share the rigid-body geometry of the action space, forcing the action decoder to implicitly recover this geometry. We propose OASIS, a visuomotor policy that aligns the intermediate representation with the action space via $SE(3)$ end-effector trajectory prediction. OASIS couples a 3D-aware feature encoder that fuses vision-language and metric-depth features with an $SE(3)$ trajectory predictor that produces a camera-frame end-effector trajectory. Conditioned on the predictor's pose-supervised hidden states, the action decoder generates action chunks consistent with rigid-body motion. Across simulation and real-world experiments, OASIS outperforms VLA and WAM baselines in success rate and out-of-distribution generalization. Our project page is available at https://npuhandsome.github.io/OASIS_web.

Voting with the Graph: Stable RLAIF via Topological Consistency Maximization

✇cs.AI, q-bio.NC updates on arXiv.org
By: Boyin Liu · Zhuo Zhang · Sen Huang · Lipeng Xie · Qingxu Fu · Haoran Chen · LI YU · Tianyi Hu · Zhaoyang Liu · Bolin Ding · Dongbin Zhao
26 May 2026 at 12:00
arXiv:2510.15514v3 Announce Type: replace Abstract: Reinforcement Learning from AI Feedback (RLAIF) relies on LLM judges as preference measurement instruments, yet these instruments are fundamentally limited by random measurement errors -- stochastic fluctuations that manifest as preference cycles (e.g., $A \succ B \succ C \succ A$), occurring in 5-9% of evaluations across state-of-the-art models. While repeated sampling mitigates noise by averaging multiple judgments, it treats each comparison in isolation and fails to exploit the structural constraints that distinguish systematic signals from random noise. We introduce Topological Consensus Rewards (TCR), a framework that leverages transitivity as a denoising mechanism via topological majority voting: systematic signals reinforce each other through transitive chains, while random errors cluster into topologically exposed cycles. TCR approximates the Maximum Acyclic Subgraph to filter stochastic noise from preference signals. We also propose Cycle Incidence Rate (CIR) as a diagnostic metric that measures the proportion of samples containing preference cycles. Under our noise model, these cycles arise primarily from stochastic measurement errors rather than genuine intransitivity. Experiments on Arena-Hard, MT-Bench, and WritingBench demonstrate that TCR consistently outperforms pairwise baselines and classical ranking algorithms, while exhibiting robust performance across different judge models.

PathMem: Toward Cognition-Aligned Memory Transformation for Pathology MLLMs

✇cs.AI, q-bio.NC updates on arXiv.org
By: Jinyue Li · Yuci Liang · Qiankun Li · Xinheng Lyu · Jiayu Qian · Huabao Chen · Kun Wang · Zhigang Zeng · Anil Anthony Bharath · Yang Liu
26 May 2026 at 12:00
arXiv:2603.09943v2 Announce Type: replace Abstract: Computational pathology demands both visual pattern recognition and dynamic integration of structured domain knowledge, including taxonomy, grading criteria, and clinical evidence. In practice, diagnostic reasoning requires linking morphological evidence with formal diagnostic and grading criteria. Although multimodal large language models (MLLMs) demonstrate strong vision language reasoning capabilities, they lack explicit mechanisms for structured knowledge integration and interpretable memory control. As a result, existing models struggle to consistently incorporate pathology-specific diagnostic standards during reasoning. Inspired by the hierarchical memory process of human pathologists, we propose PathMem, a memory-centric multimodal framework for pathology MLLMs. PathMem organizes structured pathology knowledge as a long-term memory (LTM) and introduces a Memory Transformer that models the dynamic transition from LTM to working memory (WM) through multimodal memory activation and context-aware knowledge grounding, enabling context-aware memory refinement for downstream reasoning. PathMem achieves SOTA performance across benchmarks, improving WSI-Bench report generation (12.8% WSI-Precision, 10.1% WSI-Relevance) and open-ended diagnosis by 9.7% and 8.9% over prior WSI-based models.

Topology-Driven Transferability Estimation of Medical Foundation Models for Segmentation

✇cs.AI, q-bio.NC updates on arXiv.org
By: Jiaqi Tang · Shaoyang Zhang · Xiaoqi Wang · Jiaying Zhou · Yang Liu · Qingchao Chen
26 May 2026 at 12:00
arXiv:2602.23916v2 Announce Type: replace-cross Abstract: The advent of large-scale self-supervised learning (SSL) has produced a vast zoo of medical foundation models. However, selecting optimal medical foundation models for specific segmentation tasks remains a computational bottleneck. Existing Transferability Estimation (TE) metrics, primarily designed for classification, rely on global statistical assumptions and fail to capture the topological complexity essential for dense prediction. We propose a novel Topology-Driven Transferability Estimation framework that evaluates manifold tractability rather than statistical overlap. Our approach introduces three components: (1) Global Representation Topology Divergence (GRTD), utilizing Minimum Spanning Trees to quantify feature-label structural isomorphism; (2) Local Boundary-Aware Topological Consistency (LBTC), which assesses manifold separability specifically at critical anatomical boundaries; and (3) Task-Adaptive Fusion, which dynamically integrates global and local metrics based on the semantic cardinality of the target task. Validated on the large-scale OpenMind benchmark across diverse anatomical targets and SSL foundation models, our approach significantly outperforms state-of-the-art baselines by around 31% relative improvement in the weighted Kendall metric, providing a robust, training-free proxy for efficient model selection without the cost of fine-tuning. The code will be made publicly available upon acceptance.

Cooperative Memory Paging with Keyword Bookmarks for Long-Horizon LLM Conversations

✇cs.AI, q-bio.NC updates on arXiv.org
By: Ziyang Liu
26 May 2026 at 12:00
arXiv:2604.12376v2 Announce Type: replace-cross Abstract: When LLM conversations grow beyond the context window, old content must be evicted -- but how does the model recover it when needed? We propose cooperative paging: evicted segments are replaced with minimal keyword bookmarks ([pN:keywords], ~8-24 tokens each), and the model is given a recall() tool to retrieve full content on demand. On the LoCoMo benchmark (10 real multi-session conversations, 300+ turns), cooperative paging achieves the highest answer quality among six methods -- outperforming truncation, BM25, word-overlap retrieval, a search-tool baseline, and full context -- on four models (GPT-4o-mini, DeepSeek-v3.2, Claude Haiku, GLM-5), confirmed by four independent LLM judges ($p=0.017$, paired bootstrap). We then study the paging design space with a 5x4 ablation over boundary strategies and eviction policies (3,176 synthetic probes, 1,600 LoCoMo probes). Key findings: (1) coarse fixed-size pages (fixed_20) reach 96.7% while content-aware topic_shift collapses to 56.7%; (2) eviction policy choice is data-dependent (FIFO best on synthetic, LFU on LoCoMo); (3) two bookmark generation strategies improve over the heuristic baseline (+4.4 and +8.7 E2E points); (4) the remaining bottleneck is bookmark discrimination -- the model triggers recall() 96% of the time but selects the correct page only 57% when bookmarks are insufficiently distinctive. Keyword specificity alone accounts for a 25 percentage point accuracy difference.
  • ✇cs.AI, q-bio.NC updates on arXiv.org
  • Depth Registers Unlock W4A4 on SwiGLU: A Reader/Generator Decomposition Ziyang Liu
    arXiv:2604.18128v2 Announce Type: replace-cross Abstract: We study post-training W4A4 quantization in a controlled 300M-parameter SwiGLU decoder-only language model trained on 5B tokens of FineWeb-Edu, and ask which input-activation sites dominate the error. Naive round-to-nearest W4A4 collapses validation perplexity from FP16 23.6 to 1727. A simple residual-axis training-time intervention -- Depth Registers with a register-magnitude hinge loss (DR+sink) -- reduces this to 119 (about 14x) at ma
     
  • 🏷️
  • ↗️

Depth Registers Unlock W4A4 on SwiGLU: A Reader/Generator Decomposition

✇cs.AI, q-bio.NC updates on arXiv.org
By: Ziyang Liu
26 May 2026 at 12:00
arXiv:2604.18128v2 Announce Type: replace-cross Abstract: We study post-training W4A4 quantization in a controlled 300M-parameter SwiGLU decoder-only language model trained on 5B tokens of FineWeb-Edu, and ask which input-activation sites dominate the error. Naive round-to-nearest W4A4 collapses validation perplexity from FP16 23.6 to 1727. A simple residual-axis training-time intervention -- Depth Registers with a register-magnitude hinge loss (DR+sink) -- reduces this to 119 (about 14x) at matched FP16 PPL and matched zero-shot capacity, and composes with SmoothQuant to 39.9 PPL. The residual ~2 PPL gap to FP16 is the diagnostic core. We decompose W4A4 damage by input-activation site: the five trainable linears in a SwiGLU block split into residual-axis readers (qkv, w1, w3) and block-internal generators (o_proj, w2). Elementary norm arguments show residual-axis magnitude control bounds readers tightly but leaves w2's bilinear input bounded only by the trivial product of factor bounds; empirically, DR+sink collapses reader kurtosis while leaving generators essentially unchanged, and the reader-rescued W4A4 residue is flat at ~0.28 nats across three matched checkpoints with Delta-remove(w2) dominating. We present DR+sink as a training-time probe rather than a deployment proposal: a post-hoc alternative (Per-Linear QuaRot) nearly matches it on the reader axis. Full QuaRot -- adding online per-head value Hadamard plus online w2-input rotation -- does not close the gap either, directly testing the prediction that orthogonal rotation cannot bound the bilinear SwiGLU tail. Claims are specific to our 300M, 5B-token, single-seed setting, and our experiments do not isolate the partition from the hinge.
  • ✇cs.AI, q-bio.NC updates on arXiv.org
  • Copy-as-Decode: Grammar-Constrained Parallel Prefill for LLM Editing Ziyang Liu
    arXiv:2604.18170v2 Announce Type: replace-cross Abstract: LLMs edit text and code by autoregressively regenerating the full output, even when most tokens appear verbatim in the input. We study Copy-as-Decode, a decoding-layer mechanism that recasts edit generation as structured decoding over a two-primitive grammar: references an input line range, ... emits new content. A token-level FSM guarantees syntactic validity, and a serving-layer primitive updates the KV cache for each copy span via a
     
  • 🏷️
  • ↗️

Copy-as-Decode: Grammar-Constrained Parallel Prefill for LLM Editing

✇cs.AI, q-bio.NC updates on arXiv.org
By: Ziyang Liu
26 May 2026 at 12:00
arXiv:2604.18170v2 Announce Type: replace-cross Abstract: LLMs edit text and code by autoregressively regenerating the full output, even when most tokens appear verbatim in the input. We study Copy-as-Decode, a decoding-layer mechanism that recasts edit generation as structured decoding over a two-primitive grammar: references an input line range, ... emits new content. A token-level FSM guarantees syntactic validity, and a serving-layer primitive updates the KV cache for each copy span via a single parallel-prefill forward rather than $N$ autoregressive steps -- sharing the parallel-forward kernel of speculative decoding but with input tokens as the draft and program-enforced acceptance replacing probabilistic verification. We report an upper-bound analysis that requires no end-to-end training. (i) Kernel speedup: on Qwen2.5-{1.5B, 7B}, copying $N$ tokens via parallel prefill is $6.8\times$--$303\times$ faster than autoregressive ($N \in [8, 512]$, A100 80GB bf16). (ii) Copy ceiling: on ProbeEdit and HumanEvalPack-Fix (Py/JS), $74$--$98\%$ of gold tokens are reachable under the line-level primitive; composed with the empirical kernel over each corpus's span histogram this yields a closed-form wall-clock bound of $29.0\times / 3.4\times / 4.2\times$ ($13.0\times$ pooled). A token-level extension reaches $91$--$99\%$ coverage with $4.5\times$--$6.5\times$ floors. (iii) Pipeline losslessness: oracle programs round-trip through the deterministic resolver on all $482$ cases, localizing any downstream failure to span selection rather than the mechanism. A perturbation study shows pooled EM drops from $100\%$ to $15.48\%$ under off-by-one noise. A fine-tuning pilot on Qwen2.5-Coder-1.5B lifts HEvalFix-Py EM from $0/33$ (untrained) to $12$--$17\%$, a learnability signal, not a production selector. Batched-serving integration and multi-file coverage are scoped as follow-up.

Committed SAE-Feature Traces for Audited-Session Substitution Detection in Hosted LLMs

✇cs.AI, q-bio.NC updates on arXiv.org
By: Ziyang Liu
26 May 2026 at 12:00
arXiv:2604.18179v3 Announce Type: replace-cross Abstract: Hosted-LLM providers have a silent-substitution incentive: advertise a stronger model while serving cheaper replies. Probe-after-return schemes such as SVIP leave a parallel-serve side-channel, since a dishonest provider can route the verifier's probe to the advertised model while serving ordinary users from a substitute. We propose a commit-open protocol that closes this gap. Before any opening request, the provider commits via a Merkle tree to a per-position sparse-autoencoder (SAE) feature-trace sketch of its served output at a published probe layer. A verifier opens random positions, scores them against a public named-circuit probe library calibrated with cross-backend noise, and decides with a fixed-threshold joint-consistency z-score rule. We instantiate the protocol on three backbones -- Qwen3-1.7B, Gemma-2-2B, and a 4.5x scale-up to Gemma-2-9B with a 131k-feature SAE. Of 17 attackers spanning same-family lifts, cross-family substitutes, and rank-

Chain of Evidence: Pixel-Level Visual Attribution for Iterative Retrieval-Augmented Generation

✇cs.AI, q-bio.NC updates on arXiv.org
By: Peiyang Liu · Ziqiang Cui · Xi Wang · Di Liang · Wei Ye
26 May 2026 at 12:00
arXiv:2605.01284v2 Announce Type: replace-cross Abstract: Iterative Retrieval-Augmented Generation (iRAG) has emerged as a powerful paradigm for answering complex multi-hop questions by progressively retrieving and reasoning over external documents. However, current systems predominantly operate on parsed text, which creates two critical bottlenecks: (1) \textit{Coarse-grained attribution}, where users are burdened with manually locating evidence within lengthy documents based on vague text-level citations; and (2) \textit{Visual semantic loss}, where the conversion of visually rich documents (e.g., slides, PDFs with charts) into text discards spatial logic and layout cues essential for reasoning. To bridge this gap, we present \textbf{Chain of Evidence (CoE)}, a retriever-agnostic visual attribution framework that leverages Vision-Language Models to reason directly over screenshots of retrieved document candidates. CoE eliminates format-specific parsing and outputs precise bounding boxes, visualizing the complete reasoning chain within the retrieved candidate set. We evaluate CoE on two distinct benchmarks: \textbf{Wiki-CoE}, a large-scale dataset of structured web pages derived from 2WikiMultiHopQA, and \textbf{SlideVQA}, a challenging dataset of presentation slides featuring complex diagrams and free-form layouts. Experiments demonstrate that fine-tuned Qwen3-VL-8B-Instruct achieves robust performance, significantly outperforming text-based baselines in scenarios requiring visual layout understanding, while establishing a retriever-agnostic solution for pixel-level interpretable iRAG. Our code is available at https://github.com/PeiYangLiu/CoE.git.

Exploring the prognostic role of senescence-related genes in gastric cancer through multi-omics integration and machine learning

✇Omics in Gastric
By: Yangkun Cao · Dongjie Li · Li Bao · Xiaoling Ding · Yongzhen Guo · Xiaobo Ni · Yang Liu
9 May 2026 at 18:00

Hum Genomics. 2026 May 9. doi: 10.1186/s40246-026-00979-y. Online ahead of print.

ABSTRACT

Cellular senescence plays a context-dependent role in gastric cancer (GC), functioning both through tumor-suppressive arrest and the tumor-promoting senescence-associated secretory phenotype. However, its systematic integration into prognostic models remains limited. Here, we develop a novel interpretable framework to identify and validate a robust senescence-related gene signature for GC prognosis. We first introduce a dual-model interpretable feature selection strategy that integrates a biologically informed Kolmogorov-Arnold Network with a tabular foundation model to identify cancer-associated senescence genes. From the initial candidates, an ensemble of ten machine learning algorithms distills a core 4-gene signature to construct a Senescence Risk Score (SRS). The SRS proves to be a powerful and independent prognostic indicator, effectively stratifies patients into high- and low-risk groups with distinct overall survival across multiple cohorts. High-risk patients exhibit an "immune-hot" but potentially dysfunctional tumor microenvironment, characterized by enriched immune cell infiltration, elevated checkpoint expression, and distinct metabolic reprogramming favoring pathways such as angiogenesis and epithelial-mesenchymal transition (EMT). Furthermore, the SRS correlates with differential somatic mutation profiles and suggests potential sensitivity to specific chemotherapeutic agents. In vitro functional assays confirmed the oncogenic role of SERPINE1, a top-ranked core gene, in promoting GC cell proliferation. Regulatory network analysis revealed potential upstream transcription factors and miRNAs governing the signature. Collectively, we present a validated senescence-related prognostic signature that enables effective risk stratification of patients with gastric cancer.

PMID:42106891 | DOI:10.1186/s40246-026-00979-y

XPO1 inhibitor KPT-330 disrupts the core transcriptional regulatory circuitry of dedifferentiated liposarcoma by modulating the translation process

✇Oncogene - Issue - nature.com science feeds
By: Xiaorui Fan · Ying Zhang · Zhengming Yang · Tuan Zea Tan · Xingze Huang · Suya Zheng · Jiyang Liu · Long Xie · Ting Tao · Victor Kwanmin Lee · Chao Yu · H. Phillip Koeffler · Ye Chen · Liang Xu
16 April 2026 at 08:00

Oncogene, Published online: 16 April 2026; doi:10.1038/s41388-026-03794-w

XPO1 inhibitor KPT-330 disrupts the core transcriptional regulatory circuitry of dedifferentiated liposarcoma by modulating the translation process
Received — 8 April 2026 ⏭

LLMs Judge Themselves: A Game-Theoretic Framework for Human-Aligned Evaluation

✇cs.AI, q-bio.NC updates on arXiv.org
By: Gao Yang · Yuhang Liu · Siyu Miao · Xinyue Liang · Zhengyang Liu · Heyan Huang
7 April 2026 at 12:00
arXiv:2510.15746v2 Announce Type: replace-cross Abstract: Ideal or real - that is the question.In this work, we explore whether principles from game theory can be effectively applied to the evaluation of large language models (LLMs). This inquiry is motivated by the growing inadequacy of conventional evaluation practices, which often rely on fixed-format tasks with reference answers and struggle to capture the nuanced, subjective, and open-ended nature of modern LLM behavior. To address these challenges, we propose a novel alternative: automatic mutual evaluation, where LLMs assess each other's output through self-play and peer review. These peer assessments are then systematically compared with human voting behavior to evaluate their alignment with human judgment. Our framework incorporates game-theoretic voting algorithms to aggregate peer reviews, enabling a principled investigation into whether model-generated rankings reflect human preferences. Empirical results reveal both convergences and divergences between theoretical predictions and human evaluations, offering valuable insights into the promises and limitations of mutual evaluation. To the best of our knowledge, this is the first work to jointly integrate mutual evaluation, game-theoretic aggregation, and human-grounded validation for evaluating the capabilities of LLMs.
Received — 4 April 2026 ⏭

NED-Tree: Bridging the Semantic Gap with Nonlinear Element Decomposition Tree for LLM Nonlinear Optimization Modeling

✇cs.AI, q-bio.NC updates on arXiv.org
By: Zhijing Hu · Yufan Deng · Haoyang Liu · Changjun Fan
3 April 2026 at 12:00
arXiv:2604.01588v1 Announce Type: new Abstract: Automating the translation of Operations Research (OR) problems from natural language to executable models is a critical challenge. While Large Language Models (LLMs) have shown promise in linear tasks, they suffer from severe performance degradation in real-world nonlinear scenarios due to semantic misalignment between mathematical formulations and solver codes, as well as unstable information extraction. In this study, we introduce NED-Tree, a systematic framework designed to bridge the semantic gap. NED-Tree employs (a) a sentence-by-sentence extraction strategy to ensure robust parameter mapping and traceability; and (b) a recursive tree-based structure that adaptively decomposes complex nonlinear terms into solver-compatible sub-elements. Additionally, we present NEXTOR, a novel benchmark specifically designed for complex nonlinear, extensive-constraint OR problems. Experiments across 10 benchmarks demonstrate that NED-Tree establishes a new state-of-the-art with 72.51% average accuracy, NED-Tree is the first framework that drives LLMs to resolve nonlinear modeling difficulties through element decomposition, achieving alignment between modeling semantics and code semantics. The NED-Tree framework and benchmark are accessible in the anonymous repository https://anonymous.4open.science/r/NORA-NEXTOR.

Integrated spatial transcriptomics and pan-cancer XGBoost modeling uncover spatial drivers of immune exclusion and predict immunotherapy response

✇(Multiomics OR Omics) AND (Lung OR gastric OR Hepatocellular)
By: Hongying Zhao · Wangyang Liu · Haotian Xu · Lu Wang · Zushun Chen · Yanwu Sun · Li Wang
2 April 2026 at 18:00

Cancer Immunol Immunother. 2026 Apr 2;75(4):131. doi: 10.1007/s00262-026-04374-3.

ABSTRACT

Immunotherapy has revolutionized cancer treatment, yet characterizing the spatial complexity of the tumor immune microenvironment remains a challenge. In this study, we established a comprehensive computational framework integrating multi-omics profiling across 27 cancer types to decode immune-related non-coding RNA regulatory networks. Moving beyond traditional bulk analysis, we utilized spatial transcriptomics to dissect the spatial localization of these regulators. We identified the SNHG6-BIRC5 axis as a critical driver of the "immune-cold" phenotype in lung adenocarcinoma. We provide visual evidence that this axis localizes to tumor nests and negatively correlates with T- cell infiltration, elucidating a mechanism of spatial immune exclusion. Validating the clinical relevance of these findings, genome-scale CRISPR-Cas9 screening data confirmed the functional essentiality of these targets for cancer cell survival. Furthermore, pharmacogenomic analysis revealed that high expression of this axis correlates with sensitivity to chemotherapy agents like Vinblastine, suggesting a potential stratification strategy for patients with immune-excluded tumors. To expand the clinical utility to immunotherapy prediction, we developed a pan-cancer XGBoost machine learning model incorporating 14 high-performance regulatory features. This model achieved robust performance in distinguishing immunotherapy responders from non-responders with an AUC of 0.771, outperforming traditional markers such as PD-L1. Collectively, this study highlights spatial determinants of immune exclusion and chemotherapy sensitivity- and presents a generalized machine- learning tool for precision immunotherapy stratification. The developed online resource is freely available to facilitate community-wide biomarker discovery.

PMID:41925746 | DOI:10.1007/s00262-026-04374-3

❌