❌

Normal view

Received β€” 27 May 2026 ⏭

EchoDistill:Alignment Noisy-to-Clean Self-Distillation for Robust Audio LLMs

βœ‡cs.AI, q-bio.NC updates on arXiv.org
By: Liang Lin Β· Chunxi Luo Β· Kaiwen Luo Β· Jie Zhang Β· Jin Wang Β· Yuanhe Zhang Β· Cai Yuchen Β· Qiankun Li Β· Gongli Xi Β· Zhenhong Zhou Β· Kun Wang Β· Junhao Dong
26 May 2026 at 12:00
arXiv:2605.23954v1 Announce Type: cross Abstract: Audio Large Language Models (ALLMs) are highly vulnerable to real-world noise, which often induces severe semantic drift and hallucinations. Existing robustness methods primarily rely on waveform-level acoustic enhancement, answer-level supervision, or the internal suppression of noise representations. To address these issues, we propose echodistill, an alignment-based noisy-to-clean self-distillation framework. Echodistill leverages a frozen clean-audio teacher to provide semantic references for an inference-time noisy-audio student. Specifically, the student samples candidate responses under noisy conditions to expose its test-time behavior. These trajectories are then optimized via group-relative policy optimization (GRPO), where the token-level consistency with the teacher acts as a reward bonus. By aligning the noisy student's candidate responses with clean semantic evidence, and applying audio-aware reward shaping, our method encourages reasoning trajectories that are both correct and genuinely acoustically grounded. Echodistill significantly improves the semantic reliability and task performance of Audio LLMs under complex noise, without introducing any additional inference costs. Extensive experiments show that: (I) Compared with the strongest baseline, echodistill achieves average improvements of 4.18\%$\uparrow$ in GSR under strong noise. (II) Ablation results on Qwen-Omni further show that echodistill improves over the GRPO-only variant by 3.02\%$\uparrow$ in Acc, 3.89\%$\uparrow$ in Noisy, and 4.53\%$\uparrow$ in GSR on average. Our codes are available at https://anonymous.4open.science/r/echodistill-10DE.

AutoSG: LLM-Driven Solver Generation Solely from Task Prompts for Expensive Optimization

βœ‡cs.AI, q-bio.NC updates on arXiv.org
By: Haoran Gu Β· Handing Wang Β· Yi Mei Β· Mengjie Zhang
26 May 2026 at 12:00
arXiv:2605.25658v1 Announce Type: cross Abstract: Expensive optimization tasks are ubiquitous in real-world applications, demanding highly specialized solvers. While LLM-driven automated solver generation shows promise, current paradigms face three critical issues when tackling expensive optimization: factual hallucinations due to deficient domain knowledge, the frequent dismantling of previously established locally optimal structures during refinement, and the prohibitive evaluation costs alongside restricted generalization caused by executing on training instances. To address these issues, we introduce AutoSG, a fully automated workflow directly translating natural language prompts into executable customized solvers. AutoSG features three core innovations: a retrieval-augmented solver generation module strictly grounding code in verified literature; a one-step self-refinement operator introducing task-specific improvements while preserving critical structural components; and an instance-free Elo-based LLM-as-a-Judge evaluation mechanism rapidly establishing global rankings. Extensive evaluations across diverse expensive optimization tasks confirm AutoSG significantly outperforms human-designed state-of-the-art frameworks and existing LLM-generated solvers.

HiTeC: Hierarchical Contrastive Learning on Text-Attributed Hypergraph with Semantic-Aware Augmentation

βœ‡cs.AI, q-bio.NC updates on arXiv.org
By: Mengting Pan Β· Fan Li Β· Chen Chen Β· Xiaoyang Wang Β· Wenjie Zhang
26 May 2026 at 12:00
arXiv:2508.03104v3 Announce Type: replace-cross Abstract: Contrastive learning (CL) has become a dominant paradigm for self-supervised hypergraph learning, enabling effective training without costly labels. However, node entities in real-world hypergraphs are often associated with rich textual information, which has been largely ignored in prior works. Directly applying existing CL-based methods to such text-attributed hypergraphs (TAHGs) leads to three key limitations: (1) The common use of graph-agnostic text encoders fails to capture the correlations between textual semantics and hypergraph topology, resulting in less expressive representations. (2) Their reliance on random data augmentations introduces noise and weakens the contrastive signals. (3) The primary focus on node- and hyperedge-level contrastive signals limits the ability to capture long-range dependencies, which is essential for effective representation learning. To address these challenges, we introduce HiTeC, a two-stage hierarchical contrastive learning framework for effective self-supervised learning on TAHGs. In the first stage, we pre-train the text encoder with a structure-aware contrastive objective to overcome the graph-agnostic nature of conventional methods. In the second stage, we begin by introducing semantic-aware augmentations, including structure-contextualized text augmentation and semantic-aware hyperedge dropping, to facilitate informative view generation. Subsequently, we propose a multi-scale contrastive loss with an $s$-walk-based subgraph-level objective to capture long-range dependencies. Extensive experiments on six real-world datasets validate the effectiveness of our proposed method.

HiGraph: A Large-Scale Hierarchical Graph Dataset for Malware Analysis

βœ‡cs.AI, q-bio.NC updates on arXiv.org
By: Han Chen Β· Hanchen Wang Β· Hongmei Chen Β· Ying Zhang Β· Lu Qin Β· Wenjie Zhang
26 May 2026 at 12:00
arXiv:2509.02113v2 Announce Type: replace-cross Abstract: The advancement of graph-based malware analysis is critically limited by the absence of large-scale datasets that capture the inherent hierarchical structure of software. Existing methods often oversimplify programs into single level graphs, failing to model the crucial semantic relationship between high-level functional interactions and low-level instruction logic. To bridge this gap, we introduce \dataset, the largest public hierarchical graph dataset for malware analysis, comprising over \textbf{200M} Control Flow Graphs (CFGs) nested within \textbf{595K} Function Call Graphs (FCGs). This two-level representation preserves structural semantics essential for building robust detectors resilient to code obfuscation and malware evolution. We demonstrate HiGraph's utility through a large-scale analysis that reveals distinct structural properties of benign and malicious software, establishing it as a foundational benchmark for the community. The dataset and tools are publicly available at https://higraph.org.

SNHG12 drives gastric cancer progression by activating the Wnt/beta-catenin-mediated serine synthesis pathway

βœ‡Omics in Gastric
By: Nianjie Zhang Β· Taolang Li Β· Kunming Wen
1 May 2026 at 18:00

J Transl Med. 2026 Apr 30;24(1):638. doi: 10.1186/s12967-026-08173-3.

ABSTRACT

BACKGROUND: Metabolic reprogramming is a hallmark of gastric cancer and is essential for sustaining rapid proliferation and malignant progression. The serine synthesis pathway (SSP), a key branch of glycolysis coupled to one-carbon metabolism (OCM), plays a central role in nucleotide biosynthesis, redox homeostasis, and epigenetic regulation. Although aberrant SSP activation has been implicated in gastric cancer, its upstream regulatory mechanisms remain poorly defined. Long non-coding RNAs (lncRNAs) have emerged as critical modulators of oncogenic signaling and metabolism. This study aimed to elucidate the role of the lncRNA SNHG12 in gastric cancer progression and to determine whether it drives metabolic reprogramming through the Wnt/Ξ²-catenin-SSP axis.

METHODS: SNHG12 expression and clinical relevance were analyzed using public datasets, clinical gastric cancer specimens, and cell lines. Gain- and loss-of-function experiments were performed to assess the effects of SNHG12 on proliferation, apoptosis, migration, and invasion. Transcriptomic profiling, targeted metabolomics, and integrative multi-omics analyses were used to characterize metabolic alterations. Pharmacological inhibition of SSP (NCT503) and Wnt/Ξ²-catenin signaling (IWR-1) was applied in vitro and in vivo. A subcutaneous xenograft mouse model was used to validate tumor-promoting effects and therapeutic responses.

RESULTS: SNHG12 was significantly upregulated in gastric cancer tissues and cell lines and was associated with poor overall and progression-free survival. Functionally, SNHG12 promoted gastric cancer cell proliferation, migration, and invasion while suppressing apoptosis. Transcriptomic and targeted metabolomic analyses revealed broad metabolic alterations associated with SNHG12, including changes in serine/one-carbon metabolism, purine biosynthesis, and glutathione-related pathways. Mechanistically, SNHG12 increased Wnt/Ξ²-catenin reporter activity, promoted Ξ²-catenin nuclear accumulation, and was accompanied by increased expression of key SSP-associated enzymes, including PHGDH, PSAT1, and SHMT2. Pharmacological inhibition of SSP or Wnt/Ξ²-catenin signaling partially reversed SNHG12-induced malignant phenotypes in vitro and suppressed tumor growth in xenograft models.

CONCLUSIONS: This study identifies SNHG12 as an important regulator of metabolic reprogramming in gastric cancer. Our data support a model in which SNHG12 promotes gastric cancer cell proliferation, invasion, and migration through SSP regulation, and suggest that its effects on the SSP may be mediated, at least in part, through modulation of SSP-associated enzymes via the Wnt/Ξ²-catenin pathway. These findings support SNHG12 as a candidate biomarker and a potential therapeutic target for combined metabolic and signaling-based interventions in gastric cancer.

PMID:42063161 | PMC:PMC13151230 | DOI:10.1186/s12967-026-08173-3

There are no more articles
❌