The paper shows that large language models (LLMs) naturally organize their hidden state manifolds into small‑world networks, enabling efficient multi‑hop reasoning. By converting similarity matrices into unweighted graphs, the authors trace connectivity between distant semantic anchors and find a sharp topological phase transition: deep reasoning layers compress conceptual distances into paths bounded by six semantic hops, while early syntactic layers remain fragmented. The framework is applied to zero‑shot hallucination detection in Retrieval‑Augmented Generation, revealing that factual generations preserve a ~3‑hop structure, whereas hallucinations collapse the topology.
By Md. Faiyaz Abdullah Sayeedi
The paper introduces a new object detection approach that protects sensitive visual data in test images. It is the first to apply perceptual encryption to object detection, leveraging the embedding structure of Vision Transformers and a key-based domain adaptation to maintain accuracy comparable to unprotected models. Experiments on ViTdet confirm the method’s effectiveness in both accuracy and visual protection.
By Homare Sueyoshi, Kiyoshi Nishikawa, Hitoshi Kiya
The paper introduces a topological method for detecting Hopf bifurcations directly from scalar time series. It combines delay-coordinate reconstruction with persistent homology, using the maximum persistence of one‑dimensional homology classes as a scalar descriptor of cyclic structure. Finite‑resolution persistence bounds are derived for the supercritical Hopf case, and a derivative‑based estimator localizes the critical parameter; the method is tested on the Hopf normal form, Lorenz system, and a reduced Belousov–Zhabotinsky model.
By Jhonathan Barrios, Y\'asser Ech\'avez, Carlos F. \'Alvarez
Nexus introduces a depth‑adaptive KV‑cache splicing and retrieval‑decoupled tool routing mechanism for agentic large language models that reduces the time‑to‑first‑token (TTFT) by decoupling tool routing from the expensive schema re‑encoding step. It uses an INT8 semantic lookaside buffer to select tools via retrieval and generates arguments from a compressed textual signature, maintaining about 89% routing accuracy even as the tool registry scales to 250 tools. Additionally, Nexus can splice compiled schema KV blocks into the live context, repairing the seam with a depth‑adaptive suffix redecode when rotary position embedding drift exceeds a threshold, ensuring output fidelity while achieving up to 1.7× TTFT speedup at moderate depth.
By Mustafa Arslan
arXiv:2608. 20810v1 Announce Type: cross Abstract: Multimodal information systems increasingly route generated visual content back through the same vision-language index that informed its production, so the output must remain retrievable by the queries it was meant to serve.
By Guangyuan Dong, Chuang Liu, Yangchen Zeng, Haoyu Wang, Xiaoyang Yu, Pinlong Zhao, Yuchao Hou, Ziwei Li, Zheng Lin
ZenGen is a comprehensive framework designed to develop socially intelligent large language models (LLMs). It introduces SoMBench, a psychology‑grounded benchmark with 3 primary and 17 secondary dimensions, and shows that current LLMs perform poorly on social cognition tasks. ZenGen combines supervised fine‑tuning, on‑policy distillation, and rubric‑based reinforcement learning to improve internalization, while Actio provides deployment‑time grounding with typed supports such as PRISM, Starling, SAGE, and gated RAG, yielding significant performance gains across benchmarks.
By ZenGen Team, Ao Xiang, Bi Jingping, Chen Jiahui, Chen Lehan, Chen Yilin, Cheng Xueqi, Fan Yixing, Gan Kairong, Gao Haowen, Gao Jinhua, Gao Shuxuan, Gong Chang, Guo Jiafeng, Guo Ruijie, Han Zhouyu, He Guangfu, He Yichun, Jiang Shuo, Jing Shaoling, Jing Ya, Lei Chenhao, Lei Yan, Li Anqi, Li Chengao, Li Haoyu, Li Shitian, Liang Xinjian, Liu Zhaoge, Lyu Xingyu, Nie Zhuwei, Pang Liang, Quan Zeping, Shan Shiguang, Shen Huawei, Tang Xinran, Tian Feng, Wang Qian, Wang Ruiping, Wang Xiaohong, Xia Zaiyu, Xiao Yi, Xu Jiayuan, Xu Kehan, Xu Qianqian, Xu Tianyu, Xu Yongjun, Yang Haoming, Yang Jun, Yao Di, Yu Xiaoming, Zhang Futong, Zhang Jie, Zhang Shixuan, Zhang Yuxuan, Zhao Xinyu, Zhao Zhuoran, Zhong Yunfei, Zhu Shengyu
The paper proposes a single hierarchical Semantic ID (SID) system to unify product identification across multiple merchants in e-commerce. By learning SID representations from product content, the authors demonstrate that ranking algorithms can aggregate consumer affinity and product performance over SID prefixes, improving offline relevance and online engagement. For query reformulation, SID concepts guide navigation and refinement, yielding better intent preservation and higher-quality suggestions compared to taxonomy or raw query transitions.
By Steven Xu, Sanjyot Thete, Saathvik Dirisala, Raghav Saboo, Nimesh Sinha, Leo Shao, Elyse Winer, Sudeep Das, Martin Wang, Kyle MacDonald
An integrated, web‑based platform has been developed to process diffusion‑weighted imaging (DWI) from MR‑guided linear accelerators and provide structured, literature‑grounded clinical interpretations. The system combines a deep‑learning pipeline for distortion correction, denoising, and IVIM/ADC fitting with a retrieval‑augmented generation (RAG) agent that references a curated knowledge base and traces each statement to its source. Independent expert ratings of nine glioblastoma cases showed high scores for clinical reasoning, citation quality, and overall utility, with a mean rating of 4.65 out of 5.
By Yunxiang Li, Yan Dai, Yen-Peng Liao, Jie Deng, Jill B De Vis, You Zhang
The paper proposes a method for automated research‑idea generation that preserves the typed structure of scientific papers by modeling each paper as a small category with typed research entities as objects and asserted relations as morphisms. It introduces a three‑layer algorithm—categorical signature clustering, a functor‑preservation gate, and a six‑axis LLM plausibility judge—to identify cross‑domain analogies that maintain relation chains. Experiments on tens of thousands of papers show the categorical gate filters candidates at a 17:1 ratio while keeping a falsifier rate above 83%, and it logs rejected candidates with detailed rationale.
By Yuchen Wang, Zhongzhi Luan
SENTRY is a deterministic risk assessment platform for IT change management that replaces subjective questionnaires with a machine‑learning pipeline using XGBoost and retrieval‑augmented generation. It integrates structured metadata, dependency graphs, incident history, and a hybrid semantic‑lexical search of change request text, compressing the unstructured signal into a single feature for model inference. The system delivers a ROC AUC of 0.87, 85% accuracy, and detects high‑risk changes 3.25 times faster than existing processes.
By Daniel Arulpragasam, Christer Henrysson, Ella Ly, Deepika Anbalagan, Leo Feng
arXiv:2608. 20820v1 Announce Type: new Abstract: Large language models (LLMs) are vulnerable to multi-turn jailbreak attacks that progressively manipulate conversation context.
By Yang Liu, Bin Chong, Wenkai Yang, Shuai Zhang, Yancheng Chen, Feiyu Han, GuoZhen, Cheng Zhang, Huaibing Xie, Changze Lv, Shihan Dou, Pluto Zhou
The paper introduces the Knowledge-Driven Analytics Framework (KDAF), an ontology‑driven approach for large language model analytics in enterprise finance that prioritizes auditability. KDAF constructs knowledge systems through six iterative stages and retrieves evidence via Context‑Aware Relevance Propagation (CARP), ensuring each fact includes its relationship type, confidence, and source lineage. Evaluation on FinanceBench shows that while retrieval improves accuracy marginally, KDAF significantly outperforms other methods in citation traceability and provenance completeness, demonstrating that auditability is the key advantage of ontology‑grounded retrieval.
By Sergiy Lunyakin
The paper investigates a failure mode in Graph-JEPA, a joint‑embedding predictive model trained on a large scientific‑reasoning graph. Despite achieving high linear‑probe accuracy and effective rank, the learned representation contains almost no usable instance information, as shown by retrieval metrics. The authors diagnose the issue to variance allocation in the objective, propose a repair that restores near‑perfect information recovery, and demonstrate that the problem persists even after repair, highlighting limitations in the evaluation metrics used.
By Gollam Rabby, S\"oren Auer
arXiv:2509.20377v2 Announce Type: replace-cross
Abstract: Retrieval-Augmented Generation (RAG) has significantly improved the performance of large language models (LLMs) on knowledge-intensive tasks...
By Tomoaki Isoda
The paper evaluates whether zero‑shot large language models (LLMs) can replace fine‑tuned natural language understanding (NLU) classifiers for intent detection in conversational systems. Experiments on ATIS and CLINC150 show that fine‑tuned RoBERTa outperforms Claude Haiku zero‑shot when ample in‑domain labels are available, but the LLM matches the supervised model on the broader 150‑intent CLINC150 schema. The authors identify three production‑relevant scenarios where LLMs excel—out‑of‑scope detection, robustness to ASR noise, and dynamic per‑deployment schemas—and present a decision framework to guide practitioners.
whyItMatters:"The study provides concrete evidence and a practical framework for deciding when to deploy LLMs versus fine‑tuned models in real‑world conversational applications."
By Carson Rodrigues, Oysturn Vas
arXiv:2608.21252v1 Announce Type: cross
Abstract: Question answering (QA) over long, connected documents remains challenging because relevant evidence may span multiple entities and their relationshi...
By Xuanyu Meng, Jiashuo Sun, Jash Rajesh Parekh, Jiawei Han
The paper presents an end‑to‑end framework for extracting and clustering trilingual Sri Lankan parliamentary debates in Sinhala, Tamil, and English. Using LLM‑based text extraction, multilingual embeddings, and density‑based clustering, the authors recover 30 macro‑topics with a cluster purity of 0.673. The temporal patterns of these topics align with major national events such as the 2019 Easter attacks and the 2022 economic crisis, demonstrating the method’s effectiveness where traditional LDA fails.
By Himath Dhanapala, Haren Daishika, Himandhi Kuruppu, Sithija Seneviratne, Ashini Kavindya, Patalee Narasinghe, Sandeepa Weerasekara, Nisansa de Silva, Sandareka Wickramanayake
PrimeAgentOrchestrator (PAO) is a system that spawns new instances of Claude Code, Anthropic’s terminal-based coding agent, pre‑loaded with relevant memories from a user’s personal databases. PAO queries a PostgreSQL entity‑observation database and a Cloudflare Worker semantic search index in parallel, fuses the results, and injects the compiled briefing into the agent’s filesystem to exploit its auto‑read configuration. The paper reports four months of deployment, detailing three generations of context delivery mechanisms, the failure modes that prompted each redesign, and the engineering tradeoffs involved in integrating heterogeneous memory systems instead of building a unified one.
By Myron Koch (Peak Summit Labs)
The paper argues that retrieval‑augmented question‑answering systems should perform semantic compilation at ingest time rather than re‑deriving meaning at query time. By building a maintained structure—incrementally updated embeddings and validated atomic claims—read operations become far cheaper, with experimental results showing higher accuracy and lower token usage compared to traditional chunk‑based retrieval. The authors present two proofs: cheaper incremental updates and superior performance on broadcast‑interview transcripts, suggesting a new systems agenda for compilation and read planning.
By Kyle Wild, Yusuke Takahashi, Asako Uraki
NeuroStrata is a deep learning framework that analyzes mental stress by modeling the temporal evolution of frequency‑specific directed connectivity in EEG signals using Time‑Varying Partial Directed Coherence (TV‑PDC). It transforms TV‑PDC connectivity maps into deep embeddings with pretrained CNNs and Vision Transformers, then classifies them with lightweight machine learning models. Experiments on the SAM 40 dataset show that beta‑band connectivity yields the highest accuracy (97.3 %) with a ViT backbone, while alpha‑band connectivity remains consistently stable, and that stress‑related connectivity signatures consolidate in mid‑to‑late temporal windows.
By Sayantan Acharya, Hamzeh Asgharnezhad, Abbas Khosravi, Douglas Creighton, Roohallah Alizadehsani, U Rajendra Acharya