The study audits demographic bias across four deep knowledge tracing architectures—DKT, DKVMN, SAKT, and AKT—using two large public datasets (Eedi and OULAD). It finds that bias is context‑dependent: socioeconomic bias is significant on Eedi, while gender bias appears on OULAD for most models. The most accurate model, AKT, also exhibits the greatest bias, and standard mitigation techniques such as reweighting and adversarial debiasing fail to reduce bias without sacrificing accuracy.
By Dang Quang Minh, Nguyen Dung Son, Nguyen Huu Loi, Truong Viet Vu, Nguyen Thai Anh
The paper explores subdomain-aware dimensionality reduction for pretrained image embeddings, applying techniques such as PCA and LDA to compress representations within specific image subdomains. Results show that this targeted compression reduces space and computational complexity while improving accuracy compared to using full embeddings. Additionally, the study demonstrates that the compressed representations retain transfer learning capabilities across tasks.
By Poowanut Niamluang, Jittat Fakcharoenphol
TRACE is a training‑free, agentic retrieval framework that enables accountable source discovery in historical archives, addressing challenges such as OCR degradation and genre heterogeneity. Developed for the DECIDON project on French Third Republic political discourse, it is deployed internally for 24 researchers across six institutions. On the HistoriQA‑ThirdRepublic benchmark, TRACE achieves R@10 of 0.856 and MRR of 0.653, outperforming sparse, dense, graph‑based, and other agentic RAG baselines, especially on multi‑hop and cross‑corpus questions, while costing only about $0.02 per question.
By Donghan Bian (ENC, LRE), Marie Puren (LRE, ENC), Florian Cafiero (LRE, ENC)
The paper introduces the Modality Discrepancy Transformer (MDT), a model designed to detect ambivalence and hesitancy in clinical videos by capturing cross‑modal disagreement across facial, vocal, and linguistic signals. MDT expands a 6‑token representation to 9 tokens that include modality embeddings, absolute‑difference features, and Hadamard‑product discrepancy features, which are processed through Transformer self‑attention with FiLM‑based text conditioning and LoRA fine‑tuning. On the BAH dataset from the 3rd ABAW Challenge, MDT achieves a Macro F1 score of 0.7408 on the labelled test split and 0.7368 on the private leaderboard, surpassing the strongest baseline by over 10 points while training in under 20 minutes on a single GPU.
By Shiyu Luo, Yu Wang, Jiawen Huang, Zhaoxiang Xiao, Chenxi Huang, Qi Zhang, Bin Liu
The paper investigates how large‑language‑model (LLM) based AI agents mix latency, local resource usage, and container bottlenecks when processing user requests that involve remote LLM calls and local tool execution. By measuring three representative tasks—retrieval‑augmented question answering, web search, and software coding—the authors show that agents exhibit diverse resource dynamics, with concurrent requests revealing task‑specific bottlenecks in CPU, disk I/O, and memory. Leveraging these insights, they propose CPU‑aware tool admission and task‑aware CPU allocation, achieving up to a 5.4× speed‑up for CPU‑sensitive tasks and a 32% reduction in average latency across multiple tasks.
By Wonmi Choi, Minuk Park, Zhixiong Niu, Yongqiang Xiong, Chuck Yoo, Gyeongsik Yang
The paper introduces ESIA, an Earth Surface Immune System that detects and recognizes unknown anomalies in satellite imagery without prior category knowledge. It uses a non‑specific innate stage for rapid localization and a specific adaptive stage that matches image patches to text prompts via a multi‑modal model, achieving high F1 scores. The system adapts to new scenes in seconds and has been validated on a large global dataset, with applications to farmland degradation after the Kakhovka Dam collapse and burn severity assessment from the 2025 Palisades Fire.
By Jingtao Li, Qian Zhu, Xinyu Wang, Deren Li, Liangpei Zhang, Yanfei Zhong
The paper examines the geometry of semantic neighbourhood graphs as modal logic, translating this view into a statistical framework to differentiate human from AI-generated text. It models texts as worlds in a finite frame where accessibility is defined by the k‑nearest‑neighbour relation of transformer embeddings, and measures the frequencies of modal axioms (B, 4, 5, D) as validation degrees. A prompt‑balanced comparison shows consistently higher degrees for axioms 4 and 5 in AI‑generated corpora, and the study further introduces measures of groundedness and situatedness, recasting the analysis in a sequent‑style tableau setting.
"whyItMatters":"The work provides a quantitative, modal‑logic‑based method to detect structural differences between human and machine‑generated text, offering a new lens for evaluating AI language models."
By Simone Cuconato, Donato Ferrari
The paper introduces Compressed Active Subspaces (CAS), a scalable method for Bayesian inference in high‑dimensional models. CAS first compresses model parameters via a structured isometric embedding, then constructs the active subspace in this reduced space, dramatically lowering memory requirements. Experiments on neural networks of growing size show that CAS preserves predictive performance and provides robust uncertainty estimates while enabling inference where traditional active subspace methods fail.
By Thomas Flynn, Sanket Jantre, Byung-Jun Yoon, Kibaek Kim
The paper introduces DASE, a query engine designed to efficiently link unstructured data for multi-step reasoning tasks. DASE combines a multi-step reasoning model, a sparse materialized embedding-similarity join index (SemJI), and a co‑designed execution layer to perform multi‑attribute filtering, multi‑vector search, exact relational joins, and thresholded embedding‑similarity joins. In scientific discovery workloads, DASE outperforms traditional RDBMS, rerank, and vector‑database baselines by 6x to 46x in retrieval speed while maintaining comparable recall, and it serves as a high‑recall prefilter that reduces downstream LLM evaluation cost and improves accuracy on benchmarks such as SemBench E‑Commerce.
By Jiaming Liang, Haydn Jones, Jacob R. Gardner, Mark Yatskar, Zachary Ives
JEPA-Anything is a domain‑agnostic framework that uses orthogonal predictive factorization (OPF) to decompose latent targets into complementary factors, learn them via dedicated pathways, and recombine them for shared prediction. The method is evaluated across seven diverse domains—vision, biology, clinical trajectories, control, molecular dynamics, physical fields, and weather—showing improvements on 10 dynamics tasks, reduced error on Interventional Pong, and lowest one‑step and 100‑step molecular errors among compared methods. Experimental validation includes a factor‑nominated biological intervention that succeeded in cell co‑cultures, organoids, tumor fragments, and mice, and latent orbital modes that recover the Keplerian scaling exponent.
By Taoyong Cui, Zhongyao Wang, Xinyue Xu, Weiyang Liu, Zhaochen Yu, Yuying Zhang, Qiang Gao, Mengyue Yang, Wanli Ouyang, Pheng Ann Heng, Yingcheng Wu, Zhenfei Yin, Ling Yang
The paper introduces TrioRAG, a graph-free multimodal retrieval-augmented generation framework that combines evidence from the question, an anchor image, and a VLM-enhanced query via late fusion. It also presents AutoQA, a benchmark featuring noisy web-sourced images that require reasoning across manuals. TrioRAG outperforms graph-based systems on three benchmarks while cutting costs and speeding up inference by 1.6–2.3×.
By Tithi Rakshit, Hongkuan Zhou, Lavdim Halilaj, Yuqicheng Zhu
The paper introduces an end‑to‑end framework that automatically builds a business semantic layer from raw application logs. It uses a two‑stage abstraction: first, large language models identify high‑level business features with industry knowledge, and second, a structured pipeline refines data, retrieves relevant information, filters, clusters semantically, and assigns canonical names. Evaluation on production‑scale telemetry shows significant gains in semantic quality, noise reduction, and maintenance effort, with a 0.87 Cohen’s kappa in an LLM‑as‑Judge assessment.
By Yuanzhe Jia, Ali Anaissi
The paper introduces FRESH‑GEORANGE, a semantic‑spatial range retrieval system that separates source‑watermark freshness from optional record age. It uses geographic cells and semantic microblocks for pruning, and offers an exact mode that guarantees 100% recall and a certified mode that can stop early while providing a deterministic recall lower bound. A CPU pilot on 2,500 OpenFlights airport records demonstrates the system’s correctness, achieving high recall with a modest latency overhead compared to a spatial‑first exact baseline.
By Taimoor Ahmad
The paper introduces Sequential Contextual Fit (SCF), an embedding‑based metric that quantifies how well a current information state aligns with its recent context using a recency‑weighted similarity kernel. SCF is applied across diverse domains—including language, music, audiovisual emotion, gambling, activity recognition, and EEG—showing that lower contextual fit predicts longer processing times, larger affective or behavioural transitions, and stronger neural-state changes. These findings hold even after controlling for established predictors such as surprisal, prediction error, and sensory changes.
By Kun Sun, Rong Wang
The study investigates how document segmentation and chunk representation affect retrieval-augmented generation (RAG) for chemistry texts. Using the ChemQuests corpus, the authors benchmark 41 embedding models and evaluate them across five chunking strategies, seven chunk sizes, and various overlap settings. They find that embedding choice has the largest impact, with models like E5, BGE, and Nomic performing best, and recommend medium-to-large chunks with fixed-token, recursive-token, or hierarchical-section chunking and low overlap for effective chemistry-aware RAG.
By Mahmoud Amiri, Thomas Bocklitz
The article "The Life of a Token: from Words to Bits on the Wire" explores how large language models convert text into network traffic during training. It traces the transformation from words to tokens, then to vectors, and finally to binary streams that traverse high‑performance computing systems. Using Dante’s Divine Comedy as a case study, the tutorial examines how tokenization, embeddings, and parallelization affect the volume, structure, and timing of data exchanged across the network, and provides analytical traffic models and numerical examples to clarify the communication demands of LLM training.
By Davide Avesani (CEDRIC - ROC), Pengwenlong Gu (CEDRIC - ROC), Sotiris Skaperas (Cnam), Stefano Secci (CEDRIC - ROC)
The paper examines how deliberative (System 2) reasoning affects a Retrieval-Augmented Generation (RAG) model’s vulnerability to knowledge‑poisoning attacks. Using two metrics—Cordon Rate and Leakage Rate—it evaluates six model configurations on 200 SciFact questions. Results show that enabling reasoning lowers both Cordon and Leakage Rates for DeepSeek‑V4‑Flash, indicating reduced behavioral impact from poisoned evidence, though overall attack success increases.
By Mehrdad Ghassabi, Audrina Ebrahimi, Sadra Hakim, Hamidreza Baradaran Kashani
PrefixBench-H100 is a reproducible benchmark that evaluates how reusing prompt prefixes affects LLM serving performance on NVIDIA H100 GPUs. It tests two popular runtimes (vLLM and TensorRT-LLM) across varied workloads, measuring metrics such as time-to-first-token, latency, throughput, cache hits, and GPU memory usage. The study identifies when prefix reuse significantly reduces first‑token latency and when cache pressure diminishes those gains, noting that cache effectiveness is largely unaffected by concurrency or output length, while differences arise mainly in scheduling.
By Omkar Shewale, Deepak Kumar, Divakar Kumar Yadav
The paper proposes a new framework that improves spreadsheet chunking for large language model (LLM)-driven retrieval-augmented generation (RAG) systems by adding semantic cell annotations. This approach outperforms current state‑of‑the‑art methods but is limited by the inherent two‑dimensional, unstructured nature of spreadsheets, which cannot be fully captured by finite classification categories. The authors argue that future progress requires dimensionality‑reduction techniques to flatten spreadsheets into one‑dimensional text, simplifying downstream RAG interpretation and generation.
By Zofia Smole\'n
The paper introduces a Retrieval-Augmented Generation (RAG) framework for scientific image quality assessment, targeting both the understanding (SIQA-U) and scoring (SIQA-S) tracks of the SIQA challenge. It builds a multimodal index that merges textual semantics with fine‑grained visual features and employs a multi‑route retrieval and fusion mechanism to supply large language models with relevant reference cases, improving their evaluation of complex scientific images. The approach aligns well with human expert judgment and secured first place in the SIQA-U track at the ICME 2026 Grand Challenges.
By Yinuo Zhang, Bingshuo Liu, Zhiying Tu, Dianhui Chu, Qingbin Liu, Xi Chen, Jiang Bian, Xiaoyan Yu, Dianbo Sui