hLLM (Hungarian LLM) is a decoding strategy that efficiently produces ranked lists from large language models by decoding all ordinal positions in a single forward pass. It extracts an item‑position score matrix from the model’s hidden states, then applies the Hungarian algorithm to obtain a valid permutation, avoiding the sequential token generation of traditional autoregressive decoding. Experiments show that LoRA‑based fine‑tuning with teacher ranking distillation achieves 28 ms end‑to‑end inference, a 64× speed‑up while preserving ranking quality comparable to the teacher model.
By Emil Laftchiev, Prachi Agrawal, Moe Kayali, Bixing Yan, Qi Xu, Zijie Lei, Chen Qiu, Zhi Hua, Ke Li, Luke Simon
The paper introduces UnInfo, a test‑time adaptation method for vision‑language models like CLIP that addresses image corruption—a realistic distribution shift caused by sensor conditions. UnInfo leverages uniformity‑aware confidence maximization, information‑aware loss balancing, and knowledge distillation from an EMA teacher to preserve embedding uniformity and improve zero‑shot classification accuracy. Experiments show that UnInfo outperforms existing TTA methods on corrupted image datasets.
By Kazuki Adachi, Shin'ya Yamaguchi, Tomoki Hamagami
The paper extends the h‑plot multidimensional scaling technique to handle three‑way asymmetric dissimilarities, enabling the extraction of archetypal profiles and clustering in a unified Euclidean space. It provides an explicit eigenvector‑based solution that avoids local minima, is scale‑invariant, computationally efficient, and includes a straightforward goodness‑of‑fit assessment. The method is benchmarked against existing models and demonstrated on a financial dataset, with all data and code made publicly available for reproducibility.
By Mireia Mollar-Gumbau, Aleix Alcacer, Rafael Benitez, Vicente J. Bolos, Irene Epifanio
The paper introduces QUEST, a method for uncertain knowledge graph completion that adds no trainable parameters to the standard pipeline. QUEST first initializes entity embeddings using the smallest non‑trivial eigenvectors of the confidence‑weighted graph Laplacian, thereby preserving community and hub structure before training. It then applies an unbiased mini‑batch Dirichlet energy regularizer to enforce early‑stage structural consistency, leading to improved confidence and link prediction on most metric‑dataset pairs and eliminating instability spikes on dense graphs.
By Md Abrar Jahin, Taufikur Rahman Fuad, Jay Pujara, Craig A. Knoblock
The paper introduces a method that integrates action abstraction into policy optimization for reinforcement learning and generative flow networks. By iteratively identifying frequently used action subsequences in high‑reward trajectories and treating them as single high‑level actions, the approach expands the action space and improves sample efficiency. Experiments on synthetic and real‑world tasks show that this technique discovers diverse high‑reward states more effectively, especially on challenging exploration problems, and yields interpretable abstract actions that reflect the underlying reward structure.
By Oussama Boussif, L\'ena N\'ehale Ezzine, Joseph D Viviano, Micha{\l} Koziarski, Moksh Jain, Esmeralda S. Whitammer, Emmanuel Bengio, Rim Assouel, Yoshua Bengio
The REAP system tackles the AKBC Shared Task 2026, aiming to build knowledge bases from language models without fine‑tuning and within a 32‑B parameter budget. It uses structured chain‑of‑thought reasoning, relation‑specific queries, and a reasoning‑based empty‑set gate to elicit knowledge, then directly extracts it into valid JSON arrays. Evaluated on the test set with the Mistral‑Small‑24B‑Instruct‑2501 model, REAP achieves a macro‑F1 of 0.62, notably high scores on countryLandBordersCountry (0.95), companyTradesAtStockExchange (0.73), and hasArea (0.77).
By Thanh-Dan Bui, Thanh-Trung Do, Tuan-Phong Nguyen
The paper introduces a structured reasoning framework that leverages large language models (LLMs) for root cause analysis (RCA) in telecom networks. It organizes heterogeneous network telemetry into canonical contexts, enforces decision‑path reasoning, and produces evidence‑grounded explanations to improve fault identification. Experiments on two 5G RCA datasets, TeleLogs and TelecomTS, show that this approach consistently outperforms baseline techniques in diagnostic accuracy and decision consistency.
By Hao Zhou (Jianzhong), Mandar Kulkarni (Jianzhong), Hao Chen (Jianzhong), Yan Xin (Jianzhong), Charlie (Jianzhong), Zhang
PEARL is a new framework for inductive knowledge graph completion that treats relational paths as context-conditioned reasoning signals. It builds a query‑specific contextual subgraph from the query entities’ neighborhoods and uses a large language model‑guided retriever to select semantically relevant paths. By constructing a bipartite interaction graph over paths, contextual entities, and a global subgraph representation, and applying a dual‑view contrastive objective, PEARL adapts path embeddings to local and global structural evidence, achieving the best average Hits@10 on WN18RR, FB15k‑237, and NELL‑995.
By Yunchi Yang, Longlong Li, Cunquan Qu
The paper introduces a self‑referential retrosynthesis framework for explainable AI provenance forensics that works with fixed‑generator generative models. It uses a jointly optimized encoder‑decoder pair to embed client inputs, generate high‑fidelity outputs, and then verify provenance by comparing the resynthesized image to the original query. The method eliminates the need for watermarking or generator modifications while providing interpretable evidence of a model’s output origin.
By Yijie Lin, Ching-Chun Chang, Isao Echizen, Hui Li, Chin-Chen Chang
The paper presents an LLM-driven framework that splits peer reviews into argumentative segments, detects multiple co-occurring issues such as lazy thinking and lack of specificity, and generates targeted, guideline-aware feedback using issue-specific templates. An iterative, reranking-based generation algorithm refines the feedback, and a controlled rewriting study shows it can reduce guideline violations by up to 92.4%. The authors also release LazyReviewPlus, a multi-label dataset of 1,309 sentences annotated for detecting lazy thinking and lack of specificity.
By Sukannya Purkayastha, Qile Wan, Anne Lauscher, Lizhen Qu, Iryna Gurevych
The paper introduces DPA, a diffusion-based framework that decouples product-agnostic anomaly representations to enable zero-shot anomaly generation. By reusing real anomalies from existing source products and filtering them for plausibility, DPA learns product-irrelevant anomaly embeddings that can be transferred across products. An adaptive mask-guided pipeline and a training-free labeling module further refine the realism and localization of generated anomalies, leading to improved performance on MVTec-AD, VisA, and a new anomaly-transfer benchmark.
By Hang Yao, Yansheng Fu, Ming Liu, Zifei Yan, Yanli Ji, Hongzhi Zhang, Wangmeng Zuo
The paper presents a method for selecting sub‑networks of large language models for on‑premise factory assistants, showing that after structural compression and retrieval‑grounded adaptation, model size no longer predicts answer quality. By treating deployment as a post‑adaptation selection problem, the authors commit one sub‑network per device based on judged answer quality and measured on‑device throughput, while respecting a general‑capability floor and memory budget. A weight‑shared supernetwork trained with sandwich‑style in‑place distillation makes this selection inexpensive, and a manufacturing‑manual case study demonstrates that the approach recovers most of the quality loss while operating across heterogeneous edge tiers at low power.
By Vasileios Rizeakos, Georgios Paisios, Alexandros Machairas, Michael Birbas, Athanasios Bachoumis
The paper presents a lightweight ResNet-based two-stage cascade for passive acoustic monitoring of killer whales. First, it detects vocalizations, then it classifies confident detections into five eastern North Pacific ecotypes, abstaining on ambiguous calls. The pipeline achieves high macro‑F1 scores on the DCLDE 2027 dataset and improves real‑time inference speed, while active learning adapts the detector to new acoustic environments.
By Daniela Ruiz, Manuel Castellote, Zhongqi Miao, Carl Chalmers, Bruno Demuro, Rahul Dodhia, Pablo Arbelaez, Juan M. Lavista
WeaveMark is a new multi‑bit watermarking scheme for large language models that improves payload capacity, extraction accuracy, and text quality by using coded payload spreading, soft‑decision error‑correcting codes, and unbiased multilayer reweighting. It also adds zero‑bit layers for reliable detection of watermark presence. Experiments demonstrate significant gains, achieving an 89.8% match rate for 32‑bit messages at 200 tokens and maintaining 86.0% accuracy under 10% substitution attacks on 16‑bit messages, far outperforming the BiMark baseline.
By Gang-Hyun Park, Ju-Hyeong Lee, Hee-Youl Kwak, Dae-Young Yun
MARS introduces a multi‑layer, multi‑slot embedding framework for text‑video retrieval that constructs adaptive representation slots by combining hidden states from different decoder layers. By comparing corresponding text and video slots and aggregating their similarities, MARS captures fine‑grained cues that single‑token embeddings miss. A hard‑negative‑aware slot specialization objective further encourages slots to focus on discriminative matching cues, leading to state‑of‑the‑art results on four benchmarks.
By Uicheol Jung, Juyoung Hong, Geuntaek Lim, Yukyung Choi
The paper introduces DLM-One, a score‑distillation framework that enables one‑step sequence generation with continuous diffusion language models (DLMs). By aligning a student model’s outputs with a pretrained teacher DLM’s score function in the forward‑diffused noisy space, DLM-One removes the need for iterative refinement. Experiments across various DLM architectures show up to ~2000× speedup in sampling steps and ~500× in wall‑clock time while retaining competitive performance, and the authors propose an adversarially‑regularized two‑stage training scheme to mitigate student degeneration.
By Tianqi Chen, Shujian Zhang, Mingyuan Zhou
The paper introduces TAP-PER, a prefix‑based framework that learns compact user representations for large language model personalization. By encoding user preferences into lightweight prefix embeddings and incorporating temporal signals, TAP‑PER avoids the need for heavy per‑user adapters or prompt‑serialized histories. Experiments on six LaMP tasks show that TAP‑PER outperforms both prompt‑based and model‑based baselines while using far fewer per‑user parameters, enabling scalable personalization at large user scales.
By Heng Cao, Fan Zhang, Jian Yao, Yujie Zheng, Changlin Zhao, Lu Hao, Yuxuan Wei, Wangze Ni, Huaiyu Fu, Yuqian Sun, Xuyan Mo
ExecRetrieval is a new benchmark for code‑embedding retrieval that contains 939 Python tasks, each with a verified correct implementation and up to four single‑edit buggy distractors generated mechanically. The dataset allows direct testing of a retriever’s ability to functionally discriminate correct code from near‑clone incorrect code, rather than relying on lexical similarity. Experiments on 23 dense embeddings and BM25 show that while the best system can retrieve the correct code within the top 10 results, it often fails to rank the correct implementation first, with rank‑1 errors dominated by paired buggy variants.
By Aaryan Kapoor, Md Abdullah Al Hafiz Khan
CACTUS is a new backdoor attack for decentralized federated learning that uses mask‑guided, modality‑specific operators to convert label‑consistent semantic pairs into target‑directed representation shifts. By isolating trigger effects and applying them counterfactually to clean embeddings before peer aggregation, CACTUS can propagate backdoors across repeated aggregation rounds. Experiments on speech, text, tabular, and image tasks show that with 30% malicious nodes, CACTUS achieves a mean attack success rate of 51.2% on Speech Commands and the highest mean ASR among evaluated attacks on three of four modalities, with success varying by network topology and malicious‑node ratio.
By Chao Feng, Burkhard Stiller
The paper introduces HarmReduction, a benchmark for evaluating large language models (LLMs) on their ability to provide accurate and safe harm reduction information to people who use drugs (PWUD). The benchmark, HR-Basic, contains 2,160 question‑answer‑evidence pairs covering safety boundary checks, quantitative value provision, and polysubstance risk inference. Experiments show that even state‑of‑the‑art LLMs struggle with accuracy and can pose severe safety risks, underscoring the need for a dedicated evaluation framework.
By Kaixuan Wang, Chenxin Diao, Jason T. Jacques, Zhongliang Guo, Shuai Zhao