Human - Artificial Intelligence (HAI) collaboration in writing offers opportunities to enhance efficiency and boost student confidence; however, it also carries risks, such as reduced creativity, over - reliance on AI - generated content, and academic integrity (Kim & Lee, 2023) . While the ethical use of AI in education is widely acknowledged as a way to enhance student learning (Cotton et al., 2023; Foltynek et al., 2023), the rise of Unauthorised Content Generation (UCG) presents a significant challenge to academic misconduct. Measuring the extent and nature of HAI collaboration in academic contexts remains a critical challenge for educators, particularly as generative AI (genAI) tools become increasingly available and integrated into educational settings (Atchley et al., 2024; E. Oliveira et al., 2023) . Distinguishing AI - generated text from human - authored content is necessary for understanding student learning behaviours, supporting skill development, and maintaining academic integrity. Analysing student writing patterns can help educators evaluate how st udents engage with AI tools, track their writing skill progression, and identify areas where additional support is needed (Pan et al., 2025). Existing detection tools for AI - assisted misconduct often lack reliability, explainability, and resilience to circ umvention strategies such as paraphrasing (Cotton et al., 2023) . These challenges highlight the need for innovative, transparent, and robust approaches to address the unacknowledged use of genAI in HAI collaboration within academic writing (Kasneci et al., 2023) .

Large Language Models (LLMs) have great potential in the field of health care, yet they face great challenges in adapting to rapidly evolving medical knowledge. This can lead to outdated or contradictory treatment suggestions. This study investigated how LLMs respond to evolving clinical guidelines, focusing on concept drift and internal inconsistencies. We developed the DriftMedQA benchmark to simulate guideline evolution and assessed the temporal reliability of various LLMs. Our evaluation of seven state-of-the-art models across 4,290 scenarios demonstrated difficulties in rejecting outdated recommendations and frequently endorsing conflicting guidance. Additionally, we explored two mitigation strategies: Retrieval-Augmented Generation and preference fine-tuning via Direct Preference Optimization. While each method improved model performance, their combination led to the most consistent and reliable results. These findings underscore the need to improve LLM robustness to temporal shifts to ensure more dependable applications in clinical practice. The dataset is available at https://huggingface.co/datasets/RDBH/DriftMed.

Efficient text classification is essential for handling the increasing volume of academic publications. This study explores the use of pre-trained language models (PLMs), including BERT, SciBERT, BioBERT, and BlueBERT, fine-tuned on the Web of Science (WoS-46985) dataset for scientific text classification. To enhance performance, we augment the dataset by executing seven targeted queries in the WoS database, retrieving 1,000 articles per category aligned with WoS-46985's main classes. PLMs predict labels for this unlabeled data, and a hard-voting strategy combines predictions for improved accuracy and confidence. Fine-tuning on the expanded dataset with dynamic learning rates and early stopping significantly boosts classification accuracy, especially in specialized domains. Domain-specific models like SciBERT and BioBERT consistently outperform general-purpose models such as BERT. These findings underscore the efficacy of dataset augmentation, inference-driven label prediction, hard-voting, and fine-tuning techniques in creating robust and scalable solutions for automated academic text classification.

In-context learning (ICL) has emerged as a powerful capability of large language models (LLMs), enabling them to perform new tasks based on a few provided examples without explicit fine-tuning. Despite their impressive adaptability, these models remain vulnerable to subtle adversarial perturbations and exhibit unpredictable behavior when faced with linguistic variations. Inspired by software testing principles, we introduce a software testing-inspired framework, called MMT4NL, for evaluating the trustworthiness of in-context learning by utilizing adversarial perturbations and software testing techniques. It includes diverse evaluation aspects of linguistic capabilities for testing the ICL capabilities of LLMs. MMT4NL is built around the idea of crafting metamorphic adversarial examples from a test set in order to quantify and pinpoint bugs in the designed prompts of ICL. Our philosophy is to treat any LLM as software and validate its functionalities just like testing the software. Finally, we demonstrate applications of MMT4NL on the sentiment analysis and question-answering tasks. Our experiments could reveal various linguistic bugs in state-of-the-art LLMs.

Concept Bottleneck Models (CBMs) propose to enhance the trustworthiness of AI systems by constraining their decisions on a set of human-understandable concepts. However, CBMs typically assume that datasets contain accurate concept labels-an assumption often violated in practice, which we show can significantly degrade performance (by 25% in some cases). To address this, we introduce the Concept Preference Optimization (CPO) objective, a new loss function based on Direct Preference Optimization, which effectively mitigates the negative impact of concept mislabeling on CBM performance. We provide an analysis of key properties of the CPO objective, showing it directly optimizes for the concept's posterior distribution, and contrast it against Binary Cross Entropy (BCE), demonstrating that CPO is inherently less sensitive to concept noise. We empirically confirm our analysis by finding that CPO consistently outperforms BCE on three real-world datasets, both with and without added label noise. We make our code available on Github.

Recent developments in Direct Preference Optimization (DPO) allow large language models (LLMs) to function as implicit ranking models by maximizing the margin between preferred and non-preferred responses. In practice, user feedback on such lists typically involves identifying a few relevant items in context rather than providing detailed pairwise comparisons for every possible item pair. Moreover, many complex information retrieval tasks, such as conversational agents and summarization systems, critically depend on ranking the highest-quality outputs at the top, emphasizing the need to support natural and flexible forms of user feedback. To address the challenge of limited and sparse pairwise feedback in the in-context setting, we propose an In-context Ranking Preference Optimization (IRPO) framework that directly optimizes LLMs based on ranking lists constructed during inference. To further capture flexible forms of feedback, IRPO extends the DPO objective by incorporating both the relevance of items and their positions in the list. Modeling these aspects jointly is non-trivial, as ranking metrics are inherently discrete and non-differentiable, making direct optimization difficult. To overcome this, IRPO introduces a differentiable objective based on positional aggregation of pairwise item preferences, enabling effective gradient-based optimization of discrete ranking metrics. We further provide theoretical insights showing that IRPO (i) automatically emphasizes items with greater disagreement between the model and the reference ranking, and (ii) links its gradient to an importance sampling estimator, yielding an unbiased estimator with reduced variance. Empirical results show IRPO outperforms standard DPO approaches in ranking performance, highlighting its effectiveness in aligning LLMs with direct in-context ranking preferences.

As the general capabilities of large language models (LLMs) improve and agent applications become more widespread, the underlying deception risks urgently require systematic evaluation and effective oversight. Unlike existing evaluation which uses simulated games or presents limited choices, we introduce OpenDeception, a novel deception evaluation framework with an open-ended scenario dataset. OpenDeception jointly evaluates both the deception intention and capabilities of LLM-based agents by inspecting their internal reasoning process. Specifically, we construct five types of common use cases where LLMs intensively interact with the user, each consisting of ten diverse, concrete scenarios from the real world. To avoid ethical concerns and costs of high-risk deceptive interactions with human testers, we propose to simulate the multi-turn dialogue via agent simulation. Extensive evaluation of eleven mainstream LLMs on OpenDeception highlights the urgent need to address deception risks and security concerns in LLM-based agents: the deception intention ratio across the models exceeds 80%, while the deception success rate surpasses 50%. Furthermore, we observe that LLMs with stronger capabilities do exhibit a higher risk of deception, which calls for more alignment efforts on inhibiting deceptive behaviors.

Proteins typically exist in complexes, interacting with other proteins or biomolecules to perform their specific biological roles. Research on single-chain protein modeling has been extensively and deeply explored, with advancements seen in models like the series of ESM and AlphaFold2. Despite these developments, the study and modeling of multi-chain proteins remain largely uncharted, though they are vital for understanding biological functions. Recognizing the importance of these interactions, we introduce APM (All-Atom Protein Generative Model), a model specifically designed for modeling multi-chain proteins. By integrating atom-level information and leveraging data on multi-chain proteins, APM is capable of precisely modeling inter-chain interactions and designing protein complexes with binding capabilities from scratch. It also performs folding and inverse-folding tasks for multi-chain proteins. Moreover, APM demonstrates versatility in downstream applications: it achieves enhanced performance through supervised fine-tuning (SFT) while also supporting zero-shot sampling in certain tasks, achieving state-of-the-art results. We released our code at https://github.com/bytedance/apm.

Abstract--In this paper, we show that an eXtremely Large (XL) Multiple-Input Multiple-Output (MIMO) wireless system with appropriate analog combining components exhibits the properties of a universal function approximator, similar to a feedforward neural network. By treating the channel coefficients as the random nodes of a hidden layer and the receiver's analog combiner as a trainable output layer, we cast the XL MIMO system to the Extreme Learning Machine (ELM) framework, leading to a novel formulation for Over-The-Air (OT A) edge inference without requiring traditional digital processing nor pre-processing at the transmitter . Through theoretical analysis and numerical evaluation, we showcase that XL-MIMO-ELM enables near-instantaneous training and efficient classification, even in varying fading conditions, suggesting the paradigm shift of beyond massive MIMO systems as OT A artificial neural networks alongside their profound communications role. Compared to deep learning approaches and conventional ELMs, the proposed framework achieves on par performance with orders of magnitude lower complexity, making it highly attractive for inference tasks with ultra low power wireless devices. Future device-to-device and Goal-Oriented (GO) networks will facilitate the communication of sensory data from Transmitter (Tx) to Receiver (Rx) devices, not solely for the purpose of conventional information decoding and storage, but also for extracting features that guide autonomous devices towards desired actions [1], [2].

As inference-time scaling becomes critical for enhanced reasoning capabilities, it is increasingly becoming important to build models that are efficient to infer. We introduce Nemotron-H, a family of 8B and 56B/47B hybrid Mamba-Transformer models designed to reduce inference cost for a given accuracy level. To achieve this goal, we replace the majority of self-attention layers in the common Transformer model architecture with Mamba layers that perform constant computation and require constant memory per generated token. We show that Nemotron-H models offer either better or on-par accuracy compared to other similarly-sized state-of-the-art open-sourced Transformer models (e.g., Qwen-2.5-7B/72B and Llama-3.1-8B/70B), while being up to 3$\times$ faster at inference. To further increase inference speed and reduce the memory required at inference time, we created Nemotron-H-47B-Base from the 56B model using a new compression via pruning and distillation technique called MiniPuzzle. Nemotron-H-47B-Base achieves similar accuracy to the 56B model, but is 20% faster to infer. In addition, we introduce an FP8-based training recipe and show that it can achieve on par results with BF16-based training. This recipe is used to train the 56B model. We are releasing Nemotron-H base model checkpoints with support in Hugging Face and NeMo.