Noticias de IA - ítems

arXiv:2508.07406v2 Announce Type: replace-cross Abstract: Agricultural robots are serving as powerful members across a wide range of tasks, nevertheless, they are still heavily relying on manual operations or railway systems for movement. In contrast, Vision-and-Language Navigation (VLN) enables navigating robots from starting points to target positions following natural language instructions, having shown strong performances across various domains. However, none of the existing benchmarks or methods are specifically designed for agricultural robots. To bridge this gap, we build our A2A benchmark, which consists of 1,560 episodes across 6 classes of agricultural scenes using the step-by-step format for instructions and the continuous environment for actions. We evaluate several existing state-of-the-art VLN methods on A2A, then we observe that they perform well on short instructions while perform poorly on long instructions. To alleviate this phenomenon, we design our STL module, which decomposes a complex instruction into a list of simple subtasks, thereby assisting the decision-maker to focus on the current short subtask rather than the entire long instruction at any time. We integrate STL into the NavGPT backbone to propose our AgriVLN method, achieving a sufficient improvement on SR from 0.33 to 0.47 with a slight sacrifice on NE from 2.76 m to 2.91 m. We further implement the ablation studies and the case study, discussing the effectiveness and limitations of AgriVLN on different subtask quantities and scene classes. Repository: https://github.com/AlexTraveling/AgriVLN.

arXiv:2506.16590v2 Announce Type: replace-cross Abstract: Reinforcement learning remains sample inefficient, particularly when agents must adapt over time to related but distinct tasks, as in continual learning settings. Efficiency can be improved by transferring knowledge from a pretrained teacher policy to guide exploration in new but related tasks. However, if the new task sufficiently differs from the teacher's training task, the transferred guidance may be sub-optimal and bias exploration toward low-reward behaviors. We propose an energy-based transfer learning method that uses out-of-distribution detection to selectively issue guidance, enabling the teacher to intervene only in states within its training distribution. We theoretically link energy scores to state-visitation patterns under on-policy reinforcement learning. Empirically, our method improves sample efficiency and performance across both single-task and multi-task settings.

arXiv:2506.12045v3 Announce Type: replace-cross Abstract: Much of the physical world is inaccessible to direct measurement and can be reached only through the sparse traces it leaves in other observable variables. Here we introduce the Temporal Radiation Operator Network (TRON), a spatiotemporal neural operator that learns proxy-to-field mappings, reconstructing otherwise unobserved physical fields from sparse, indirect sensor histories. For cosmic-radiation mapping, TRON reconstructs sea-level effective dose at 65,341 locations from histories of 12 neutron monitors, achieving 0.096\% mean relative $L_2$ error against EXPACS-generated reference fields in a chronologically held-out year; batched GPU inference evaluates a global field in approximately $\sim$1~ms, compared with $\sim$57.6~s for EXPACS on one CPU thread. On measured weather networks, TRON reconstructs temperature at held-out stations from thermometer or barometer histories, outperforming interpolation, calendar-based and hypsometric baselines. These results establish proxy-to-field reconstruction as a general route to recovering spatially distributed physical information beyond the direct reach of sparse sensor networks.

arXiv:2506.07121v3 Announce Type: replace-cross Abstract: Ensuring the safety and robustness of large language models (LLMs) is a fundamental challenge and a critical prerequisite for the responsible deployment of artificial intelligence. Red-teaming, a systematic framework to identify adversarial prompts that elicit harmful responses from target LLMs, has emerged as a crucial safety evaluation paradigm. Within this framework, the diversity of adversarial prompts is critical for a comprehensive safety assessment. However, previous red-teaming approaches often pursue diversity through simplistic metrics such as word frequency or sentence embedding similarity, which may not capture meaningful variation in attack strategies. In addition, the common practice of training a single attacker model restricts coverage across all potential attack styles and risk categories. This paper introduces Quality-Diversity Red-Teaming (QDRT), a new framework designed to address these limitations. QDRT achieves goal-driven diversity through behavior-conditioned training and implements a behavioral replay buffer in an open-ended manner. In addition, it trains multiple specialized attackers capable of generating high-quality attacks across diverse styles and risk categories. Our empirical evaluation demonstrates that QDRT generates attacks that are both more diverse and more effective against a wide range of target LLMs, including open-source models \texttt{GPT-2}, \texttt{Llama-3}, \texttt{Gemma-2}, \texttt{Qwen2.5}, and commercial models \texttt{GPT-4.1} and \texttt{GPT-5-Chat}. This work advances the field of LLM safety by providing a systematic and effective approach to automated red-teaming, ultimately supporting the responsible deployment of LLMs. Our code is available at https://github.com/lamda-bbo/QDRT.

arXiv:2505.24255v3 Announce Type: replace-cross Abstract: Large Language Models (LLMs) have shown potential in simulating human behaviors and performing theory-of-mind (ToM) reasoning, crucial for complex social interactions. We investigate ToM reasoning's role in aligning agentic behaviors with human norms in negotiation tasks, using the ultimatum game as our referenced task. We initialized LLM agents with different prosocial beliefs (Greedy, Fair, Selfless) and reasoning methods (chain of thought and ToM reasoning of varying levels), examining their decision-making processes and outcomes across multiple LLMs, including reasoning models like o3-mini and DeepSeek-R1 Distilled Qwen 32B. We perform 2,700 simulations to show that ToM reasoning enhances behavioral alignment with human norms, decision-making consistency, and negotiation outcomes. Consistent with prior findings, reasoning LLMs exhibit limited capability compared to ToM-enhanced LLMs, with different game roles benefiting from different ToM orders. Agents almost always chose strategies consistent with their assigned beliefs, and doing so aligned better with human norms in nearly every role, yet Greedy and Selfless agents still fell short of their norms in outcomes. Human verification further revealed that Llama 3.3 70B produces reasoning most consistent with its actions and beliefs. Our findings advance understanding of ToM's role in human-AI interaction and cooperative decision-making. The code used for our experiments can be found at https://github.com/Stealth-py/UltimatumToM.

arXiv:2505.06335v3 Announce Type: replace-cross Abstract: Federated Learning (FL) has the potential for simultaneous global learning amongst a large number of parallel agents, enabling emerging AI systems to be trained across demographically diverse data. Central to this being efficient is the ability for FL to perform sparse gradient updates and remote direct memory access at the central server. Most of the research in FL security focuses on protecting data privacy at the edge client or in the communication channels between the client and server. Client-facing attacks on the server are less well investigated as the assumption is that a large collective of clients offer resilience. Here, for the first time, we show that by attacking clients that lead to repetitive clustered memory updates in the server, we can remote initiate a rowhammer attack to corrupt the server memory with induced bit-flips. A reinforcement learning (RL) attacker without backdoor access to the server can learn how to maximize server repetitive clustered memory updates by manipulating the clients' sensor observation. We demonstrate the feasibility of our attack using FL systems with sparse updates, tasked for automatic speech recognition (ASR) as well as CIFAR10 classification, our adversarial attacking agent can achieve around 70% repeated update rate (RUR) in the targeted server model, effectively inducing bit flips on server DRAM. The security implications are that such adversarial observations can cause disruptions to learning or may inadvertently cause elevated privilege, which could pave the way for further research on practical mitigation strategies in FL and hardware design.

arXiv:2505.04841v3 Announce Type: replace-cross Abstract: Data imputation is a critical preprocessing step, particularly in clinical datasets where missingness can bias downstream inference. We propose a quantum-inspired imputation framework and evaluate it on the UCI Pima Indians Diabetes dataset, where several physiological variables contain structural zeros that are treated as missing. The approach combines Principal Component Analysis (PCA) with quantum rotation-based reconstruction, where rotation angles are selected via gradient-free optimizers (Simulated Annealing, COBYLA, and Differential Evolution). For robustness, we retain the top three optimizer solutions per method and reconstruct missing values under distribution-aware bounds (e.g., $\pm 2$ standard deviations) and avoid unrealistic collapse toward central tendency. We also benchmark against (i) classical baselines (Mean, KNN, MICE, Decision Tree), (ii) modern matrix/ML imputers (SoftImpute, MissForest, VAE), and (iii) rotation/PCA controls (random orthogonal rotations and bootstrap PCA). Across the evaluated features, quantum-inspired variants achieve substantially improved distributional fidelity under Wasserstein W1 relative to modern baselines (e.g., the best quantum setting reduces average W1 by $\sim$66% vs. SoftImpute, $\sim$84% vs. MissForest, and $\sim$90% vs. VAE; and by $\sim$60% vs. Mean imputation), while remaining competitive with strong classical/control configurations. Dispersion is better preserved under the quantum-inspired methods, with variance/SD ratios closer to the observed data than (SoftImpute, MissForest, and VAE) and typically better than (Mean, KNN, MICE, and Decision Tree) on average. Overall, the results support quantum-inspired, optimization-guided rotations as a statistically grounded and scalable strategy for realistic imputation in healthcare and general AI pipelines, with a design that can be mapped to future quantum hardware as it matures.

arXiv:2504.03690v2 Announce Type: replace-cross Abstract: We consider multiple transmitters aiming to communicate their source signals (e.g., images) over a multiple access channel (MAC). Conventional communication systems minimize interference by orthogonally allocating resources (time and/or bandwidth) among users, which limits their capacity. We introduce a machine learning (ML)-aided wireless image transmission method that merges compression and channel coding using a multi-view autoencoder, which allows the transmitters to use all the available channel resources simultaneously, resulting in a non-orthogonal multiple access (NOMA) scheme. The receiver must recover all the images from the received superposed signal, while also associating each image with its transmitter. Traditional ML models deal with individual samples, whereas our model allows signals from different users to interfere in order to leverage gains from NOMA under limited bandwidth and power constraints. We introduce a progressive fine-tuning algorithm that doubles the number of users at each iteration, maintaining initial performance with orthogonalized user-specific projections, which is then improved through fine-tuning steps. Remarkably, our method supports up to 64 users with potential for further scaling, with only a 2.4% increase in the number of trainable parameters compared to a single-user model, significantly enhancing recovered image quality and outperforming existing NOMA-based methods over a wide range of datasets, metrics, and channel conditions. Our approach paves the way for more efficient and robust multi-user communication systems, leveraging innovative ML components and strategies.

arXiv:2503.19394v2 Announce Type: replace-cross Abstract: Current machine learning approaches to medical diagnosis often rely on correlational patterns between symptoms and diseases, risking misdiagnoses when symptoms are ambiguous or common across multiple conditions. In this work, we move beyond correlation to investigate the causal influence of key symptoms-specifically "chest pain" on diagnostic predictions. Leveraging the CausaLM framework, we generate counterfactual text representations in which target concepts are effectively "forgotten" enabling a principled estimation of the causal effect of that concept on a model's predicted disease distribution. By employing Textual Representation-based Average Treatment Effect (TReATE), we quantify how the presence or absence of a symptom shapes the model's diagnostic outcomes, and contrast these findings against correlation-based baselines such as CONEXP. Our results offer deeper insight into the decision-making behavior of clinical NLP models and have the potential to inform more trustworthy, interpretable, and causally-grounded decision support tools in medical practice.

arXiv:2503.00128v2 Announce Type: replace-cross Abstract: Language models offer a way to scale costly, labour-intensive expert annotation, but their ability to reproduce expert judgements remains uncertain. This is particularly challenging in law, where annotators must distinguish the reasoning supporting judicial decisions from other legally relevant discussion. We present AnnoCaseLaw as a densely annotated benchmark comprising $471$ U.S. civil-negligence court opinions with more than $12,000$ expert highlights across five annotation categories. We evaluate proprietary frontier, open-weight, legal-specialised, and decision models on their ability to reproduce these annotations. Models perform better on Facts and Outcome than on Application of Law to Facts, Procedural History, and Relevant Precedents. On Application of Law to Facts, proprietary frontier models achieve the strongest balance of reasoning coverage and precision, while the open-weight model Gemma 4 achieves comparable Macro-F1 to proprietary frontier models at a fraction of the estimated inference cost. Our findings highlight the importance of evaluating annotation coverage and cost alongside aggregate scores when assessing the feasibility of automated legal annotation.