Prompt injection attacks pose serious security risks across a wide range of real-world applications.While receiving increasing attention, the community faces a critical gap: the lack of a unified platform for prompt injection evaluation.This makes it challenging to reliably compare defenses, understand their true robustness under diverse attacks, or assess how well they generalize across tasks and benchmarks.For instance, many defenses initially reported as effective were later found to exhibit limited robustness on diverse datasets and attacks.To bridge this gap, we introduce PIArena, a unified and extensible platform for prompt injection evaluation that enables users to easily integrate state-of-the-art attacks and defenses and evaluate them across a variety of existing and new benchmarks.We also design a dynamic strategy-based attack that adaptively optimizes injected prompts based on defense feedback.Through comprehensive evaluation using PIArena, we uncover critical limitations of state-of-the-art defenses: limited generalizability across tasks, vulnerability to adaptive attacks, and fundamental challenges when an injected task aligns with the target task.The code and datasets are available at https://github.com/sleeepeer/PIArena.

Enhancing the reasoning capabilities of Large Language Models (LLMs) is a key strategy for building Agents that "think then act."However, recent observations, like OpenAI's o3, suggest a paradox: stronger reasoning often coincides with increased hallucination, yet no prior work has systematically examined whether reasoning enhancement itself causes tool hallucination.To address this gap, we pose the central question: Does strengthening reasoning increase tool hallucination of LLM Agents?We address this gap by introducing SIMPLETOOL-HALLUBENCH, a diagnostic benchmark measuring tool hallucination.Through controlled experiments, we show that across RL, distillation, and toggleable reasoning modes, gains in task performance are consistently accompanied by higher tool hallucination rates.This effect is training method-agnostic and transcends simple overfitting, as training even on non-tool-related tasks (e.g., mathematics) still amplifies tool hallucination.Controlled ablations further reveal that the reasoning itself, rather than RL training in general, is most closely associated with the hallucination increase.We evaluate mitigation strategies including Prompt Engineering and Direct Preference Optimization (DPO), revealing a fundamental reliability-capability trade-off: reducing hallucination unavoidably degrades utility.Our findings demonstrate that under current reasoning enhancement methods, improved reasoning is systematically associated with increased tool hallucination, highlighting the need for training objectives that jointly optimize capability and reliability.Our

Pre-trained Language Models (PLMs), such as ChatGPT, have significantly advanced the field of natural language processing. This progress has inspired a series of innovative studies that explore the adaptation of PLMs to time series analysis, intending to create a unified foundation model that addresses various time series analytical tasks. However, these efforts predominantly focus on Regularly Sampled Time Series (RSTS), neglecting the unique challenges posed by Irregularly Sampled Time Series (ISTS), which are characterized by uneven sampling intervals and prevalent missing data. To bridge this gap, this work takes the first step in exploring the potential of PLMs for ISTS analysis. We begin by investigating the effect of various methods for representing ISTS, aiming to maximize the efficacy of PLMs in the analysis. Furthermore, we propose a unified PLM-based framework, named ISTS-PLM, to address diverse ISTS analytical tasks. It integrates novel time-aware and variable-aware PLMs tailored to tackle the intractable intra- and inter-time series modeling in ISTS. Finally, extensive experiments on a comprehensive benchmark demonstrate that the ISTS-PLM, utilizing a structured and effective series-based representation for ISTS, consistently achieves state-of-the-art performance across various analytical tasks, such as classification, interpolation, extrapolation, few-shot and zero-shot learning scenarios, spanning scientific domains like healthcare, biomechanics, and climate science.

Proactive caching scheme proactively caches data before the request comes, which is essential for optimizing of the resource utilization and ensuring the stability of cached contents. However, current proactive caching methods suffer from high computational complexity and often fall into suboptimal solutions. This study proposes a hierarchical reinforcement learning (HRL)-based proactive caching scheme to determine whether to update the caching queue and how to update the caching queue. Specifically, we build a bi-level optimization model to maximize the content cache hit ratio while minimizing the content cache cost. We formulate the upper-level and lower-level optimization problems as Markov decision process (MDP) models and employ double deep Q-learning methods to solve the optimal caching strategy. In particular, the content history request features and the user preference generated by the collaborative filtering-based recommendation system are used as the state inputs of caching decisions. In addition, we design a two-stage curriculum learning paradigm for training the agents from easy tasks to hard tasks. At the first stage, a greedy decision tree with a posterior knowledge is designed to provide the upper-level decisions for training the lower-level agent in the easy task. Then, we train all agents together in the hard task while freezing the trained lower-level agent. We conduct both the simulation and on-road experiments to validate the effectiveness of our method, and results indicate that our method outperforms the baseline methods by 2.2%–37.2% in terms of cache hit ratio and 75.4%–77.6% in terms of cache cost.

Forecasting of Irregular Multivariate Time Series (IMTS) is critical for numerous areas, such as healthcare, biomechanics, climate science, and astronomy. Despite existing research addressing irregularities in time series through ordinary differential equations, the challenge of modeling correlations between asynchronous IMTS remains underexplored. To bridge this gap, this study proposes Transformable Patching Graph Neural Networks (t-PatchGNN), which transforms each univariate irregular time series into a series of transformable patches encompassing a varying number of observations with uniform temporal resolution. It seamlessly facilitates local semantics capture and inter-time series correlation modeling while avoiding sequence length explosion in aligned IMTS. Building on the aligned patching outcomes, we then present time-adaptive graph neural networks to model dynamic intertime series correlation based on a series of learned time-varying adaptive graphs. We demonstrate the remarkable superiority of t-PatchGNN on a comprehensive IMTS forecasting benchmark we build, which contains four real-world scientific datasets covering healthcare, biomechanics and climate science, and seventeen competitive baselines adapted from relevant research fields.

Increasingly more AGVs (Automated Guided Vehicles) are being utilized in ACTs (Automated Container Terminals) in recent years. Intersection collaboration of AGVs is crucial for improving the efficiency of containers transportation. Most of the existing methods model intersections as topological nodes which allows only one AGV to pass through at a time, or as grids which may lead to heavy computation workload. Moreover, these methods usually do not consider the mutual influences of the collaboration decisions between adjacent intersections and can hardly realize the full potential of AGVs. To address this issue, this paper proposes an MCTS (Monte Carlo Tree Search)- based intersection collaboration method for AGVs in ACTs. The proposed method models each intersection with a polygon and periodically assigns priorities to AGVs waiting for passing through intersections. The priorities of AGVs are determined by simultaneously applying MCTS to multiple adjacent intersections considering conflicts avoidance, queue alleviation and congestion prevention. The results of experiments show that the proposed method achieves better average intersection crossing time and total tasks completion time.