Multimodal Retrieval-Augmented Generation (mRAG) has emerged as a promising solution to address the temporal limitations of Multimodal Large Language Models (MLLMs) in real-world scenarios like news analysis and trending topics. However, existing approaches often suffer from rigid retrieval strategies and under-utilization of visual information. To bridge this gap, we propose E-Agent, an agent framework featuring two key innovations: a mRAG planner trained to dynamically orchestrate multimodal tools based on contextual reasoning, and a task executor employing tool-aware execution sequencing to implement optimized mRAG workflows. E-Agent adopts a one-time mRAG planning strategy that enables efficient information retrieval while minimizing redundant tool invocations. To rigorously assess the planning capabilities of mRAG systems, we introduce the Real-World mRAG Planning (RemPlan) benchmark. This novel benchmark contains both retrieval-dependent and retrieval-independent question types, systematically annotated with essential retrieval tools required for each instance. The benchmark's explicit mRAG planning annotations and diverse question design enhance its practical relevance by simulating real-world scenarios requiring dynamic mRAG decisions. Experiments across RemPlan and three established benchmarks demonstrate E-Agent's superiority: 13% accuracy gain over state-of-the-art mRAG methods while reducing redundant searches by 37%.

Effective information seeking in the vast and ever-growing digital landscape requires balancing expansive search with strategic reasoning. Current large language model (LLM)-based agents struggle to achieve this balance due to limitations in search breadth and reasoning depth, where slow, serial querying restricts coverage of relevant sources and noisy raw inputs disrupt the continuity of multi-step reasoning. To address these challenges, we propose BrowseMaster, a scalable framework built around a programmatically augmented planner-executor agent pair. The planner formulates and adapts search strategies based on task constraints, while the executor conducts efficient, targeted retrieval to supply the planner with concise, relevant evidence. This division of labor preserves coherent, long-horizon reasoning while sustaining broad and systematic exploration, overcoming the trade-off that limits existing agents. Extensive experiments on challenging English and Chinese benchmarks show that BrowseMaster consistently outperforms open-source and proprietary baselines, achieving scores of 30.0 on BrowseComp-en and 46.5 on BrowseComp-zh, which demonstrates its strong capability in complex, reasoning-heavy information-seeking tasks at scale.

In recent years, deep neural networks (DNNs) have shown competitive results in many fields. Despite this success, they often suffer from poor calibration, especially in safety-critical scenarios such as autonomous driving and healthcare, where unreliable confidence estimates can lead to serious consequences. Recent studies have focused on improving calibration by modifying the classifier, yet such efforts remain limited. Moreover, most existing approaches overlook calibration errors caused by underconfidence, which can be equally detrimental. To address these challenges, we propose MaC-Cal, a novel mask-based classifier calibration method that leverages stochastic sparsity to enhance the alignment between confidence and accuracy. MaC-Cal adopts a two-stage training scheme with adaptive sparsity, dynamically adjusting mask retention rates based on the deviation between confidence and accuracy. Extensive experiments show that MaC-Cal achieves superior calibration performance and robustness under data corruption, offering a practical and effective solution for reliable confidence estimation in DNNs.

Large vision language models exhibit notable limitations on Geometry Problem Solving (GPS) because of their unreliable diagram interpretation and pure natural-language reasoning. A recent line of work mitigates this by using symbolic solvers: the model directly generates a formal program that a geometry solver can execute. However, this direct program generation lacks intermediate reasoning, making the decision process opaque and prone to errors. In this work, we explore a new approach that integrates Chain-of-Thought (CoT) with formal language. The model interleaves natural language reasoning with incremental emission of solver-executable code, producing a hybrid reasoning trace in which critical derivations are expressed in formal language. To teach this behavior at scale, we combine (1) supervised fine-tuning on an 11K newly developed synthetic dataset with interleaved natural language reasoning and automatic formalization, and (2) solver-in-the-loop reinforcement learning that jointly optimizes both the CoT narrative and the resulting program through outcome-based rewards. Built on Qwen2.5-VL-7B, our new model, named GF-Reasoner, achieves up to 15% accuracy improvements on standard GPS benchmarks, surpassing both 7B-scale peers and the much larger model Qwen2.5-VL-72B. By exploiting high-order geometric knowledge and offloading symbolic computation to the solver, the generated reasoning traces are noticeably shorter and cleaner. Furthermore, we present a comprehensive analysis of method design choices (e.g., reasoning paradigms, data synthesis, training epochs, etc.), providing actionable insights for future research.

We introduce Chi-Geometry - a library that generates graph data for testing and benchmarking GNNs' ability to predict chirality. Chi-Geometry generates synthetic graph samples with (i) user-specified geometric and topological traits to isolate certain types of samples and (ii) randomized node positions and species to minimize extraneous correlations. Each generated graph contains exactly one chiral center labeled either R or S, while all other nodes are labeled N/A (non-chiral). The generated samples are then combined into a cohesive dataset that can be used to assess a GNN's ability to predict chirality as a node classification task. Chi-Geometry allows more interpretable and less confounding benchmarking of GNNs for prediction of chirality in the graph samples which can guide the design of new GNN architectures with improved predictive performance. We illustrate Chi-Geometry's efficacy by using it to generate synthetic datasets for benchmarking various state-of-the-art (SOTA) GNN architectures. The conclusions of these benchmarking results guided our design of two new GNN architectures. The first GNN architecture established all-to-all connections in the graph to accurately predict chirality across all challenging configurations where previously tested SOTA models failed, but at a computational cost (both for training and inference) that grows quadratically with the number of graph nodes. The second GNN architecture avoids all-to-all connections by introducing a virtual node in the original graph structure of the data, which restores the linear scaling of training and inference computational cost with respect to the number of nodes in the graph, while still ensuring competitive accuracy in detecting chirality with respect to SOTA GNN architectures.

Intrusion Detection Systems (IDS) are a vital part of a network-connected device. In this paper, we develop a deep learning based intrusion detection system that is deployed in a distributed setup across devices connected to a network. Our aim is to better equip deep learning models against unknown attacks using knowledge from known attacks. To this end, we develop algorithms to maximize the number of transferability relationships. We propose a Convolutional Neural Network (CNN) model, along with two algorithms that maximize the number of relationships observed. One is a two step data pre-processing stage, and the other is a Block-Based Smart Aggregation (BBSA) algorithm. The proposed system succeeds in achieving superior transferability performance while maintaining impressive local detection rates. We also show that our method is generalizable, exhibiting transferability potential across datasets and even with different backbones. The code for this work can be found at https://github.com/ghosh64/tabfidsv2.

The nodes represent the devices chosen to participate in federated training and the local models are the models used on the devices to train with their local, private data.Figure 1: Architecture of the proposed intrusion detection system. A. Nodes We consider a network of Internet of Things devices connected to each other. Each node monitors the data packets that are flowing through the network and detect malicious data packets injected by attackers using a deep learning based intrusion detection model deployed on the local devices. The intrusion detection systems are always collecting local data, training their models on local data and participating in communication rounds with the central server. B. Attention Based Deep Learning Model The deep learning model is an attention based model, presented in Figure 1. The input is a 1D vector that contains the features of the network traffic data. Attention models are typically aimed at drawing the focus of the model on the contextually important parts of an input. The multihead attention function consists of several attention heads that focus the attention of the model on different aspects or different features of the data. The computed attention matrix tells the model which part of the input is important in relation to itself.

The rapid emergence of large language models (LLMs) has raised urgent questions across the modern workforce about this new technology's strengths, weaknesses, and capabilities. For privacy professionals, the question is whether these AI systems can provide reliable support on regulatory compliance, privacy program management, and AI governance. In this study, we evaluate ten leading open and closed LLMs, including models from OpenAI, Anthropic, Google DeepMind, Meta, and DeepSeek, by benchmarking their performance on industry-standard certification exams: CIPP/US, CIPM, CIPT, and AIGP from the International Association of Privacy Professionals (IAPP). Each model was tested using official sample exams in a closed-book setting and compared to IAPP's passing thresholds. Our findings show that several frontier models such as Gemini 2.5 Pro and OpenAI's GPT-5 consistently achieve scores exceeding the standards for professional human certification - demonstrating substantial expertise in privacy law, technical controls, and AI governance. The results highlight both the strengths and domain-specific gaps of current LLMs and offer practical insights for privacy officers, compliance leads, and technologists assessing the readiness of AI tools for high-stakes data governance roles. This paper provides an overview for professionals navigating the intersection of AI advancement and regulatory risk and establishes a machine benchmark based on human-centric evaluations.

As large language models (LLMs) become more integrated into societal systems, the risk of them perpetuating and amplifying harmful biases becomes a critical safety concern. Traditional methods for mitigating bias often rely on data filtering or post-hoc output moderation, which treat the model as an opaque black box. In this work, we introduce a complete, end-to-end system that uses techniques from mechanistic interpretability to both identify and actively mitigate bias directly within a model's internal workings. Our method involves two primary stages. First, we train linear "probes" on the internal activations of a model to detect the latent representations of various biases (e.g., gender, race, age). Our experiments on \texttt{gpt2-large} demonstrate that these probes can identify biased content with near-perfect accuracy, revealing that bias representations become most salient in the model's later layers. Second, we leverage these findings to compute "steering vectors" by contrasting the model's activation patterns for biased and neutral statements. By adding these vectors during inference, we can actively steer the model's generative process away from producing harmful, stereotypical, or biased content in real-time. We demonstrate the efficacy of this activation steering technique, showing that it successfully alters biased completions toward more neutral alternatives. We present our work as a robust and reproducible system that offers a more direct and interpretable approach to building safer and more accountable LLMs.

Designing mechanical mechanisms to trace specific paths is a classic yet notoriously difficult engineering problem, characterized by a vast and complex search space of discrete topologies and continuous parameters. We introduce MechaFormer, a Transformer-based model that tackles this challenge by treating mechanism design as a conditional sequence generation task. Our model learns to translate a target curve into a domain-specific language (DSL) string, simultaneously determining the mechanism's topology and geometric parameters in a single, unified process. MechaFormer significantly outperforms existing baselines, achieving state-of-the-art path-matching accuracy and generating a wide diversity of novel and valid designs. We demonstrate a suite of sampling strategies that can dramatically improve solution quality and offer designers valuable flexibility. Furthermore, we show that the high-quality outputs from MechaFormer serve as excellent starting points for traditional optimizers, creating a hybrid approach that finds superior solutions with remarkable efficiency.