arXiv subjects

Weijie Fu

Publications and source records attributed to Weijie Fu.

5 recordsLinked to original sources

arXiv2026

SpatialCLI: Learning to Reason With Spatial Tools, Then Without Them

Vision-language models (VLMs) are increasingly used in embodied agents to interpret visual inputs, reason about spatial relationships, and make task-level decisions based on that reasoning. However, a fundamental capability mismatch remains: general VLMs can reason about the overall task but often miss the visual details that determine success, while specialist vision models can capture those details but cannot translate them into task-level decisions. In this work, we propose SpatialCLI, a framework that teaches VLMs to reason with spatial tools and progressively internalize the specialist perceptual capabilities they provide. SpatialCLI proceeds in three stages: (1) Call exposes specialist vision models as spatial tools to augment the VLM's perception; (2) Learn uses Cold-Start SFT and agentic RL to improve tool use; and (3) Internalize verbalizes successful tool-use trajectories to internalize specialist perceptual capabilities. We further introduce SpatialCLI-Bench, a 516-example benchmark for compositional perception across localization, segmentation, depth, and pose. On MindCube, SpatialCLI raises Qwen3-VL-8B-Instruct from 29.3% to 84.6% with tools, surpassing GPT-5.6 Sol with tools (72.1%), while retaining 73.8% without tools after internalization.

cs.AI↗

arXiv2025

Dilepton emission assists the search for the QCD critical point

In this work, we propose that dilepton emission rate (DER) could possibly carry the characteristic structures associated with the chiral criticality of the QCD system based on the extended Polyakov-quark-meson model. The model could successfully capture two main mechanisms for dilepton production, $\pi^+\pi^-$ and quark-antiquark annihilations on one hand, and self-consistently account for chiral transition and (de-)confinement on the other hand. All the moments of the DER peak exhibit extremal features similar to those of light-quark mass along the chemical freeze-out lines, thus the DER can reflect the change of chiral symmetry and criticality. However, the DER is relatively too low to produce a sufficient number of thermal dileptons in each heavy-ion collision, which prevents the implementation of event-by-event measurements of the moments. As an alternative, we propose to study the baseline-subtracted DER: For a given low center of mass of dileptons, the reduced baseline-subtracted DER exhibits a nonmonotonic behavior in the region where the light-quark mass changes most rapidly.

nucl-th↗

arXiv2022

Worst-case Design for RIS-aided Over-the-air Computation with Imperfect CSI

Over-the-air computation (AirComp) enables fast wireless data aggregation at the receiver through concurrent transmission by sensors in the application of Internet-of-Things (IoT). To further improve the performance of AirComp under unfavorable propagation channel conditions, we consider the problem of computation distortion minimization in a reconfigurable intelligent surface (RIS)-aided AirComp system. In particular, we take into account an additive bounded uncertainty of the channel state information (CSI) and the total power constraint, and jointly optimize the transceiver (Tx-Rx) and the RIS phase design from the perspective of worst-case robustness by minimizing the mean squared error (MSE) of the computation. To solve this intractable nonconvex problem, we develop an efficient alternating algorithm where both solutions to the robust sub-problem and to the joint design of Tx-Rx and RIS are obtained in closed forms. Simulation results demonstrate the effectiveness of the proposed method.

cs.IT↗

arXiv2021

ExAD: An Ensemble Approach for Explanation-based Adversarial Detection

Recent research has shown Deep Neural Networks (DNNs) to be vulnerable to adversarial examples that induce desired misclassifications in the models. Such risks impede the application of machine learning in security-sensitive domains. Several defense methods have been proposed against adversarial attacks to detect adversarial examples at test time or to make machine learning models more robust. However, while existing methods are quite effective under blackbox threat model, where the attacker is not aware of the defense, they are relatively ineffective under whitebox threat model, where the attacker has full knowledge of the defense. In this paper, we propose ExAD, a framework to detect adversarial examples using an ensemble of explanation techniques. Each explanation technique in ExAD produces an explanation map identifying the relevance of input variables for the model's classification. For every class in a dataset, the system includes a detector network, corresponding to each explanation technique, which is trained to distinguish between normal and abnormal explanation maps. At test time, if the explanation map of an input is detected as abnormal by any detector model of the classified class, then we consider the input to be an adversarial example. We evaluate our approach using six state-of-the-art adversarial attacks on three image datasets. Our extensive evaluation shows that our mechanism can effectively detect these attacks under blackbox threat model with limited false-positives. Furthermore, we find that our approach achieves promising results in limiting the success rate of whitebox attacks.

cs.LG↗

arXiv2020

A Survey on Large-scale Machine Learning

Machine learning can provide deep insights into data, allowing machines to make high-quality predictions and having been widely used in real-world applications, such as text mining, visual classification, and recommender systems. However, most sophisticated machine learning approaches suffer from huge time costs when operating on large-scale data. This issue calls for the need of {Large-scale Machine Learning} (LML), which aims to learn patterns from big data with comparable performance efficiently. In this paper, we offer a systematic survey on existing LML methods to provide a blueprint for the future developments of this area. We first divide these LML methods according to the ways of improving the scalability: 1) model simplification on computational complexities, 2) optimization approximation on computational efficiency, and 3) computation parallelism on computational capabilities. Then we categorize the methods in each perspective according to their targeted scenarios and introduce representative methods in line with intrinsic strategies. Lastly, we analyze their limitations and discuss potential directions as well as open issues that are promising to address in the future.

cs.LG↗