Zhang Wei

Renrui Zhang, Zhang Wei, Rongyao Fang et al.

Contrastive Vision-Language Pre-training, known as CLIP, has provided a new paradigm for learning visual representations using large-scale image-text pairs. It shows impressive performance on downstream tasks by zero-shot knowledge transfer. To further enhance CLIP's adaption capability, existing methods proposed to fine-tune additional learnable modules, which significantly improves the few-shot performance but introduces extra training time and computational resources. In this paper, we propose a training-free adaption method for CLIP to conduct few-shot classification, termed as Tip-Adapter, which not only inherits the training-free advantage of zero-shot CLIP but also performs comparably to those training-required approaches. Tip-Adapter constructs the adapter via a key-value cache model from the few-shot training set, and updates the prior knowledge encoded in CLIP by feature retrieval. On top of that, the performance of Tip-Adapter can be further boosted to be state-of-the-art on ImageNet by fine-tuning the cache model for 10$\times$ fewer epochs than existing methods, which is both effective and efficient. We conduct extensive experiments of few-shot classification on 11 datasets to demonstrate the superiority of our proposed methods. Code is released at https://github.com/gaopengcuhk/Tip-Adapter.

Zhang Wei, Hanxuan Chen, Peilu Hu et al.

The increasing deployment of large language models (LLMs) in safety-critical applications raises fundamental challenges in systematically evaluating robustness against adversarial behaviors. Existing red-teaming practices are largely manual and expert-driven, which limits scalability, reproducibility, and coverage in high-dimensional prompt spaces. We formulate automated LLM red-teaming as a structured adversarial search problem and propose a learning-driven framework for scalable vulnerability discovery. The approach combines meta-prompt-guided adversarial prompt generation with a hierarchical execution and detection pipeline, enabling standardized evaluation across six representative threat categories, including reward hacking, deceptive alignment, data exfiltration, sandbagging, inappropriate tool use, and chain-of-thought manipulation. Extensive experiments on GPT-OSS-20B identify 47 vulnerabilities, including 21 high-severity failures and 12 previously undocumented attack patterns. Compared with manual red-teaming under matched query budgets, our method achieves a 3.9$\times$ higher discovery rate with 89\% detection accuracy, demonstrating superior coverage, efficiency, and reproducibility for large-scale robustness evaluation.

Zheng Xiangyu, He Songcheng, Li Wanyun et al.

Unsupervised Video Object Segmentation (UVOS) aims to predict pixel-level masks for the most salient objects in videos without any prior annotations. While memory mechanisms have been proven critical in various video segmentation paradigms, their application in UVOS yield only marginal performance gains despite sophisticated design. Our analysis reveals a simple but fundamental flaw in existing methods: over-reliance on memorizing high-level semantic features. UVOS inherently suffers from the deficiency of lacking fine-grained information due to the absence of pixel-level prior knowledge. Consequently, memory design relying solely on high-level features, which predominantly capture abstract semantic cues, is insufficient to generate precise predictions. To resolve this fundamental issue, we propose a novel hierarchical memory architecture to incorporate both shallow- and high-level features for memory, which leverages the complementary benefits of pixel and semantic information. Furthermore, to balance the simultaneous utilization of the pixel and semantic memory features, we propose a heterogeneous interaction mechanism to perform pixel-semantic mutual interactions, which explicitly considers their inherent feature discrepancies. Through the design of Pixel-guided Local Alignment Module (PLAM) and Semantic-guided Global Integration Module (SGIM), we achieve delicate integration of the fine-grained details in shallow-level memory and the semantic representations in high-level memory. Our Hierarchical Memory with Heterogeneous Interaction Network (HMHI-Net) consistently achieves state-of-the-art performance across all UVOS and video saliency detection benchmarks. Moreover, HMHI-Net consistently exhibits high performance across different backbones, further demonstrating its superiority and robustness. Project page: https://github.com/ZhengxyFlow/HMHI-Net .

Li Kai, Li Ning, Zhang Wei et al.

Based on the traditional VAE, a novel neural network model is presented, with the latest molecular representation, SELFIES, to improve the effect of generating new molecules. In this model, multi-layer convolutional network and Fisher information are added to the original encoding layer to learn the data characteristics and guide the encoding process, which makes the features of the data hiding layer more aggregated, and integrates the Long Short Term Memory neural network (LSTM) into the decoding layer for better data generation, which effectively solves the degradation phenomenon generated by the encoding layer and decoding layer of the original VAE model. Through experiments on zinc molecular data sets, it is found that the similarity in the new VAE is 8.47% higher than that of the original ones. SELFIES are better at generating a variety of molecules than the traditional molecular representation, SELFIES. Experiments have shown that using SELFIES and the new VAE model presented in this paper can improve the effectiveness of generating new molecules.

Yunyi Zhao, Zhang Wei, Qingyu Yan et al.

Battery health monitoring and prediction are critically important in the era of electric mobility with a huge impact on safety, sustainability, and economic aspects. Existing research often focuses on prediction accuracy but tends to neglect practical factors that may hinder the technology's deployment in real-world applications. In this paper, we address these practical considerations and develop models based on the Bayesian neural network for predicting battery end-of-life. Our models use sensor data related to battery health and apply distributions, rather than single-point, for each parameter of the models. This allows the models to capture the inherent randomness and uncertainty of battery health, which leads to not only accurate predictions but also quantifiable uncertainty. We conducted an experimental study and demonstrated the effectiveness of our proposed models, with a prediction error rate averaging 13.9%, and as low as 2.9% for certain tested batteries. Additionally, all predictions include quantifiable certainty, which improved by 66% from the initial to the mid-life stage of the battery. This research has practical values for battery technologies and contributes to accelerating the technology adoption in the industry.

Alexis Decurninge, Luis García Ordóñez, Paul Ferrand et al.

We report on experimental results on the use of a learning-based approach to infer the location of a mobile user of a cellular network within a cell, for a 5G-type Massive multiple input, multiple output (MIMO) system. We describe how the sample spatial covariance matrix computed from the CSI can be used as the input to a learning algorithm which attempts to relate it to user location. We discuss several learning approaches, and analyze in depth the application of extreme learning machines, for which theoretical approximate performance benchmarks are available, to the localization problem. We validate the proposed approach using experimental data collected on a Huawei 5G testbed, provide some performance and robustness benchmarks, and discuss practical issues related to the deployment of such a technique in 5G networks.