Weakly supervised video anomaly detection (WVAD) aims to locate events or behaviors that deviate from normal patterns in untrimmed videos using video-level labels. Recent studies typically utilize supplementary modalities to assist anomaly detection. However, these methods suffer from two main issues: (1) The limitations of long-duration anomaly event temporal modeling. The model struggles to consistently maintain key information, resulting in the forgetting phenomenon, which affects the tracking of the event’s overall dynamic evolution and complicates anomaly event analysis and understanding. (2) The multi-modal fusion strategy is insufficient, particularly when there is temporal inconsistency between visual and audio information, causing the model to overlook key information, directly affecting the accurate detection and recognition of anomalous events. To address these issues, we propose a visual-guided long-term temporal context learning network (LTCLNet). The network consists of three key components: a cross-modal interaction module, a multi-modal fusion module, and a visual-guided parameter optimization strategy. First, to address the forgetting issue in long-duration anomaly detection, we designed a cross-modal interaction module. The key part of this module is the establishment of a cross-matrix mechanism. This mechanism achieves bidirectional temporal guidance across modalities. It allows the temporal modeling of each modality to dynamically integrate information from the other modality. This enables the model to continuously track the dynamic evolution of the event. The tracking is facilitated through shared temporal information between the visual and audio modalities. Secondly, to fully exploit the complementary characteristics between different modalities, we introduced a novel temporal reversal integration method in the multi-modal fusion module. This method reverses the feature sequences of each modality to enhance the model’s perception of temporal dynamic changes. By fusing the modality features before and after reversal, the shared temporal structure between modalities is strengthened, improving the model’s ability to capture anomalous information. Additionally, our proposed visual-guided parameter optimization strategy trains a parallel visual modality network as a semantic anchor, ensuring that the model stays aligned with a semantically stable and structurally clear visual flow during the learning process, thus ensuring stability and semantic coherence in the training. Extensive experiments on datasets such as XD-Violence demonstrate that our method significantly outperforms existing approaches, particularly achieving notable improvements in the accuracy and stability of long-term anomaly detection. Our code is publicly available at https://github.com/ibliever/LTCLNet .

With the rapid advancements in information technologies, global data volumes have increased exponentially. Deoxyribonucleic acid (DNA)-based data storage has gained significant attention due to its superior storage capacity and long-term preservation potential compared to traditional media. However, during this process, DNA sequences may be affected by errors such as deletions, insertions, and substitutions, resulting in altered copies of the original sequence. Furthermore, the sequences in the DNA pool lack inherent order, preventing direct access to specific fragments. In this paper, we propose GTree, a trie-based DNA clustering framework tailored for DNA data storage. GTree achieves both high accuracy and low runtime complexity. It introduces a lightweight pre-screening module that combines Q-gram feature extraction, MinHash signatures, and Locality Sensitive Hashing (LSH) indexing to efficiently reduce the comparison space. A majority-label voting mechanism is also integrated to enhance clustering robustness in the presence of sequencing errors. Finally, we validate the performance of GTree through clustering experiments on both real biological and simulated datasets, demonstrating its superior performance.

Post-training quantization (PTQ) for transformer-based large foundation models (LFMs) significantly accelerates model inference and relieves memory constraints, without incurring model training. However, existing methods face three main issues: 1) The scaling factors, which are commonly used in scale reparameterization based weight-activation quantization for mitigating the quantization errors, are mostly hand-crafted defined which may lead to suboptimal results; 2) The formulation of current quantization error defined by L2-norm ignores the directional shifts after quantization; 3) Most methods are devised tailored for single scenario, i.e., only evaluated on LLMs or only designed for weight-only quantization, which lacks of a comprehensive evaluation on diverse benchmarks and a broad application scope. To address these challenges, this paper introduces a unified Learnable and Robust post-training Quantization framework for transformer based LFMs and various quantization scenarios, called LRQuant. First, we consider an efficient block-wise learnable paradigm to find optimal scaling factors which are initialized by logarithmic activation equivalent and get suitable clipping range of quantization steps. In addition, we empirically find that only relying on MSE loss could hardly lead to optimal quantization results, so we reformulate the quantization error and then propose a novel loss function based on the negative logarithm of cosine similarity (NLC loss) between outputs of full-precision and quantized block. To fully investigate the potentiality of our learnable paradigm, we propose a more superior version LRQuant+. Specifically, we first propose a dynamically weighted scheme to balance MSE and NLC loss, and then devise learnable rotation vectors to further directly reduce directional gaps. In addition, we improve the block-wise optimization framework into a novel two-branch nature which jointly considers the error propagation and homologous reconstruction error. Extensive experiments demonstrate the superiority of our LRQuantand LRQuant+, as well as their unified effectiveness across various LFMs for both weight-activation and weight-only quantization, especially under challenging quantization scenarios, i.e., W4A4 and W2A16 on LLMs, ViTS, and MLLMs.

Abstract Siamese network based the tracker is a hot topic in the field of visual object tracking. However, Siamese trackers still have a robustness gap compared with state‐of‐the‐art algorithms. Therefore, focusing on the issue, this letter adds Frequency Channel Attention (FCA) and adaptive template feature map to the framework of Siamese neural network. FCA can enhance feature representation of effective channels and improve feature discrimination by modeling the correlation between each channel of the image. In this algorithm, by theoretical analysis and experimental validation, restriction is broken through a simple yet effective FCA network sampling strategy and a Siamese‐FCA tracker with significant performance gain is successfully trained. Meanwhile, in order to better adjust the proportion between target and background, the tracker selects suitable size of the target feature map. Moreover, extensive ablation studies are conducted to demonstrate the effectiveness of the proposed tracker. Fairly, the experimental results of five test benchmarks, including OTB2013, OTB2015, VOT2016, VOT2018 and UAV123 datasets, shows that the proposed algorithm performs outstanding. In particular, the issue of similarity and small target tracking failure is overcome. The average running frame rate reaches 86 frames per second, which can meet the real‐time requirements.