Xiangmin Xu

Yirong Chen, Weiquan Fan, Xiaofen Xing et al. · tsinghua

Human language expression is based on the subjective construal of the situation instead of the objective truth conditions, which means that speakers' personalities and emotions after cognitive processing have an important influence on conversation. However, most existing datasets for conversational AI ignore human personalities and emotions, or only consider part of them. It's difficult for dialogue systems to understand speakers' personalities and emotions although large-scale pre-training language models have been widely used. In order to consider both personalities and emotions in the process of conversation generation, we propose CPED, a large-scale Chinese personalized and emotional dialogue dataset, which consists of multi-source knowledge related to empathy and personal characteristic. These knowledge covers gender, Big Five personality traits, 13 emotions, 19 dialogue acts and 10 scenes. CPED contains more than 12K dialogues of 392 speakers from 40 TV shows. We release the textual dataset with audio features and video features according to the copyright claims, privacy issues, terms of service of video platforms. We provide detailed description of the CPED construction process and introduce three tasks for conversational AI, including personality recognition, emotion recognition in conversations as well as personalized and emotional conversation generation. Finally, we provide baseline systems for these tasks and consider the function of speakers' personalities and emotions on conversation. Our motivation is to propose a dataset to be widely adopted by the NLP community as a new open benchmark for conversational AI research. The full dataset is available at https://github.com/scutcyr/CPED.

Jiahao Sun, Chunmei Qing, Junpeng Tan et al.

Most existing methods realize 3D instance segmentation by extending those models used for 3D object detection or 3D semantic segmentation. However, these non-straightforward methods suffer from two drawbacks: 1) Imprecise bounding boxes or unsatisfactory semantic predictions limit the performance of the overall 3D instance segmentation framework. 2) Existing method requires a time-consuming intermediate step of aggregation. To address these issues, this paper proposes a novel end-to-end 3D instance segmentation method based on Superpoint Transformer, named as SPFormer. It groups potential features from point clouds into superpoints, and directly predicts instances through query vectors without relying on the results of object detection or semantic segmentation. The key step in this framework is a novel query decoder with transformers that can capture the instance information through the superpoint cross-attention mechanism and generate the superpoint masks of the instances. Through bipartite matching based on superpoint masks, SPFormer can implement the network training without the intermediate aggregation step, which accelerates the network. Extensive experiments on ScanNetv2 and S3DIS benchmarks verify that our method is concise yet efficient. Notably, SPFormer exceeds compared state-of-the-art methods by 4.3% on ScanNetv2 hidden test set in terms of mAP and keeps fast inference speed (247ms per frame) simultaneously. Code is available at https://github.com/sunjiahao1999/SPFormer.

Weidong Chen, Xiaofen Xing, Xiangmin Xu et al.

Paralinguistic speech processing is important in addressing many issues, such as sentiment and neurocognitive disorder analyses. Recently, Transformer has achieved remarkable success in the natural language processing field and has demonstrated its adaptation to speech. However, previous works on Transformer in the speech field have not incorporated the properties of speech, leaving the full potential of Transformer unexplored. In this paper, we consider the characteristics of speech and propose a general structure-based framework, called SpeechFormer++, for paralinguistic speech processing. More concretely, following the component relationship in the speech signal, we design a unit encoder to model the intra- and inter-unit information (i.e., frames, phones, and words) efficiently. According to the hierarchical relationship, we utilize merging blocks to generate features at different granularities, which is consistent with the structural pattern in the speech signal. Moreover, a word encoder is introduced to integrate word-grained features into each unit encoder, which effectively balances fine-grained and coarse-grained information. SpeechFormer++ is evaluated on the speech emotion recognition (IEMOCAP & MELD), depression classification (DAIC-WOZ) and Alzheimer's disease detection (Pitt) tasks. The results show that SpeechFormer++ outperforms the standard Transformer while greatly reducing the computational cost. Furthermore, it delivers superior results compared to the state-of-the-art approaches.

Weidong Chen, Xiaofen Xing, Peihao Chen et al.

This paper presents a paradigm that adapts general large-scale pretrained models (PTMs) to speech emotion recognition task. Although PTMs shed new light on artificial general intelligence, they are constructed with general tasks in mind, and thus, their efficacy for specific tasks can be further improved. Additionally, employing PTMs in practical applications can be challenging due to their considerable size. Above limitations spawn another research direction, namely, optimizing large-scale PTMs for specific tasks to generate task-specific PTMs that are both compact and effective. In this paper, we focus on the speech emotion recognition task and propose an improved emotion-specific pretrained encoder called Vesper. Vesper is pretrained on a speech dataset based on WavLM and takes into account emotional characteristics. To enhance sensitivity to emotional information, Vesper employs an emotion-guided masking strategy to identify the regions that need masking. Subsequently, Vesper employs hierarchical and cross-layer self-supervision to improve its ability to capture acoustic and semantic representations, both of which are crucial for emotion recognition. Experimental results on the IEMOCAP, MELD, and CREMA-D datasets demonstrate that Vesper with 4 layers outperforms WavLM Base with 12 layers, and the performance of Vesper with 12 layers surpasses that of WavLM Large with 24 layers.

Yirong Chen, Xiaofen Xing, Jingkai Lin et al.

Large language models (LLMs) have been widely applied in various fields due to their excellent capability for memorizing knowledge and chain of thought (CoT). When these language models are applied in the field of psychological counseling, they often rush to provide universal advice. However, when users seek psychological support, they need to gain empathy, trust, understanding and comfort, rather than just reasonable advice. To this end, we constructed a multi-turn empathetic conversation dataset of more than 2 million samples, in which the input is the multi-turn conversation context, and the target is empathetic responses that cover expressions such as questioning, comfort, recognition, listening, trust, emotional support, etc. Experiments have shown that the empathy ability of LLMs can be significantly enhanced when finetuning by using multi-turn dialogue history and responses that are closer to the expression of a psychological consultant.

Yirong Chen, Zhenyu Wang, Xiaofen Xing et al.

Large language models (LLMs) have performed well in providing general and extensive health suggestions in single-turn conversations, exemplified by systems such as ChatGPT, ChatGLM, ChatDoctor, DoctorGLM, and etc. However, the limited information provided by users during single turn results in inadequate personalization and targeting of the generated suggestions, which requires users to independently select the useful part. It is mainly caused by the missing ability to engage in multi-turn questioning. In real-world medical consultations, doctors usually employ a series of iterative inquiries to comprehend the patient's condition thoroughly, enabling them to provide effective and personalized suggestions subsequently, which can be defined as chain of questioning (CoQ) for LLMs. To improve the CoQ of LLMs, we propose BianQue, a ChatGLM-based LLM finetuned with the self-constructed health conversation dataset BianQueCorpus that is consist of multiple turns of questioning and health suggestions polished by ChatGPT. Experimental results demonstrate that the proposed BianQue can simultaneously balance the capabilities of both questioning and health suggestions, which will help promote the research and application of LLMs in the field of proactive health.

Zhuowen Yin, Xinyao Ding, Xin Zhang et al.

Autism Spectrum Disorder (ASD) has been emerging as a growing public health threat. Early diagnosis of ASD is crucial for timely, effective intervention and treatment. However, conventional diagnosis methods based on communications and behavioral patterns are unreliable for children younger than 2 years of age. Given evidences of neurodevelopmental abnormalities in ASD infants, we resort to a novel deep learning-based method to extract key features from the inherently scarce, class-imbalanced, and heterogeneous structural MR images for early autism diagnosis. Specifically, we propose a Siamese verification framework to extend the scarce data, and an unsupervised compressor to alleviate data imbalance by extracting key features. We also proposed weight constraints to cope with sample heterogeneity by giving different samples different voting weights during validation, and we used Path Signature to unravel meaningful developmental features from the two-time point data longitudinally. We further extracted machine learning focused brain regions for autism diagnosis. Extensive experiments have shown that our method performed well under practical scenarios, transcending existing machine learning methods and providing anatomical insights for autism early diagnosis.

Shuaiqi Chen, Xiaofen Xing, Weibin Zhang et al.

Speech emotion recognition is crucial to human-computer interaction. The temporal regions that represent different emotions scatter in different parts of the speech locally. Moreover, the temporal scales of important information may vary over a large range within and across speech segments. Although transformer-based models have made progress in this field, the existing models could not precisely locate important regions at different temporal scales. To address the issue, we propose Dynamic Window transFormer (DWFormer), a new architecture that leverages temporal importance by dynamically splitting samples into windows. Self-attention mechanism is applied within windows for capturing temporal important information locally in a fine-grained way. Cross-window information interaction is also taken into account for global communication. DWFormer is evaluated on both the IEMOCAP and the MELD datasets. Experimental results show that the proposed model achieves better performance than the previous state-of-the-art methods.

Yan Wang, Yawen Zeng, Jingsheng Zheng et al.

Multimodal large language models (MLLMs) are flourishing, but mainly focus on images with less attention than videos, especially in sub-fields such as prompt engineering, video chain-of-thought (CoT), and instruction tuning on videos. Therefore, we try to explore the collection of CoT datasets in videos to lead to video OpenQA and improve the reasoning ability of MLLMs. Unfortunately, making such video CoT datasets is not an easy task. Given that human annotation is too cumbersome and expensive, while machine-generated is not reliable due to the hallucination issue, we develop an automatic annotation tool that combines machine and human experts, under the active learning paradigm. Active learning is an interactive strategy between the model and human experts, in this way, the workload of human labeling can be reduced and the quality of the dataset can be guaranteed. With the help of the automatic annotation tool, we strive to contribute three datasets, namely VideoCoT, TopicQA, TopicCoT. Furthermore, we propose a simple but effective benchmark based on the collected datasets, which exploits CoT to maximize the complex reasoning capabilities of MLLMs. Extensive experiments demonstrate the effectiveness our solution.

Pucheng Zhai, Kailing Guo, Fang Liu et al.

Structured pruning is a commonly used convolutional neural network (CNN) compression approach. Pruning rate setting is a fundamental problem in structured pruning. Most existing works introduce too many additional learnable parameters to assign different pruning rates across different layers in CNN or cannot control the compression rate explicitly. Since too narrow network blocks information flow for training, automatic pruning rate setting cannot explore a high pruning rate for a specific layer. To overcome these limitations, we propose a novel framework named Layer Adaptive Progressive Pruning (LAPP), which gradually compresses the network during initial training of a few epochs from scratch. In particular, LAPP designs an effective and efficient pruning strategy that introduces a learnable threshold for each layer and FLOPs constraints for network. Guided by both task loss and FLOPs constraints, the learnable thresholds are dynamically and gradually updated to accommodate changes of importance scores during training. Therefore the pruning strategy can gradually prune the network and automatically determine the appropriate pruning rates for each layer. What's more, in order to maintain the expressive power of the pruned layer, before training starts, we introduce an additional lightweight bypass for each convolutional layer to be pruned, which only adds relatively few additional burdens. Our method demonstrates superior performance gains over previous compression methods on various datasets and backbone architectures. For example, on CIFAR-10, our method compresses ResNet-20 to 40.3% without accuracy drop. 55.6% of FLOPs of ResNet-18 are reduced with 0.21% top-1 accuracy increase and 0.40% top-5 accuracy increase on ImageNet.

Kan Wang, Changxing Ding, Jianxin Pang et al.

Video-based person re-identification (ReID) is challenging due to the presence of various interferences in video frames. Recent approaches handle this problem using temporal aggregation strategies. In this work, we propose a novel Context Sensing Attention Network (CSA-Net), which improves both the frame feature extraction and temporal aggregation steps. First, we introduce the Context Sensing Channel Attention (CSCA) module, which emphasizes responses from informative channels for each frame. These informative channels are identified with reference not only to each individual frame, but also to the content of the entire sequence. Therefore, CSCA explores both the individuality of each frame and the global context of the sequence. Second, we propose the Contrastive Feature Aggregation (CFA) module, which predicts frame weights for temporal aggregation. Here, the weight for each frame is determined in a contrastive manner: i.e., not only by the quality of each individual frame, but also by the average quality of the other frames in a sequence. Therefore, it effectively promotes the contribution of relatively good frames. Extensive experimental results on four datasets show that CSA-Net consistently achieves state-of-the-art performance.

Kailing Guo, Zhenquan Lin, Canyang Chen et al.

Low-rankness plays an important role in traditional machine learning, but is not so popular in deep learning. Most previous low-rank network compression methods compress networks by approximating pre-trained models and re-training. However, the optimal solution in the Euclidean space may be quite different from the one with low-rank constraint. A well-pre-trained model is not a good initialization for the model with low-rank constraints. Thus, the performance of a low-rank compressed network degrades significantly. Compared with other network compression methods such as pruning, low-rank methods attract less attention in recent years. In this paper, we devise a new training method, low-rank projection with energy transfer (LRPET), that trains low-rank compressed networks from scratch and achieves competitive performance. We propose to alternately perform stochastic gradient descent training and projection of each weight matrix onto the corresponding low-rank manifold. Compared to re-training on the compact model, this enables full utilization of model capacity since solution space is relaxed back to Euclidean space after projection. The matrix energy (the sum of squares of singular values) reduction caused by projection is compensated by energy transfer. We uniformly transfer the energy of the pruned singular values to the remaining ones. We theoretically show that energy transfer eases the trend of gradient vanishing caused by projection. In modern networks, a batch normalization (BN) layer can be merged into the previous convolution layer for inference, thereby influencing the optimal low-rank approximation of the previous layer. We propose BN rectification to cut off its effect on the optimal low-rank approximation, which further improves the performance.

Jialong Mai, Xiaofen Xing, Xiangmin Xu

Fine-grained local timing control is still absent from modern text-to-speech systems: existing approaches typically provide only utterance-level duration or global speaking-rate control, while precise token-level timing manipulation remains unavailable. To the best of our knowledge, MAGIC-TTS is the first TTS model with explicit local timing control over token-level content duration and pause. MAGIC-TTS is enabled by explicit token-level duration conditioning, carefully prepared high-confidence duration supervision, and training mechanisms that correct zero-value bias and make the model robust to missing local controls. On our timing-control benchmark, MAGIC-TTS substantially improves token-level duration and pause following over spontaneous synthesis. Even when no timing control is provided, MAGIC-TTS maintains natural high-quality synthesis. We further evaluate practical local editing with a scenario-based benchmark covering navigation guidance, guided reading, and accessibility-oriented code reading. In this setting, MAGIC-TTS realizes a reproducible uniform-timing baseline and then moves the edited regions toward the requested local targets with low mean bias. These results show that explicit fine-grained controllability can be implemented effectively in a high-quality TTS system and can support realistic local timing-editing applications.

Zhaojie Chu, Kailing Guo, Xiaofen Xing et al.

Speech-driven 3D facial animation is a challenging cross-modal task that has attracted growing research interest. During speaking activities, the mouth displays strong motions, while the other facial regions typically demonstrate comparatively weak activity levels. Existing approaches often simplify the process by directly mapping single-level speech features to the entire facial animation, which overlook the differences in facial activity intensity leading to overly smoothed facial movements. In this study, we propose a novel framework, CorrTalk, which effectively establishes the temporal correlation between hierarchical speech features and facial activities of different intensities across distinct regions. A novel facial activity intensity metric is defined to distinguish between strong and weak facial activity, obtained by computing the short-time Fourier transform of facial vertex displacements. Based on the variances in facial activity, we propose a dual-branch decoding framework to synchronously synthesize strong and weak facial activity, which guarantees wider intensity facial animation synthesis. Furthermore, a weighted hierarchical feature encoder is proposed to establish temporal correlation between hierarchical speech features and facial activity at different intensities, which ensures lip-sync and plausible facial expressions. Extensive qualitatively and quantitatively experiments as well as a user study indicate that our CorrTalk outperforms existing state-of-the-art methods. The source code and supplementary video are publicly available at: https://zjchu.github.io/projects/CorrTalk/

Sijin Yu, Zijiao Chen, Zhenyu Yang et al.

Current fMRI decoders face a performance-fidelity trade-off where efficient ID encoders outperform geometrically faithful surface-based models. We argue this is partly driven by inefficient surface tokenization and the failure to use anatomy as a predictive signal. We present NeurIPS, a framework that improves surface-based decoding by reframing anatomical variation from a nuisance to a powerful inductive prior. NeurIPS unites two innovations: a Selective ROI Spherical Tokenizer (SRST) for efficient geometric encoding, and a Structure-Guided Mixture of Experts (SG-MoE) that explicitly models individual anatomy using cortical features. On the Natural Scenes Dataset, NeurIPS establishes a new state-of-the-art for surface decoders and achieves performance comparable to strong 1D baselines. This is achieved with unprecedented efficiency, as the model converges dramatically faster (10 vs. 600 epochs). This efficiency enables rapid adaptation to new subjects using only 20% of data and ensures robust scalability as the training cohort is expanded. Ablations provide causal evidence that these gains are driven by the model's use of cortical features, not by memorizing subject IDs. By leveraging anatomical priors, NeurIPS provides a principled and scalable path toward robust, generalizable brain decoding.

Zhili Lai, Chunmei Qing, Junpeng Tan et al.

Utilizing functional near-infrared spectroscopy (fNIRS) signals for emotion recognition is a significant advancement in understanding human emotions. However, due to the lack of artificial intelligence data and algorithms in this field, current research faces the following challenges: 1) The portable wearable devices have higher requirements for lightweight models; 2) The objective differences of physiology and psychology among different subjects aggravate the difficulty of emotion recognition. To address these challenges, we propose a novel cross-subject fNIRS emotion recognition method, called the Online Multi-level Contrastive Representation Distillation framework (OMCRD). Specifically, OMCRD is a framework designed for mutual learning among multiple lightweight student networks. It utilizes multi-level fNIRS feature extractor for each sub-network and conducts multi-view sentimental mining using physiological signals. The proposed Inter-Subject Interaction Contrastive Representation (IS-ICR) facilitates knowledge transfer for interactions between student models, enhancing cross-subject emotion recognition performance. The optimal student network can be selected and deployed on a wearable device. Some experimental results demonstrate that OMCRD achieves state-of-the-art results in emotional perception and affective imagery tasks.

Sihang Nie, Xiaofen Xing, Jingyuan Xing et al.

Large Language Model (LLM)-based Text-to-Speech (TTS) models have already reached a high degree of naturalness. However, the precision control of TTS inference is still challenging. Although instruction-based Text-to-Speech (Instruct-TTS) models are proposed, these models still lack fine-grained control due to the modality gap between single-level text instructions and multilevel speech tokens. To address this limitation, we propose HD-PPT, a framework that transforms speech synthesis into a structured, hierarchical task. To enable fine-grained control, we introduce a novel speech codec to extract distinct prompt-preference and content-preference tokens from the complex speech tokens, supervised by automatic speech recognition (ASR) and cross-lingual audio-text pre-training (CLAP) objectives. To bridge the modality gap of these tokens, we propose a hierarchical decoding strategy, where the LLM generates tokens in a structured order: first semantic, then fine-grained style, and finally complete acoustic representation. Extensive experiments demonstrate that this hierarchical paradigm significantly improves instruction adherence and achieves state-of-the-art naturalness, validating our approach for precise and controllable speech synthesis. Audio samples are available at https://xxh333.github.io/.

Zhenyu Wang, Xiaofen Xing, Yirong Chen et al.

Multimodal Role-Playing Agents (MRPAs) are attracting increasing attention due to their ability to deliver more immersive multimodal emotional interactions. However, existing studies still rely on pure textual benchmarks to evaluate the text responses of MRPAs, while delegating the assessment of their multimodal expressions solely to modality-synthesis metrics. This evaluation paradigm, on the one hand, entangles semantic assessment with modality generation, leading to ambiguous error attribution, and on the other hand remains constrained by the heavy reliance on human judgment. To this end, we propose MERRY, a semantically decoupled evaluation framework for assessing Multimodal Emotional and Role consistencies of Role-playing agents. This framework introduce five refined metrics for EC and three for RC. Notably, we transform the traditional subjective scoring approach into a novel bidirectional-evidence-finding task, significantly improving the human agreement of LLM-as-Judge evaluations. Based on MERRY, we conduct extensive evaluations. Our empirical results primarily reveal that: (1) Training on synthetic datasets tends to reduce emotional consistency, whereas training on real-world datasets improves it; (2) Existing models suffer from emotional templatization and simplification, exhibiting positive-bias and performance bottleneck in fine-grained negative emotions; (3) Simple prompting method strengthens the weak models but constrains the strong ones, while simple fine-tuning method suffers from poor role generalization. Codes and dataset are available.

Junpeng Tan, Chunmei Qing, Xiangmin Xu

Compressed Sensing (CS) significantly speeds up Magnetic Resonance Image (MRI) processing and achieves accurate MRI reconstruction from under-sampled k-space data. According to the current research, there are still several problems with dynamic MRI k-space reconstruction based on CS. 1) There are differences between the Fourier domain and the Image domain, and the differences between MRI processing of different domains need to be considered. 2) As three-dimensional data, dynamic MRI has its spatial-temporal characteristics, which need to calculate the difference and consistency of surface textures while preserving structural integrity and uniqueness. 3) Dynamic MRI reconstruction is time-consuming and computationally resource-dependent. In this paper, we propose a novel robust low-rank dynamic MRI reconstruction optimization model via highly under-sampled and Discrete Fourier Transform (DFT) called the Robust Depth Linear Error Decomposition Model (RDLEDM). Our method mainly includes linear decomposition, double Total Variation (TV), and double Nuclear Norm (NN) regularizations. By adding linear image domain error analysis, the noise is reduced after under-sampled and DFT processing, and the anti-interference ability of the algorithm is enhanced. Double TV and NN regularizations can utilize both spatial-temporal characteristics and explore the complementary relationship between different dimensions in dynamic MRI sequences. In addition, Due to the non-smoothness and non-convexity of TV and NN terms, it is difficult to optimize the unified objective model. To address this issue, we utilize a fast algorithm by solving a primal-dual form of the original problem. Compared with five state-of-the-art methods, extensive experiments on dynamic MRI data demonstrate the superior performance of the proposed method in terms of both reconstruction accuracy and time complexity.

Kaijun Gong, Zhuowen Yin, Yushu Li et al.

The redundancy of Convolutional neural networks not only depends on weights but also depends on inputs. Shuffling is an efficient operation for mixing channel information but the shuffle order is usually pre-defined. To reduce the data-dependent redundancy, we devise a dynamic shuffle module to generate data-dependent permutation matrices for shuffling. Since the dimension of permutation matrix is proportional to the square of the number of input channels, to make the generation process efficiently, we divide the channels into groups and generate two shared small permutation matrices for each group, and utilize Kronecker product and cross group shuffle to obtain the final permutation matrices. To make the generation process learnable, based on theoretical analysis, softmax, orthogonal regularization, and binarization are employed to asymptotically approximate the permutation matrix. Dynamic shuffle adaptively mixes channel information with negligible extra computation and memory occupancy. Experiment results on image classification benchmark datasets CIFAR-10, CIFAR-100, Tiny ImageNet and ImageNet have shown that our method significantly increases ShuffleNets' performance. Adding dynamic generated matrix with learnable static matrix, we further propose static-dynamic-shuffle and show that it can serve as a lightweight replacement of ordinary pointwise convolution.

Zhuolong Li, Xingao Li, Changxing Ding et al.

Detecting human-object interaction (HOI) has long been limited by the amount of supervised data available. Recent approaches address this issue by pre-training according to pseudo-labels, which align object regions with HOI triplets parsed from image captions. However, pseudo-labeling is tricky and noisy, making HOI pre-training a complex process. Therefore, we propose an efficient disentangled pre-training method for HOI detection (DP-HOI) to address this problem. First, DP-HOI utilizes object detection and action recognition datasets to pre-train the detection and interaction decoder layers, respectively. Then, we arrange these decoder layers so that the pre-training architecture is consistent with the downstream HOI detection task. This facilitates efficient knowledge transfer. Specifically, the detection decoder identifies reliable human instances in each action recognition dataset image, generates one corresponding query, and feeds it into the interaction decoder for verb classification. Next, we combine the human instance verb predictions in the same image and impose image-level supervision. The DP-HOI structure can be easily adapted to the HOI detection task, enabling effective model parameter initialization. Therefore, it significantly enhances the performance of existing HOI detection models on a broad range of rare categories. The code and pre-trained weight are available at https://github.com/xingaoli/DP-HOI.

Junpeng Tan, Xin Zhang, Yao Lv et al.

Although the use of multiple stacks can handle slice-to-volume motion correction and artifact removal problems, there are still several problems: 1) The slice-to-volume method usually uses slices as input, which cannot solve the problem of uniform intensity distribution and complementarity in regions of different fetal MRI stacks; 2) The integrity of 3D space is not considered, which adversely affects the discrimination and generation of globally consistent information in fetal MRI; 3) Fetal MRI with severe motion artifacts in the real-world cannot achieve high-quality super-resolution reconstruction. To address these issues, we propose a novel fetal brain MRI high-quality volume reconstruction method, called the Radiation Diffusion Generation Model (RDGM). It is a self-supervised generation method, which incorporates the idea of Neural Radiation Field (NeRF) based on the coordinate generation and diffusion model based on super-resolution generation. To solve regional intensity heterogeneity in different directions, we use a pre-trained transformer model for slice registration, and then, a new regionally Consistent Implicit Neural Representation (CINR) network sub-module is proposed. CINR can generate the initial volume by combining a coordinate association map of two different coordinate mapping spaces. To enhance volume global consistency and discrimination, we introduce the Volume Diffusion Super-resolution Generation (VDSG) mechanism. The global intensity discriminant generation from volume-to-volume is carried out using the idea of diffusion generation, and CINR becomes the deviation intensity generation network of the volume-to-volume diffusion model. Finally, the experimental results on real-world fetal brain MRI stacks demonstrate the state-of-the-art performance of our method.

Yuanbo Fang, Haoze Sun, Jun Liu et al.

End-to-end speech large language models ((LLMs)) extend the capabilities of text-based models to directly process and generate audio tokens. However, this often leads to a decline in reasoning and generation performance compared to text input, a phenomenon referred to as intelligence degradation. To systematically evaluate this gap, we propose S2SBench, a benchmark designed to quantify performance degradation in Speech LLMs. It includes diagnostic datasets targeting sentence continuation and commonsense reasoning under audio input. We further introduce a pairwise evaluation protocol based on perplexity differences between plausible and implausible samples to measure degradation relative to text input. We apply S2SBench to analyze the training process of Baichuan-Audio, which further demonstrates the benchmark's effectiveness. All datasets and evaluation code are available at https://github.com/undobug/S2SBench.

Zhenquan Lin, Kailing Guo, Xiaofen Xing et al.

To obtain good performance, convolutional neural networks are usually over-parameterized. This phenomenon has stimulated two interesting topics: pruning the unimportant weights for compression and reactivating the unimportant weights to make full use of network capability. However, current weight reactivation methods usually reactivate the entire filters, which may not be precise enough. Looking back in history, the prosperity of filter pruning is mainly due to its friendliness to hardware implementation, but pruning at a finer structure level, i.e., weight elements, usually leads to better network performance. We study the problem of weight element reactivation in this paper. Motivated by evolution, we select the unimportant filters and update their unimportant elements by combining them with the important elements of important filters, just like gene crossover to produce better offspring, and the proposed method is called weight evolution (WE). WE is mainly composed of four strategies. We propose a global selection strategy and a local selection strategy and combine them to locate the unimportant filters. A forward matching strategy is proposed to find the matched important filters and a crossover strategy is proposed to utilize the important elements of the important filters for updating unimportant filters. WE is plug-in to existing network architectures. Comprehensive experiments show that WE outperforms the other reactivation methods and plug-in training methods with typical convolutional neural networks, especially lightweight networks. Our code is available at https://github.com/BZQLin/Weight-evolution.

Weiquan Fan, Xiangmin Xu, Xiaofen Xing et al.

Speech emotion recognition is a vital contributor to the next generation of human-computer interaction (HCI). However, current existing small-scale databases have limited the development of related research. In this paper, we present LSSED, a challenging large-scale english speech emotion dataset, which has data collected from 820 subjects to simulate real-world distribution. In addition, we release some pre-trained models based on LSSED, which can not only promote the development of speech emotion recognition, but can also be transferred to related downstream tasks such as mental health analysis where data is extremely difficult to collect. Finally, our experiments show the necessity of large-scale datasets and the effectiveness of pre-trained models. The dateset will be released on https://github.com/tobefans/LSSED.

Xiangmin Xu, Zhen Meng, Yichi Zhang et al.

Real-time three-dimensional (3D) scene representations serve as one of the building blocks that bolster various innovative applications, e.g., digital manufacturing, Virtual/Augmented/Extended/Mixed Reality (VR/AR/XR/MR), and the metaverse. Despite substantial efforts that have been made to real-time communications and computing, real-time 3D scene representations remain a challenging task. This paper investigates the tradeoff between timeliness and fidelity in real-time 3D scene representations. Specifically, we establish a framework to evaluate the impact of communication delay on the tradeoff, where the real-world scenario is monitored by multiple cameras that communicate with an edge server. To improve fidelity for 3D scene representations, we propose to use a single-step Proximal Policy Optimization (PPO) method that leverages the Age of Information (AoI) to decide if the received image needs to be involved in 3D scene representations and rendering. We test our framework and the proposed approach with different well-known 3D scene representation methods. Simulation results reveal that real-time 3D scene representation can be sensitively affected by communication delay, and our proposed method can achieve optimal 3D scene representation results.

Xu Yang, Changxing Ding, Zhibin Hong et al.

Image-based virtual try-on is an increasingly important task for online shopping. It aims to synthesize images of a specific person wearing a specified garment. Diffusion model-based approaches have recently become popular, as they are excellent at image synthesis tasks. However, these approaches usually employ additional image encoders and rely on the cross-attention mechanism for texture transfer from the garment to the person image, which affects the try-on's efficiency and fidelity. To address these issues, we propose an Texture-Preserving Diffusion (TPD) model for virtual try-on, which enhances the fidelity of the results and introduces no additional image encoders. Accordingly, we make contributions from two aspects. First, we propose to concatenate the masked person and reference garment images along the spatial dimension and utilize the resulting image as the input for the diffusion model's denoising UNet. This enables the original self-attention layers contained in the diffusion model to achieve efficient and accurate texture transfer. Second, we propose a novel diffusion-based method that predicts a precise inpainting mask based on the person and reference garment images, further enhancing the reliability of the try-on results. In addition, we integrate mask prediction and image synthesis into a single compact model. The experimental results show that our approach can be applied to various try-on tasks, e.g., garment-to-person and person-to-person try-ons, and significantly outperforms state-of-the-art methods on popular VITON, VITON-HD databases.

Haojie Xie, Yirong Chen, Xiaofen Xing et al.

Currently, large language models (LLMs) have made significant progress in the field of psychological counseling. However, existing mental health LLMs overlook a critical issue where they do not consider the fact that different psychological counselors exhibit different personal styles, including linguistic style and therapy techniques, etc. As a result, these LLMs fail to satisfy the individual needs of clients who seek different counseling styles. To help bridge this gap, we propose PsyDT, a novel framework using LLMs to construct the Digital Twin of Psychological counselor with personalized counseling style. Compared to the time-consuming and costly approach of collecting a large number of real-world counseling cases to create a specific counselor's digital twin, our framework offers a faster and more cost-effective solution. To construct PsyDT, we utilize dynamic one-shot learning by using GPT-4 to capture counselor's unique counseling style, mainly focusing on linguistic style and therapy techniques. Subsequently, using existing single-turn long-text dialogues with client's questions, GPT-4 is guided to synthesize multi-turn dialogues of specific counselor. Finally, we fine-tune the LLMs on the synthetic dataset, PsyDTCorpus, to achieve the digital twin of psychological counselor with personalized counseling style. Experimental results indicate that our proposed PsyDT framework can synthesize multi-turn dialogues that closely resemble real-world counseling cases and demonstrate better performance compared to other baselines, thereby show that our framework can effectively construct the digital twin of psychological counselor with a specific counseling style.

Xiangyu Li, Yawen Zeng, Xiaofen Xing et al.

As automated trading gains traction in the financial market, algorithmic investment strategies are increasingly prominent. While Large Language Models (LLMs) and Agent-based models exhibit promising potential in real-time market analysis and trading decisions, they still experience a significant -20% loss when confronted with rapid declines or frequent fluctuations, impeding their practical application. Hence, there is an imperative to explore a more robust and resilient framework. This paper introduces an innovative multi-agent system, HedgeAgents, aimed at bolstering system robustness via ``hedging'' strategies. In this well-balanced system, an array of hedging agents has been tailored, where HedgeAgents consist of a central fund manager and multiple hedging experts specializing in various financial asset classes. These agents leverage LLMs' cognitive capabilities to make decisions and coordinate through three types of conferences. Benefiting from the powerful understanding of LLMs, our HedgeAgents attained a 70% annualized return and a 400% total return over a period of 3 years. Moreover, we have observed with delight that HedgeAgents can even formulate investment experience comparable to those of human experts (https://hedgeagents.github.io/).

Baozhu Zhao, Qiwei Xiong, Xiaohan Zhang et al.

Three-dimensional point cloud anomaly detection that aims to detect anomaly data points from a training set serves as the foundation for a variety of applications, including industrial inspection and autonomous driving. However, existing point cloud anomaly detection methods often incorporate multiple feature memory banks to fully preserve local and global representations, which comes at the high cost of computational complexity and mismatches between features. To address that, we propose an unsupervised point cloud anomaly detection framework based on joint local-global features, termed PointCore. To be specific, PointCore only requires a single memory bank to store local (coordinate) and global (PointMAE) representations and different priorities are assigned to these local-global features, thereby reducing the computational cost and mismatching disturbance in inference. Furthermore, to robust against the outliers, a normalization ranking method is introduced to not only adjust values of different scales to a notionally common scale, but also transform densely-distributed data into a uniform distribution. Extensive experiments on Real3D-AD dataset demonstrate that PointCore achieves competitive inference time and the best performance in both detection and localization as compared to the state-of-the-art Reg3D-AD approach and several competitors.

Andrea Avogaro, Andrea Toaiari, Federico Cunico et al.

We introduce HARPER, a novel dataset for 3D body pose estimation and forecast in dyadic interactions between users and Spot, the quadruped robot manufactured by Boston Dynamics. The key-novelty is the focus on the robot's perspective, i.e., on the data captured by the robot's sensors. These make 3D body pose analysis challenging because being close to the ground captures humans only partially. The scenario underlying HARPER includes 15 actions, of which 10 involve physical contact between the robot and users. The Corpus contains not only the recordings of the built-in stereo cameras of Spot, but also those of a 6-camera OptiTrack system (all recordings are synchronized). This leads to ground-truth skeletal representations with a precision lower than a millimeter. In addition, the Corpus includes reproducible benchmarks on 3D Human Pose Estimation, Human Pose Forecasting, and Collision Prediction, all based on publicly available baseline approaches. This enables future HARPER users to rigorously compare their results with those we provide in this work.

Weiying Xue, Qi Liu, Qiwei Xiong et al.

Human-object interaction (HOI) detection aims to locate human-object pairs and identify their interaction categories in images. Most existing methods primarily focus on supervised learning, which relies on extensive manual HOI annotations. In this paper, we propose a novel framework, termed Knowledge Integration to HOI (KI2HOI), that effectively integrates the knowledge of visual-language model to improve zero-shot HOI detection. Specifically, the verb feature learning module is designed based on visual semantics, by employing the verb extraction decoder to convert corresponding verb queries into interaction-specific category representations. We develop an effective additive self-attention mechanism to generate more comprehensive visual representations. Moreover, the innovative interaction representation decoder effectively extracts informative regions by integrating spatial and visual feature information through a cross-attention mechanism. To deal with zero-shot learning in low-data, we leverage a priori knowledge from the CLIP text encoder to initialize the linear classifier for enhanced interaction understanding. Extensive experiments conducted on the mainstream HICO-DET and V-COCO datasets demonstrate that our model outperforms the previous methods in various zero-shot and full-supervised settings.

Jiayu Jiang, Changxing Ding, Wentao Tan et al.

Text-to-image person re-identification (ReID) aims to retrieve the images of an interested person based on textual descriptions. One main challenge for this task is the high cost in manually annotating large-scale databases, which affects the generalization ability of ReID models. Recent works handle this problem by leveraging Multi-modal Large Language Models (MLLMs) to describe pedestrian images automatically. However, the captions produced by MLLMs lack diversity in description styles. To address this issue, we propose a Human Annotator Modeling (HAM) approach to enable MLLMs to mimic the description styles of thousands of human annotators. Specifically, we first extract style features from human textual descriptions and perform clustering on them. This allows us to group textual descriptions with similar styles into the same cluster. Then, we employ a prompt to represent each of these clusters and apply prompt learning to mimic the description styles of different human annotators. Furthermore, we define a style feature space and perform uniform sampling in this space to obtain more diverse clustering prototypes, which further enriches the diversity of the MLLM-generated captions. Finally, we adopt HAM to automatically annotate a massive-scale database for text-to-image ReID. Extensive experiments on this database demonstrate that it significantly improves the generalization ability of ReID models.

Huachao Yan, Kailing Guo, Shiwei Song et al.

Diagnosing seizure onset zone (SOZ) is a challenge in neurosurgery, where stereoelectroencephalography (sEEG) serves as a critical technique. In sEEG SOZ identification, the existing studies focus solely on the intra-patient representation of epileptic information, overlooking the general features of epilepsy across patients and feature interdependencies between feature elements in each contact site. In order to address the aforementioned challenges, we propose the shared attention-based autoencoder (sATAE). sATAE is trained by sEEG data across all patients, with attention blocks introduced to enhance the representation of interdependencies between feature elements. Considering the spatial diversity of sEEG across patients, we introduce graph-based method for identification SOZ of each patient. However, the current graph-based methods for sEEG SOZ identification rely exclusively on static graphs to model epileptic networks. Inspired by the finding of neuroscience that epileptic network is intricately characterized by the interplay of sophisticated equilibrium between fluctuating and stable states, we design the hierarchical fusion-based graph convolution network (HFGCN) to identify the SOZ. HFGCN integrates the dynamic and static characteristics of epileptic networks through hierarchical weighting across different hierarchies, facilitating a more comprehensive learning of epileptic features and enriching node information for sEEG SOZ identification. Combining sATAE and HFGCN, we perform comprehensive experiments with sATAE-HFGCN on the self-build sEEG dataset, which includes sEEG data from 17 patients with temporal lobe epilepsy. The results show that our method, sATAE-HFGCN, achieves superior performance for identifying the SOZ of each patient, effectively addressing the aforementioned challenges, providing an efficient solution for sEEG-based SOZ identification.

Huimin Zheng, Xiaofeng Xing, Xiangmin Xu

We present a novel approach to personalized sleep health management using few-shot Chain-of-Thought (CoT) distillation, enabling small-scale language models (> 2B parameters) to rival the performance of large language models (LLMs) in specialized health domains. Our method simultaneously distills problem-solving strategies, long-tail expert knowledge, and personalized recommendation capabilities from larger models into more efficient, compact models. Unlike existing systems, our approach offers three key functionalities: generating personalized sleep health recommendations, supporting user-specific follow-up inquiries, and providing responses to domain-specific knowledge questions. We focus on sleep health due to its measurability via wearable devices and its impact on overall well-being. Our experimental setup, involving GPT-4o for data synthesis, Qwen-max for instruction set creation, and Qwen2.5 1.5B for model distillation, demonstrates significant improvements over baseline small-scale models in penalization, reasoning, and knowledge application. Experiments using 100 simulated sleep reports and 1,000 domain-specific questions shows our model achieves comparable performance to larger models while maintaining efficiency for real-world deployment. This research not only advances AI-driven health management but also provides a novel approach to leveraging LLM capabilities in resource-constrained environments, potentially enhancing the accessibility of personalized healthcare solutions.

Xiangyu Li, Yawen Zeng, Xiaofen Xing et al.

LLM-based financial agents have attracted widespread excitement for their ability to trade like human experts. However, most systems exhibit a "profit mirage": dazzling back-tested returns evaporate once the model's knowledge window ends, because of the inherent information leakage in LLMs. In this paper, we systematically quantify this leakage issue across four dimensions and release FinLake-Bench, a leakage-robust evaluation benchmark. Furthermore, to mitigate this issue, we introduce FactFin, a framework that applies counterfactual perturbations to compel LLM-based agents to learn causal drivers instead of memorized outcomes. FactFin integrates four core components: Strategy Code Generator, Retrieval-Augmented Generation, Monte Carlo Tree Search, and Counterfactual Simulator. Extensive experiments show that our method surpasses all baselines in out-of-sample generalization, delivering superior risk-adjusted performance.

Sijin Yu, Zijiao Chen, Wenxuan Wu et al.

Reconstructing visual stimuli from human brain activity (e.g., fMRI) bridges neuroscience and computer vision by decoding neural representations. However, existing methods often overlook critical brain structure-function relationships, flattening spatial information and neglecting individual anatomical variations. To address these issues, we propose (1) a novel sphere tokenizer that explicitly models fMRI signals as spatially coherent 2D spherical data on the cortical surface; (2) integration of structural MRI (sMRI) data, enabling personalized encoding of individual anatomical variations; and (3) a positive-sample mixup strategy for efficiently leveraging multiple fMRI scans associated with the same visual stimulus. Collectively, these innovations enhance reconstruction accuracy, biological interpretability, and generalizability across individuals. Experiments demonstrate superior reconstruction performance compared to SOTA methods, highlighting the effectiveness and interpretability of our biologically informed approach.

Lianlian Liu, YongKang He, Zhaojie Chu et al.

Generating stylized 3D human motion from speech signals presents substantial challenges, primarily due to the intricate and fine-grained relationships among speech signals, individual styles, and the corresponding body movements. Current style encoding approaches either oversimplify stylistic diversity or ignore regional motion style differences (e.g., upper vs. lower body), limiting motion realism. Additionally, motion style should dynamically adapt to changes in speech rhythm and emotion, but existing methods often overlook this. To address these issues, we propose MimicParts, a novel framework designed to enhance stylized motion generation based on part-aware style injection and part-aware denoising network. It divides the body into different regions to encode localized motion styles, enabling the model to capture fine-grained regional differences. Furthermore, our part-aware attention block allows rhythm and emotion cues to guide each body region precisely, ensuring that the generated motion aligns with variations in speech rhythm and emotional state. Experimental results show that our method outperforming existing methods showcasing naturalness and expressive 3D human motion sequences.

Xiangyu Li, Yawen Zeng, Xiaofen Xing et al.

In this paper, our objective is to develop a multi-agent financial system that incorporates simulated trading, a technique extensively utilized by financial professionals. While current LLM-based agent models demonstrate competitive performance, they still exhibit significant deviations from real-world fund companies. A critical distinction lies in the agents' reliance on ``post-reflection'', particularly in response to adverse outcomes, but lack a distinctly human capability: long-term prediction of future trends. Therefore, we introduce QuantAgents, a multi-agent system integrating simulated trading, to comprehensively evaluate various investment strategies and market scenarios without assuming actual risks. Specifically, QuantAgents comprises four agents: a simulated trading analyst, a risk control analyst, a market news analyst, and a manager, who collaborate through several meetings. Moreover, our system incentivizes agents to receive feedback on two fronts: performance in real-world markets and predictive accuracy in simulated trading. Extensive experiments demonstrate that our framework excels across all metrics, yielding an overall return of nearly 300% over the three years (https://quantagents.github.io/).

Mingyu Chen, Jingkai Lin, Zhaojie Chu et al.

Recently, advancements in AI counseling based on large language models have shown significant progress. However, existing studies employ a one-time generation approach to synthesize multi-turn dialogue samples, resulting in low therapy fidelity and failing to capture the decision-making rationale behind each response. In this work, we propose CATCH, a novel data synthesis framework designed to address these challenges. Specifically, to improve therapy fidelity, we introduce the Progressive Dialogue Synthesis strategy, which extracts goals, resources, and solutions from a client's self-report, organizes them into structured outlines, and then incrementally generates stage-aligned counseling dialogues. To capture decision-making rationale behind each response, we propose the Memory-Driven Dynamic Planning thinking pattern that integrates memory enhancement, global planning, and strategy reasoning; a collaborative multi-agent optimizer then leverages MDP to attach explicit chain-of-thought to each dialogue turn. Extensive experiments and human evaluations demonstrate that CATCH significantly enhances fidelity and logical coherence in AI counseling.

Xiangmin Xu, Zhen Meng, Kan Chen et al.

Real-time Three-dimensional (3D) scene representation is a foundational element that supports a broad spectrum of cutting-edge applications, including digital manufacturing, Virtual, Augmented, and Mixed Reality (VR/AR/MR), and the emerging metaverse. Despite advancements in real-time communication and computing, achieving a balance between timeliness and fidelity in 3D scene representation remains a challenge. This work investigates a wireless network where multiple homogeneous mobile robots, equipped with cameras, capture an environment and transmit images to an edge server over channels for 3D representation. We propose a contextual-bandit Proximal Policy Optimization (PPO) framework incorporating both Age of Information (AoI) and semantic information to optimize image selection for representation, balancing data freshness and representation quality. Two policies -- the $ω$-threshold and $ω$-wait policies -- together with two benchmark methods are evaluated, timeliness embedding and weighted sum, on standard datasets and baseline 3D scene representation models. Experimental results demonstrate improved representation fidelity while maintaining low latency, offering insight into the model's decision-making process. This work advances real-time 3D scene representation by optimizing the trade-off between timeliness and fidelity in dynamic environments.

Jialong Mai, Jinxin Ji, Xiaofen Xing et al.

Mainstream Automatic Speech Recognition (ASR) systems excel at transcribing lexical content, but largely fail to recognize nonverbal vocalizations (NVs) embedded in speech, such as sighs, laughs, and coughs. This capability is important for a comprehensive understanding of human communication, as NVs convey crucial emotional and intentional cues. Progress in NV-aware ASR has been hindered by the lack of high-quality, well-annotated datasets. To address this gap, we introduce MNV-17, a 7.55-hour performative Mandarin speech dataset. Unlike most existing corpora that rely on model-based detection, MNV-17's performative nature ensures high-fidelity, clearly articulated NV instances. To the best of our knowledge, MNV-17 provides the most extensive set of nonverbal vocalization categories, comprising 17 distinct and well-balanced classes of common NVs. We benchmarked MNV-17 on four mainstream ASR architectures, evaluating their joint performance on semantic transcription and NV classification. The dataset and the pretrained model checkpoints will be made publicly available to facilitate future research in expressive ASR.

Kan Chen, Zhen Meng, Xiangmin Xu et al.

Real-time human-device interaction in industrial Metaverse faces challenges such as high computational load, limited bandwidth, and strict latency. This paper proposes a task-oriented edge-assisted cross-system framework using digital twins (DTs) to enable responsive interactions. By predicting operator motions, the system supports: 1) proactive Metaverse rendering for visual feedback, and 2) preemptive control of remote devices. The DTs are decoupled into two virtual functions-visual display and robotic control-optimizing both performance and adaptability. To enhance generalizability, we introduce the Human-In-The-Loop Model-Agnostic Meta-Learning (HITL-MAML) algorithm, which dynamically adjusts prediction horizons. Evaluation on two tasks demonstrates the framework's effectiveness: in a Trajectory-Based Drawing Control task, it reduces weighted RMSE from 0.0712 m to 0.0101 m; in a real-time 3D scene representation task for nuclear decommissioning, it achieves a PSNR of 22.11, SSIM of 0.8729, and LPIPS of 0.1298. These results show the framework's capability to ensure spatial precision and visual fidelity in real-time, high-risk industrial environments.

Jialong Mai, Xiaofen Xing, Yawei Li et al.

Recent research has focused on applying speech large language model (SLLM) to improve speech emotion recognition (SER). However, the inherently high frame rate in speech modality severely limits the signal processing and understanding capabilities of SLLM. For example, a SLLM with a 4K context window can only process 80 seconds of audio at 50Hz feature sampling rate before reaching its capacity limit. Input token compression methods used in SLLM overlook the continuity and inertia of emotions across multiple conversation turns. This paper proposes a Dynamic Parameter Memory (DPM) mechanism with contextual semantics and sentence-level emotion encoding, enabling processing of unlimited-length audio with limited context windows in SLLM. Specifically, DPM progressively encodes sentence-level information and emotions into a temporary LoRA module during inference to effectively "memorize" the contextual information. We trained an emotion SLLM as a backbone and incorporated our DPM into inference for emotion recognition in conversation (ERC). Experimental results on the IEMOCAP dataset show that DPM significantly improves the emotion recognition capabilities of SLLM when processing long audio sequences, achieving state-of-the-art performance.

Yupeng Hu, Changxing Ding, Chang Sun et al.

Open vocabulary Human-Object Interaction (HOI) detection is a challenging task that detects all <human, verb, object> triplets of interest in an image, even those that are not pre-defined in the training set. Existing approaches typically rely on output features generated by large Vision-Language Models (VLMs) to enhance the generalization ability of interaction representations. However, the visual features produced by VLMs are holistic and coarse-grained, which contradicts the nature of detection tasks. To address this issue, we propose a novel Bilateral Collaboration framework for open vocabulary HOI detection (BC-HOI). This framework includes an Attention Bias Guidance (ABG) component, which guides the VLM to produce fine-grained instance-level interaction features according to the attention bias provided by the HOI detector. It also includes a Large Language Model (LLM)-based Supervision Guidance (LSG) component, which provides fine-grained token-level supervision for the HOI detector by the LLM component of the VLM. LSG enhances the ability of ABG to generate high-quality attention bias. We conduct extensive experiments on two popular benchmarks: HICO-DET and V-COCO, consistently achieving superior performance in the open vocabulary and closed settings. The code will be released in Github.

Yifeng Gu, Zicong Jiang, Jianxiu Jin et al.

Large Language Models (LLMs) have significantly advanced the field of Artificial Intelligence. However, their deployment is resource-intensive, not only due to the large number of model parameters but also because the (Key-Value) KV cache consumes a lot of memory during inference. While several works propose reducing the KV cache by evicting the unnecessary tokens, these approaches rely on accumulated attention score as eviction score to quantify the importance of the token. We identify the accumulated attention score is biased and it decreases with the position of the tokens in the mathematical expectation. As a result, the retained tokens concentrate on the initial positions, limiting model's access to global contextual information. To address this issue, we propose Adaptive holistic attention KV (AhaKV), it addresses the bias of the accumulated attention score by adaptively tuning the scale of softmax according the expectation of information entropy of attention scores. To make use of the holistic attention information in self-attention mechanism, AhaKV utilize the information of value vectors, which is overlooked in previous works, to refine the adaptive score. We show theoretically that our method is well suited for bias reduction. We deployed AhaKV on different models with a fixed cache budget. Experiments show that AhaKV successfully mitigates bias and retains crucial tokens across global context and achieve state-of-the-art results against other related work on several benchmark tasks.

Zhen Meng, Kan Chen, Xiangmin Xu et al.

Active 3D scene representation is pivotal in modern robotics applications, including remote inspection, manipulation, and telepresence. Traditional methods primarily optimize geometric fidelity or rendering accuracy, but often overlook operator-specific objectives, such as safety-critical coverage or task-driven viewpoints. This limitation leads to suboptimal viewpoint selection, particularly in constrained environments such as nuclear decommissioning. To bridge this gap, we introduce a novel framework that integrates expert operator preferences into the active 3D scene representation pipeline. Specifically, we employ Reinforcement Learning from Human Feedback (RLHF) to guide robotic path planning, reshaping the reward function based on expert input. To capture operator-specific priorities, we conduct interactive choice experiments that evaluate user preferences in 3D scene representation. We validate our framework using a UR3e robotic arm for reactor tile inspection in a nuclear decommissioning scenario. Compared to baseline methods, our approach enhances scene representation while optimizing trajectory efficiency. The RLHF-based policy consistently outperforms random selection, prioritizing task-critical details. By unifying explicit 3D geometric modeling with implicit human-in-the-loop optimization, this work establishes a foundation for adaptive, safety-critical robotic perception systems, paving the way for enhanced automation in nuclear decommissioning, remote maintenance, and other high-risk environments.

Yuehao Song, Xinggang Wang, Jingfeng Yao et al.

Gaze following aims to interpret human-scene interactions by predicting the person's focal point of gaze. Prevailing approaches often adopt a two-stage framework, whereby multi-modality information is extracted in the initial stage for gaze target prediction. Consequently, the efficacy of these methods highly depends on the precision of the preceding modality extraction. Others use a single-modality approach with complex decoders, increasing network computational load. Inspired by the remarkable success of pre-trained plain vision transformers (ViTs), we introduce a novel single-modality gaze following framework called ViTGaze. In contrast to previous methods, it creates a novel gaze following framework based mainly on powerful encoders (relative decoder parameters less than 1%). Our principal insight is that the inter-token interactions within self-attention can be transferred to interactions between humans and scenes. Leveraging this presumption, we formulate a framework consisting of a 4D interaction encoder and a 2D spatial guidance module to extract human-scene interaction information from self-attention maps. Furthermore, our investigation reveals that ViT with self-supervised pre-training has an enhanced ability to extract correlation information. Many experiments have been conducted to demonstrate the performance of the proposed method. Our method achieves state-of-the-art (SOTA) performance among all single-modality methods (3.4% improvement in the area under curve (AUC) score, 5.1% improvement in the average precision (AP)) and very comparable performance against multi-modality methods with 59% number of parameters less.

Weidong Chen, Xiaofeng Xing, Xiangmin Xu et al.

Speech emotion recognition is a challenging research topic that plays a critical role in human-computer interaction. Multimodal inputs further improve the performance as more emotional information is used. However, existing studies learn all the information in the sample while only a small portion of it is about emotion. The redundant information will become noises and limit the system performance. In this paper, a key-sparse Transformer is proposed for efficient emotion recognition by focusing more on emotion related information. The proposed method is evaluated on the IEMOCAP and LSSED. Experimental results show that the proposed method achieves better performance than the state-of-the-art approaches.

Weirui Lu, Xiaofen Xing, Bolun Cai et al.

Rank-based Learning with deep neural network has been widely used for image cropping. However, the performance of ranking-based methods is often poor and this is mainly due to two reasons: 1) image cropping is a listwise ranking task rather than pairwise comparison; 2) the rescaling caused by pooling layer and the deformation in view generation damage the performance of composition learning. In this paper, we develop a novel model to overcome these problems. To address the first problem, we formulate the image cropping as a listwise ranking problem to find the best view composition. For the second problem, a refined view sampling (called RoIRefine) is proposed to extract refined feature maps for candidate view generation. Given a series of candidate views, the proposed model learns the Top-1 probability distribution of views and picks up the best one. By integrating refined sampling and listwise ranking, the proposed network called LVRN achieves the state-of-the-art performance both in accuracy and speed.