We propose unsupervised multi-scenario (UMS) person re-identification (ReID) as a new task that expands ReID across diverse scenarios (cross-resolution, clothing change, etc.) within a single coherent framework. To tackle UMS-ReID, we introduce image-text knowledge modeling (ITKM) -- a three-stage framework that effectively exploits the representational power of vision-language models. We start with a pre-trained CLIP model with an image encoder and a text encoder. In Stage I, we introduce a scenario embedding in the image encoder and fine-tune the encoder to adaptively leverage knowledge from multiple scenarios. In Stage II, we optimize a set of learned text embeddings to associate with pseudo-labels from Stage I and introduce a multi-scenario separation loss to increase the divergence between inter-scenario text representations. In Stage III, we first introduce cluster-level and instance-level heterogeneous matching modules to obtain reliable heterogeneous positive pairs (e.g., a visible image and an infrared image of the same person) within each scenario. Next, we propose a dynamic text representation update strategy to maintain consistency between text and image supervision signals. Experimental results across multiple scenarios demonstrate the superiority and generalizability of ITKM; it not only outperforms existing scenario-specific methods but also enhances overall performance by integrating knowledge from multiple scenarios.

The de novo generation of molecules with desirable properties is a critical challenge, where diffusion models are computationally intensive and autoregressive models struggle with error propagation. In this work, we introduce the Graph VQ-Transformer (GVT), a two-stage generative framework that achieves both high accuracy and efficiency. The core of our approach is a novel Graph Vector Quantized Variational Autoencoder (VQ-VAE) that compresses molecular graphs into high-fidelity discrete latent sequences. By synergistically combining a Graph Transformer with canonical Reverse Cuthill-McKee (RCM) node ordering and Rotary Positional Embeddings (RoPE), our VQ-VAE achieves near-perfect reconstruction rates. An autoregressive Transformer is then trained on these discrete latents, effectively converting graph generation into a well-structured sequence modeling problem. Crucially, this mapping of complex graphs to high-fidelity discrete sequences bridges molecular design with the powerful paradigm of large-scale sequence modeling, unlocking potential synergies with Large Language Models (LLMs). Extensive experiments show that GVT achieves state-of-the-art or highly competitive performance across major benchmarks like ZINC250k, MOSES, and GuacaMol, and notably outperforms leading diffusion models on key distribution similarity metrics such as FCD and KL Divergence. With its superior performance, efficiency, and architectural novelty, GVT not only presents a compelling alternative to diffusion models but also establishes a strong new baseline for the field, paving the way for future research in discrete latent-space molecular generation.

Accurate prediction of changes in protein-protein binding affinity influenced by mutations (i.e.,$\Delta\Delta G)$is essential for understanding the structure and function of proteins and elucidating the underlying mechanisms of complex diseases. We introduce a novel multi-scale feature fusion network for protein-protein$\Delta\Delta G$prediction, aiming to reduce the dependency of previous methods on intricate biological features and expert-driven knowledge, and demonstrate its effectiveness in the challenging and important domain of protein data. Specifically, we employ multi-scale modeling of protein complexes, and introduce self-attention mechanisms and various extraction modules to comprehensively capture the features. The proposed method effectively learns the complete biological regulation of complexes and incorporates interactions between amino acids. Extensive experiments on three benchmark datasets demonstrate that our proposed framework significantly outperforms the state-of-the-art methods for$\Delta\Delta G$prediction while also providing excellent performance in the domain of membrane protein design.

Computational methods for predicting drug-target binding affinity (DTA) are critical for large-scale screening of prospective therapeutic compounds during drug discovery. Deep neural networks (DNNs) have recently shown significant promise for DTA prediction. By leveraging available data for training, DNNs can expand the use of DTA prediction to situations where only sequence information is available for potential drug molecules and their targets, and there is no prior knowledge regarding the molecular geometric conformations. We propose DHAG-DTA, a general dynamic hierarchical affinity graph DNN approach, for DTA prediction using molecular sequence information and already known drug-target interactions. DHAG-DTA introduces a two-level hierarchical graph structure: at the upper level, interactions between drug and target molecules are represented via an affinity graph and at the lower level, embedded molecular graphs represent interactions within the individual molecules. This allows for integration of information from both inter and intra molecular interactions for DTA prediction, which has also been addressed in other recent independent work. The fundamental innovations introduced by DHAG-DTA include: (a) a single overall hierarchical graph that allows better assimilation of information during the learning process compared with loosely-coupled individual graphs, (b) dynamic determination of the affinity graph structure via the introduction of unlabeled edges and a maximum entropy criterion for active edge selection, (c) skip connections in the DNN for fusing intra and inter molecular information, and (d) fusion of both model-based and similarity-based feature embeddings to get robust embeddings of unseen molecules. Experimental results on two common benchmark datasets demonstrate that DHAG-DTA outperforms other existing models on multiple evaluation metrics, achieving state-of-the-art performance.

Clothing change person re-identification (CC-ReID) is a crucial task in intelligent surveillance, aiming to match images of the same person wearing different clothing. Promising performance in existing CC-ReID methods is achieved at the cost of labor-intensive manual annotation of identity labels. While some researchers have explored unsupervised CC-ReID, these methods still depend on additional deep learning models for preprocessing. To eliminate the need for additional models and improve performance, we propose a joint augmentation and part learning (JAPL) framework that obtains clothing change positive pairs in an unsupervised fashion by synergistically combining augmentation-based invariant learning (AugIL) and part-based invariant learning (ParIL). AugIL first constructs clothing change pseudo-positive pairs and then encourages the model to focus on clothing-invariant information by enhancing feature consistency between the pseudo-positive pairs. ParIL beneficially encourages high similarity between inter-cluster clothing change positive pair using part images and a prediction sharpening loss. PartIL also introduces a soft consistency loss that promotes clothing-invariant feature learning by encouraging consistency of class vectors between the real features actually used for CC-ReID and the part features. Experimental results on multiple ReID datasets demonstrate that the proposed JAPL not only surpasses existing unsupervised methods but also achieves competitive performance compared to some supervised CC-ReID methods.

Cross-resolution person re-identification (CR-ReID) aims to match low-resolution (LR) and high-resolution (HR) images of the same individual. To reduce the cost of manual annotation, existing unsupervised CR-ReID methods typically rely on cross-resolution fusion to obtain pseudo-labels and resolution-invariant features. However, the fusion process requires two encoders and a fusion module, which significantly increases computational complexity and reduces efficiency. To address this issue, we propose a robust labeling and invariance modeling (RLIM) framework, which utilizes a single encoder to tackle the unsupervised CR-ReID problem. To obtain pseudo-labels robust to resolution gaps, we develop cross-resolution robust labeling (CRL), which utilizes two clustering criteria to encourage cross-resolution positive pairs to cluster together and exploit the reliable relationships between images. We also introduce random texture augmentation (TexA) to enhance the model's robustness to noisy textures related to artifacts and backgrounds by randomly adjusting texture strength. During the optimization process, we introduce the resolution-cluster consistency loss, which promotes resolution-invariant feature learning by aligning inter-resolution distances with intra-cluster distances. Experimental results on multiple datasets demonstrate that RLIM not only surpasses existing unsupervised methods, but also achieves performance close to some supervised CR-ReID methods. Code is available at https://github.com/zqpang/RLIM.

Policy constraint methods in offline reinforcement learning employ additional regularization techniques to constrain the discrepancy between the learned policy and the offline dataset. However, these methods tend to result in overly conservative policies that resemble the behavior policy, thus limiting their performance. We investigate this limitation and attribute it to the static nature of traditional constraints. In this paper, we propose a novel dynamic policy constraint that restricts the learned policy on the samples generated by the exponential moving average of previously learned policies. By integrating this self-constraint mechanism into off-policy methods, our method facilitates the learning of non-conservative policies while avoiding policy collapse in the offline setting. Theoretical results show that our approach results in a nearly monotonically improved reference policy. Extensive experiments on the D4RL MuJoCo domain demonstrate that our proposed method achieves state-of-the-art performance among the policy constraint methods.

BACKGROUND: Molecular biology is crucial for drug discovery, protein design, and human health. Due to the vastness of the drug-like chemical space, depending on biomedical experts to manually design molecules is exceedingly expensive. Utilizing generative methods with deep learning technology offers an effective approach to streamline the search space for molecular design and save costs. This paper introduces a novel E(3)-equivariant score-based diffusion framework for 3D molecular generation via SDEs, aiming to address the constraints of unified Gaussian diffusion methods. Within the proposed framework EMDS, the complete diffusion is decomposed into separate diffusion processes for distinct components of the molecular feature space, while the modeling processes also capture the complex dependency among these components. Moreover, angle and torsion angle information is integrated into the networks to enhance the modeling of atom coordinates and utilize spatial information more effectively. RESULTS: Experiments on the widely utilized QM9 dataset demonstrate that our proposed framework significantly outperforms the state-of-the-art methods in all evaluation metrics for 3D molecular generation. Additionally, ablation experiments are conducted to highlight the contribution of key components in our framework, demonstrating the effectiveness of the proposed framework and the performance improvements of incorporating angle and torsion angle information for molecular generation. Finally, the comparative results of distribution show that our method is highly effective in generating molecules that closely resemble the actual scenario. CONCLUSION: Through the experiments and comparative results, our framework clearly outperforms previous 3D molecular generation methods, exhibiting significantly better capacity for modeling chemically realistic molecules. The excellent performance of EMDS in 3D molecular generation brings novel and encouraging opportunities for tackling challenging biomedical molecule and protein scenarios.

Visible-infrared person re-identification (VI-ReID) is a challenging cross-modality image retrieval task. Compared to visible modality person re-identification that handles only the intra-modality discrepancy, VI-ReID suffers from an additional modality gap. Most existing VI-ReID methods achieve promising accuracy in a supervised setting, but the high annotation cost limits their scalability to real-world scenarios. Although a few unsupervised VI-ReID methods already exist, they typically rely on intra-modality initialization and cross-modality instance selection, despite the additional computational time required for intra-modality initialization. In this paper, we study the fully unsupervised VI-ReID problem and propose a novel cross-modality hierarchical clustering and refinement (CHCR) method by promoting modality-invariant feature learning and improving the reliability of pseudo-labels. Unlike conventional VI-ReID methods, CHCR does not rely on any manual identity annotation and intra-modality initialization. First, we design a simple and effective cross-modality clustering baseline that clusters between modalities. Then, to provide sufficient inter-modality positive sample pairs for modality-invariant feature learning, we propose a cross-modality hierarchical clustering algorithm to promote the clustering of inter-modality positive samples into the same cluster. In addition, we develop an inter-channel pseudo-label refinement algorithm to eliminate unreliable pseudo-labels by checking the clustering results of three channels in the visible modality. Extensive experiments demonstrate that CHCR outperforms state-of-the-art unsupervised methods and achieves performance competitive with many supervised methods.

RGB (visible), near-infrared (NI), and thermal infrared (TI) imaging modalities are commonly combined for round-the-clock surveillance. We introduce a novel unsupervised multi-modality person re-identification (MM-ReID) task, which, based on an individual’s image in any one modality, seeks to identify matches in the other two modalities. Compared to prior MM-ReID problem formulations, unsupervised MM-ReID significantly reduces labeling cost and imaging constraints. To address the unsupervised MM-ReID task, we propose a novel inter-modality similarity learning (IMSL) framework consisting of four synergistic interconnected modules: modality mean clustering (MMC), multi-modality reliability estimation (MMRE), shape-based mutual reinforcement (SMR), and modality-aware invariant learning (MIL). MMC iterates with SMR and MIL in a mutually beneficial manner to provide pseudo-labels that are robust to modality gap. MMRE normalizes sample weights, mitigating the impact of noisy labels in the multi-modality setting. SMR emphasizes shape information to implicitly enhance the model’s robustness to the modality gap and is additionally guided by pseudo-labels provided by MMC to attend to identity-related details. MIL explicitly encourages learning of modality-invariant and identity-related features via contrastive feedback for the MMC module. Extensive experimental results on the multi-modality and cross-modality datasets demonstrate that IMSL provides substantial performance gains over existing methods. Code is made available at https://github.com/zqpang/IMSL.

We introduce a novel graph-based framework for alleviating key challenges in distantly-supervised relation extraction and demonstrate its effectiveness in the challenging and important domain of biomedical data. Specifically, we propose a graph view of sentence bags referring to an entity pair, which enables message-passing based aggregation of information related to the entity pair over the sentence bag. The proposed framework alleviates the common problem of noisy labeling in distantly supervised relation extraction and also effectively incorporates inter-dependencies between sentences within a bag. Extensive experiments on two large-scale biomedical relation datasets and the widely utilized NYT dataset demonstrate that our proposed framework significantly outperforms the state-of-the-art methods for biomedical distant supervision relation extraction while also providing excellent performance for relation extraction in the general text mining domain.

Label noise and long-tailed distributions are two major challenges in distantly supervised relation extraction. Recent studies have shown great progress on denoising, but paid little attention to the problem of long-tailed relations. In this paper, we introduce a constraint graph to model the dependencies between relation labels. On top of that, we further propose a novel constraint graph-based relation extraction framework(CGRE) to handle the two challenges simultaneously. CGRE employs graph convolution networks to propagate information from data-rich relation nodes to data-poor relation nodes, and thus boosts the representation learning of long-tailed relations. To further improve the noise immunity, a constraint-aware attention module is designed in CGRE to integrate the constraint information. Extensive experimental results indicate that CGRE achieves significant improvements over the previous methods for both denoising and long-tailed relation extraction.

In disease research, the study of gene-disease correlation has always been an important topic. With the emergence of large-scale connected data sets in biology, we use known correlations between the entities, which may be from different sets, to build a biological heterogeneous network and propose a new network embedded representation algorithm to calculate the correlation between disease and genes, using the correlation score to predict pathogenic genes. Then, we conduct several experiments to compare our method to other state-of-the-art methods. The results reveal that our method achieves better performance than the traditional methods.

MicroRNAs (miRNAs) regulate genes that are associated with various diseases. To better understand miRNAs, the miRNA regulatory mechanism needs to be investigated and the real targets identified. Here, we present miRTDL, a new miRNA target prediction algorithm based on convolutional neural network (CNN). The CNN automatically extracts essential information from the input data rather than completely relying on the input dataset generated artificially when the precise miRNA target mechanisms are poorly known. In this work, the constraint relaxing method is first used to construct a balanced training dataset to avoid inaccurate predictions caused by the existing unbalanced dataset. The miRTDL is then applied to 1,606 experimentally validated miRNA target pairs. Finally, the results show that our miRTDL outperforms the existing target prediction algorithms and achieves significantly higher sensitivity, specificity and accuracy of 88.43, 96.44, and 89.98 percent, respectively. We also investigate the miRNA target mechanism, and the results show that the complementation features are more important than the others.

As a representative deep learning model, Convolutional Neural Networks (CNNs) can provide good features to represent the objects in image, and has made a great achievement in image classification and object detection. However, CNNs requires resizing the input images to a fixed size, which may affect the performance of the model due to information loss and distortion. To overcome the limitation, we replace the last pooling layer with topic model-LDA (Latent Dirichlet Allocation) to get a fixed-size output without resizing the input images, and we call it Topic Network. With Topic Network, the input images can be images of an arbitrary size and ratio without resizing, but the output is a k-dimension vector which represents the distribution of topics in image (k is the number of topics). Topic Network performs well in image classification task on Caltech101 and VOC2007 datasets.

MicroRNAs (miRNAs) are small non-coding RNAs which cause target genes degradation or translational inhibition. Constructing functional miRNAs regulatory module can be a significant step towards the discovery of their regulatory roles in various development programs. In this paper, we present a Correlated Correspondence Regulatory Module model which builds on modified Correlated Topic Model (CTM). We apply the proposed method to the expression profiles of miRNAs and genes on 89 human cancer samples. The approach computationally predicts miRNA-gene interactions according to the negative or positive correlation relationship between miRNA and gene expression data and identifies functional miRNA regulatory modules from which we can infer multiple and dynamic miRNA function according to the known elements, the result shows consistency with published literature and database. Furthermore, a miRNA regulatory network is constructed in order to study the associations among various regulatory modules, we can detect evolution of miRNA function in biological process according these associations, which solve restriction of traditional methods that only focus on static miRNA function in single regulatory module. Online services can be accessed at the website (http://nclab.hit.edu.cn/CCRM).

Lung nodule detection in thin section computerized tomography (CT) images is a useful but challenging task in the development of computer aided diagnosis (CAD) system for lung cancer. In order to improve sensitivity and reduce false positive, we consider a 3D nodule as a 2D region of interest (ROI) sequence and utilize a discriminative sequence model called hidden conditional random field to capture the correlations and transitions of a nodule's ROIs on several consecutive slices. First, we use region growing and thresholding to segment lung parenchyma. Second, selective enhancement filter is employed on 2D images to get 2D ROIs and after that, we match these ROIs on consecutive images based on a simple but effective criteria to get 2D ROI sequence(3D candidate) of a nodule. Third, given these ROI sequences, hidden conditional random field is devised to classify whether some 3D candidates are nodules or not based on these sequences. The proposed system is validated on 24 patients' scans which contain 59 nodules in total from Lung Image Database Consortium (LIDC) dataset. Experimental results demonstrate that our approach achieves high sensitivity and reduces false positive significantly.

MOTIVATION: The rapid accumulation of microRNAs (miRNAs) and experimental evidence for miRNA interactions has ushered in a new area of miRNA research that focuses on network more than individual miRNA interaction, which provides a systematic view of the whole microRNome. So it is a challenge to infer miRNA functional interactions on a system-wide level and further draw a miRNA functional network (miRFN). A few studies have focused on the well-studied human species; however, these methods can neither be extended to other non-model organisms nor take fully into account the information embedded in miRNA-target and target-target interactions. Thus, it is important to develop appropriate methods for inferring the miRNA network of non-model species, such as soybean (Glycine max), without such extensive miRNA-phenotype associated data as miRNA-disease associations in human. RESULTS: Here we propose a new method to measure the functional similarity of miRNAs considering both the site accessibility and the interactive context of target genes in functional gene networks. We further construct the miRFNs of soybean, which is the first study on soybean miRNAs on the network level and the core methods can be easily extended to other species. We found that miRFNs of soybean exhibit a scale-free, small world and modular architecture, with their degrees fit best to power-law and exponential distribution. We also showed that miRNA with high degree tends to interact with those of low degree, which reveals the disassortativity and modularity of miRFNs. Our efforts in this study will be useful to further reveal the soybean miRNA-miRNA and miRNA-gene interactive mechanism on a systematic level. AVAILABILITY AND IMPLEMENTATION: A web tool for information retrieval and analysis of soybean miRFNs and the relevant target functional gene networks can be accessed at SoymiRNet: http://nclab.hit.edu.cn/SoymiRNet.

The identification of communities is an important step in understanding of the complex network. Comparative studies suggest that the development of accurate and efficient methods to infer the communities is still in its early stages. Label propagation algorithm (LPA) that detects communities by propagating labels among vertices, attracts a great deal of attention recently. However, the communities detected by most LPAs are disjointed. Due to communities are often overlapping in real world networks, we show a multi-label propagation algorithm (MLPA) to detect overlapping communities. The inspiration is that the more people are familiar, the more they trust each other. To simulate the confidence of human communication, propagating intensity (PI) is defined to describe the confidence extent of the label propagated by neighboring vertices. The PI is then used to guide the propagation, with the purpose to make the detection more accurate. The results of extensive experiments both on synthetic and real networks show that the proposed MLPA outperforms many other methods. The effectiveness of MLPA can be attributed to its multi-label propagating strategy.

This paper addresses the problem of building up effective training sets at minimal labeling cost for object detection. This problem occurs in the situation that the part-based detector is trained on a group of positive examples with bounding box labels, but the images selected by uniform sampling do not reflect the desired training distribution and need additional labeling cost in order to obtain enough positive examples. We study the active training process in which some object windows are sampled from a pool of unlabeled candidate windows, and then their corresponding bounding annotations are queried. We derive an effective training set by selecting a group of most uncertain object windows according to the current detector. Our approach has been empirically demonstrated on the object detection task of PASCAL VOC dataset. The experiment results show that our proposed algorithm outperforms common uniform sampling within the same labeling cost.

MOTIVATION: The evolutionary history of species is traditionally represented with a rooted phylogenetic tree. Each tree comprises a set of clusters, i.e. subsets of the species that are descended from a common ancestor. When rooted phylogenetic trees are built from several different datasets (e.g. from different genes), the clusters are often conflicting. These conflicting clusters cannot be expressed as a simple phylogenetic tree; however, they can be expressed in a phylogenetic network. Phylogenetic networks are a generalization of phylogenetic trees that can account for processes such as hybridization, horizontal gene transfer and recombination, which are difficult to represent in standard tree-like models of evolutionary histories. There is currently a large body of research aimed at developing appropriate methods for constructing phylogenetic networks from cluster sets. The Cass algorithm can construct a much simpler network than other available methods, but is extremely slow for large datasets or for datasets that need lots of reticulate nodes. The networks constructed by Cass are also greatly dependent on the order of input data, i.e. it generally derives different phylogenetic networks for the same dataset when different input orders are used. RESULTS: In this study, we introduce an improved Cass algorithm, Lnetwork, which can construct a phylogenetic network for a given set of clusters. We show that Lnetwork is significantly faster than Cass and effectively weakens the influence of input data order. Moreover, we show that Lnetwork can construct a much simpler network than most of the other available methods. AVAILABILITY: Lnetwork has been built as a Java software package and is freely available at http://nclab.hit.edu.cn/∼wangjuan/Lnetwork/. CONTACT: [email protected] SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

In this paper, we present a system for exhibiting a Chinese landscape painting about 900 years old. There are three parts in our system: (1) we allocate a voice dubbing or background music, which is treated as a point sound source, onto the 2D painting and obtain its position in the 2D space. All of the audio data are then located in a 3D hidden space, by projecting their 2D positions to the 3D space through a projection model. (2) A two-layer directed graph structure is proposed to well organize the audio data in a 4D space (with 1D temporal and 3D spatial). (3) The exhibition is defined as an active exploration in a viewpoint space, which faces both the image and the 3D world where the sound sources reside. The 3D space and the two-layer graph structure generate a natural and meaningful stereo audio field. Meanwhile, compared to videos with guided walk through, the active exploration makes the exhibition more attractive.

BACKGROUND: Inference of evolutionary trees using the maximum likelihood principle is NP-hard. Therefore, all practical methods rely on heuristics. The topological transformations often used in heuristics are Nearest Neighbor Interchange (NNI), Subtree Prune and Regraft (SPR) and Tree Bisection and Reconnection (TBR). However, these topological transformations often fall easily into local optima, since there are not many trees accessible in one step from any given tree. Another more exhaustive topological transformation is p-Edge Contraction and Refinement (p-ECR). However, due to its high computation complexity, p-ECR has rarely been used in practice. RESULTS: To make the p-ECR move more efficient, this paper proposes a new method named p-ECRNJ. The main idea of p-ECRNJ is to use neighbor joining (NJ) to refine the unresolved nodes produced in p-ECR. CONCLUSION: Experiments with real datasets show that p-ECRNJ can find better trees than the best known maximum likelihood methods so far and can efficiently improve local topological transforms in reasonable time.

Playfield detection is a key step in sports video content analysis, since many semantic clues could be inferred from it. In this paper we propose an adaptive GMM based algorithm for playfield detection. Its advantages are twofold. First, it can update model parameters by the incremental expectation maximization (IEM) algorithm, which enables the model to adapt to the playfield variation with time; Second, online training is performed, which saves buffer for training samples. Then, the playfield detection results are applied in recognizing the key zone of the current playfield in soccer video, in which a fast algorithm based on playfield contour and least square is proposed. Experimental results show that the proposed algorithms are encouraging.

Particle filter is a powerful visual tracking tool based on sequential Monte Carlo framework, and it needs large numbers of samples to properly approximate the posterior density of the state evolution. However, its efficiency degenerates if too many samples are applied. In this paper, an improved particle filter is proposed by integrating support vector regression into sequential Monte Carlo framework to enhance the performance of particle filter with small sample set. The proposed particle filter utilizes an SVR based re-weighting scheme to re-approximate the posterior density and avoid sample impoverishment. Firstly, a regression function is obtained by support vector regression method over the weighted sample set. Then, each sample is re-weighted via the regression function. Finally, ameliorative posterior density of the state is re-approximated to maintain the effectiveness and diversity of samples. Experimental results demonstrate that the proposed particle filter improves the efficiency of tracking system effectively and outperforms classical particle filter.

In this demonstration, a prototype system for generating 3D cartoon from broadcast soccer video is proposed. This system takes advantage of computer vision (CV) and computer graphics (CG) techniques to provide users new experience that can not be obtained from original video. Firstly, it uses CV techniques to obtain 3D positions of the players and ball. Then, CG techniques are applied to model the playfield, players, and ball. Finally, 3D cartoon is generated. Our system allows users to watch the game at any point of view using a 3D viewer based on OpenGL.

This paper presents a novel shot segmentation algorithm on the H.264/AVC video, which operates in the compressed domain. First, the algorithm exploits the intra prediction mode histogram to locate those potential GOPs, where shot transitions occur with great probability. Secondly, to further find shot boundaries at the frame level, we count the number of macroblocks with different inter prediction modes as the features and exploit HMMs to automatically model different cases in which shot transitions can occur among I, P and B frames. Since H.264/AVC provides more motion compensation modes, using HMMs can avoid the tediousness of manually tuning multiple thresholds simultaneously. The experimental results show that the algorithm is efficient and robust and it can not only locate cuts, but also work for gradual shot transitions.

The balance between exploration and exploitation is one of the key problems of action selection in Q-learning. Pure exploitation causes the agent to reach the locally optimal policies quickly, whereas excessive exploration degrades the performance of the Q-learning algorithm even if it may accelerate the learning process and allow avoiding the locally optimal policies. In this paper, finding the optimum policy in Q-learning is described as search for the optimum solution in combinatorial optimization. The Metropolis criterion of simulated annealing algorithm is introduced in order to balance exploration and exploitation of Q-learning, and the modified Q-learning algorithm based on this criterion, SA-Q-learning, is presented. Experiments show that SA-Q-learning converges more quickly than Q-learning or Boltzmann exploration, and that the search does not suffer of performance degradation due to excessive exploration.