Generative adversarial networks (GANs) have achieved remarkable progress in image generation tasks, yet issues such as mode collapse and limited sample diversity constrain their broader applicability. This paper proposes a novel multi-objective conditional GAN that integrates residual attention mechanisms and topological constraints to address these challenges. The model employs an evolutionary optimization strategy based on two objective functions to encourage high image fidelity and mode diversity: a quality fitness score based on discriminator output, and a mode diversity score measuring intra-class variability. To enhance conditional generation, a novel generator architecture is designed by incorporating a conditional residual block that embeds label information via learnable feature-wise modulation. In addition, self-attention and channel attention mechanisms are employed to capture global structure and highlight class-relevant features. The discriminator adopts a symmetric architecture with attention-enhanced residual blocks and spectral normalization to ensure semantic consistency and stable adversarial training. Furthermore, a topological loss function based on persistent homology is introduced to enforce structural consistency by regulating the number of connected components and loops in the generated samples. Experimental results on MNIST, CIFAR-10, and a molecular dataset demonstrate that the proposed model achieves superior image fidelity and diversity compared to existing methods, effectively mitigating mode collapse while preserving class-specific characteristics.
For violence detection on event streams in privacy-constrained settings, many methods increase model complexity to gain accuracy, which conflicts with edge efficiency and interpretability. This work revisits time-series anomaly detection and introduces a two-stage system named TEV (Trees for Event-stream Violence Detection). The event-stream analysis is formulated as a weight-allocation problem over a long sequence defined by sliding windows. In Stage 1, the event-camera data are processed by a pretrained feature extractor. A dual-branch architecture conducts initial weight allocation, which is further refined using a graph self-attention network with a concatenation strategy. Center loss and video-level label loss are employed to produce both binary and multi-class predictions. In Stage 2, a tree-structured allocator is built on top of Stage 1 outputs. Unlike standard decision trees, the split function fuses feature and topological information and introduces a pointer mechanism that allows split nodes to reuse information computed in earlier splits, enabling task-adaptive behavior. Consequently, at inference time, the system uses the frozen feature extractor, compact graph readout/global head, and lightweight tree branch rather than a heavy video backbone. Experiments on multiple benchmark datasets show that the proposed TEV achieves higher performance than prior methods while remaining lightweight. Beyond video-level recognition, the evaluation includes frame-/segment-level AUC, AP, false-alarm analysis, component ablations, Top-$K$ and fusion-weight sensitivity, and on-device latency, memory, and power measurements on an NVIDIA Jetson Xavier AGX. These results show that the graph-to-tree migration improves localization, interpretability, and edge deployability without relying on a heavy video backbone.
Mars hyperspectral imaging enables mineralogical mapping by exploiting diagnostic absorption characteristics in high-dimensional reflectance spectra. However, compared with Earth HSI classification, which often benefits from richer land-cover annotations and stronger spatial or material semantics, few-shot Martian mineral classification is more constrained by scarce expert labels, high inter-class spectral similarity, and weak local absorption differences among minerals. Existing hyperspectral masked autoencoders mainly learn generic spectral–spatial continuity, but uniform reconstruction may underemphasize mineral-discriminative absorption structures in CRISM spectra. To address this issue, we propose an absorption-aware masked autoencoder for few-shot Mars hyperspectral classification. The framework preserves raw spectra through a raw reconstruction branch, emphasizes local absorption residuals through a continuum-guided branch, and introduces a derivative-aware constraint to retain spectral slope and absorption-edge variations. All reconstruction targets are derived from unlabeled Mars hyperspectral observations, so the pre-training remains self-supervised without using mineral labels. The learned encoder is then fine-tuned with only a few labeled pixels per class. Experiments on three HyMars scenes show that the proposed method improves accuracy and stability over supervised CNN/Transformer baselines, Mars-oriented methods, and generic hyperspectral masked modeling approaches, demonstrating the effectiveness of absorption-aware self-supervised pre-training for highly similar Martian mineral classes.
Data-driven optimal control approaches have been significantly ameliorated by adaptive dynamic programming. In this paper, a novel evolutionary Q-learning (EQL) algorithm, coupled with an advanced neural network (NN), is developed to address the optimal tracking issue for nonlinear discrete-time systems. In conventional Q-learning algorithms, policy improvement is constructed by the gradient descent method. To effectively tackle situations where gradient information is inaccessible, a swarm optimizer leveraging evolutionary computation is provided to explore and identify improved policies during the policy optimization process. Besides, a model network is generally constructed to solve the feedforward control in optimal tracking issues. To further enhance tracking performance, the advanced NN based on the brain storm optimization method is introduced to accurately approximate the reference dynamics. Moreover, the monotonicity and convergence properties of EQL are rigorously established. Finally, a nonlinear simulation is provided to validate the performance of the established data-driven control algorithm.
Learning discriminative and informative pixel-level feature representations in a self-supervised setting is crucial for downstream computer vision applications. The introduction of the batch-contrastive approach within a self-supervised paradigm sparked a new trend in learning highly discriminative feature representations. For pixel-level batch-contrastive training, existing object-centric approaches use mid-level visual priors, such as salience masks, to sample positive and negative pixel pairs. However, the mid-level visual priors contain object-level information but lack class-level information. Meanwhile, class-level information can provide a more precise boundary for grouping pixels for contrastive training. Also, recent advances suggest that pixels from different objects within the same visual class exhibit higher intra-class variance than those from the same object. Therefore, we propose a novel self-supervised method, OPCL, to overcome the limitations of the traditional batch-contrastive approach for object-centric Unsupervised Semantic Segmentation. The OPCL introduces a self-labeling step to generate high-level visual priors for training images, incorporating object- and class-level information. Also, it employs a novel pseudo-supervised loss that jointly leverages object- and class-level information within a contrastive learning framework to learn discriminative pixel embeddings. The proposed method surpasses the baseline pixel-level contrastive approach for object-centric semantic segmentation in an unsupervised setting, achieving a relative improvement of 17.14% in mIoU over the PASCAL VOC dataset. It also surpasses existing methods for auxiliary tasks such as Semantic Segmentation Retrieval, and Unsupervised Saliency Detection across multiple datasets, including CUB, DUTS, and ECSSD.
