Object concepts play a foundational role in human visual cognition, enabling perception, memory, and interaction in the physical world. Inspired by findings in developmental psychology—where infants are shown to acquire object understanding through observation of motion—we propose a biologically inspired framework for learning object-centric visual representations in an unsupervised manner. We were inspired by the insight that motion boundary serves as a strong signal for object-level grouping, which can be used to derive pseudo-instance supervision from raw videos. Concretely, we generate motion-based instance masks using off-the-shelf optical flow and clustering algorithms, and use them to train visual encoders via contrastive learning. Our framework is fully label-free and does not rely on camera calibration, making it scalable to large-scale unstructured video data. We evaluate our approach on three downstream tasks spanning both low-level (monocular depth estimation) and high-level (3D object detection and occupancy prediction) vision. Our models outperform previous supervised and self-supervised baselines and demonstrate strong generalization to unseen scenes. These results suggest that motion-induced object representations offer a compelling alternative to existing vision foundation models, capturing a crucial but overlooked level of abstraction: the visual instance. The implementation can be found here: https://github.com/yulemao/Object_Concepts_Emerge_from_Motion

From the perspective of the user experience of media products, the study primarily focuses on the influence of users’ interaction mode on cognitive absorption. Specifically, 80 participants were recruited and randomly assigned to different interactive modes: a conversational interactive news interface and a webpage-based interactive news interface. They needed to be recorded with an EEG during the interaction and completion of the cognitive absorption (CA) scale after the interaction. The results demonstrated that there were significant differences in CA between the webpage-based interaction group and the conversational interaction group. Meanwhile, the PSD of theta and delta bands under the conversational interaction group appeared to be significantly higher. Besides, the frontal EEG asymmetry in the conversational interaction group was significantly higher than in the webpage-based interaction group. In contrast, the individual alpha frequency of the webpage-based interaction group was higher than that of the conversational interaction group. These results suggest that conversational interaction reduces the user’s reading difficulty while also reduces the cognitive load in the process of task. This helps improve user’s level of being immersed, enjoyment, sense of control, and curiosity while also improving users’ cognitive absorption as they also exhibit stronger positive emotional experiences.

In this paper, we present Auto-KD, the first automated search framework for optimal knowledge distillation design. Traditional distillation techniques typically require handcrafted designs by experts and extensive tuning costs for different teacher-student pairs. To address these issues, we empirically study different distillers, finding that they can be decomposed, combined, and simplified. Based on these observations, we build our uniform search space with advanced operations in transformations, distance functions, and hyperparameters components. For instance, the transformation parts are optional for global, intra-spatial, and inter-spatial operations, such as attention, mask, and multi-scale. Then, we introduce an effective search strategy based on the Monte Carlo tree search, modeling the search space as a Monte Carlo Tree (MCT) to capture the dependency among options. The MCT is updated using test loss and representation gap of student trained by candidate distillers as the reward for better exploration-exploitation balance. To accelerate the search process, we exploit offline processing without teacher inference, sparse training for student, and proxy settings based on distillation properties. In this way, our Auto-KD only needs small costs to search for optimal distillers before the distillation phase. Moreover, we expand Auto-KD for multi-layer and multi-teacher scenarios with training-free weighted factors. Our method is promising yet practical, and extensive experiments demonstrate that it generalizes well to different CNNs and Vision Transformer models and attains state-of-the-art performance across a range of vision tasks, including image classification, object detection, and semantic segmentation. Code is provided at https://github.com/lilujunai/Auto-KD.

Knowledge distillation (KD) has emerged as an effective technique for compressing models that can enhance the lightweight model. Conventional KD methods propose various designs to allow student model to imitate the teacher better. However, these handcrafted KD designs heavily rely on expert knowledge and may be sub-optimal for various teacher-student pairs. In this paper, we present a novel framework, KD-Zero, which utilizes evolutionary search to automatically discover promising distiller from scratch for any teacher-student architectures. Specifically, we first decompose the generalized distiller into knowledge transformations, distance functions, and loss weights. Then, we construct our distiller search space by selecting advanced operations for these three components. With sharpness and represent gap as fitting objectives, we evolve candidate populations and generate better distillers by crossover and mutation. To ensure efficient searching, we employ the loss-rejection protocol, search space shrinkage, and proxy settings during the search process. In this manner, the discovered distiller can address the capacity gap and cross-architecture challenges for any teacher-student pairs in the final distillation stage. Comprehensive experiments reveal that KD-Zero consistently outperforms other state-of-the-art methods across diverse architectures on classification, detection, and segmentation tasks. Noticeably, we provide some practical insights in designing the distiller by analyzing the distiller discovered. Codes are available in supplementary materials.

Feature distillation is a widely used training method to transfer feature information from a teacher to a student network. Current methods seek to minimize the reconstruction error of hidden feature maps between teacher-student models by explicitly optimizing distillation loss. However, some feature loss methods require complex transformations, which are not easy to optimize. In this paper, we propose a novel and effective feature distillation method, which learns to transfer knowledge by applying feature fusion as an alternative to distillation loss. Specifically, we fuse the intermediate feature of the student model to the attention teacher network, which has better representation and relatively less training cost. During training, this separable feature fusion can effectively transfer feature knowledge and is easy to optimize without complex transformation. After training, the feature fusion and the teacher network can be discarded, and the student network can be used separately in inference. Equipped with auxiliary classifier for ensemble logits distillation, our Separable Feature Knowledge Distillation (SFKD) obtains state-of-the-art performance. In experiments, SFKD achieves 4% performance improvement on CIFAR-100 and 2% on ImageNet for ResNet models, which substantially outperforms other feature distillation methods.

Feature distillation always leads to significant performance improvements, but requires extra training budgets. To address the problem, we propose TFD, a simple and effective Teacher-Free Distillation framework, which seeks to reuse the privileged features within the student network itself. Specifically, TFD squeezes feature knowledge in the deeper layers into the shallow ones by minimizing feature loss. Thanks to the narrow gap of these self-features, TFD only needs to adopt a simple l2loss without complex transformations. Extensive experiments on recognition benchmarks show that our framework can achieve superior performance than teacher-based feature distillation methods. On the ImageNet dataset, our approach achieves 0.8% gains for ResNet18, which surpasses other state-of-the-art training techniques.

MOTIVATION: Quality of gene expression analyses using de novo assembled transcripts in species that experienced recent polyploidization remains unexplored. RESULTS: Differential gene expression (DGE) analyses using putative genes inferred by Trinity, Corset and Grouper performed slightly differently across five plant species that experienced various polyploidy histories. In species that lack recent polyploidy events that occurred in the past several millions of years, DGE analyses using de novo assembled transcriptomes identified 54-82% of the differentially expressed genes recovered by mapping reads to the reference genes. However, in species that experienced more recent polyploidy events, the percentage decreased to 21-65%. Gene co-expression network analyses using de novo assemblies versus mapping to the reference genes recovered the same module that significantly correlated with treatment in one species that lacks recent polyploidization. AVAILABILITY AND IMPLEMENTATION: Commands and scripts used in this study are available at https://bitbucket.org/lychen83/chen_et_al_2018_benchmark_dge/; Analysis files are available at Dryad doi: 10.5061/dryad.4p6n481. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

With the widespread use of smart appliances in the residential area, the load side of the microgrid is becoming more and more flexible. As we all know, the renewable energy generation is usually uncontrollable, so how to dispatch the controllable generation devices, storage system and flexible loads to guarantee the safe and economical operation of the microgrid is a very important problem to be solved. In this paper, a two-phase optimal strategy is proposed to deal with the optimal problem. In the first phase, the flexible loads are optimized to maximize the power consumption provided by renewable energy generation system. Then the day-ahead economic optimal dispatch problem is established to minimize the operating costs and environmental costs based on the optimized loads demand, and the Quantum behaved particle swarm optimization (QPSO) algorithm is adopted to solve it. Finally, a typical grid-connecting microgrid is taken as the study case, and the simulation results show that the reasonable use of the flexible loads can not only play the role of peak shifting and valley filling in electricity, but also increase the utilization rate of renewable energy, thereby reducing operating and environmental governance costs.

Modern Brain-computer interface (BCI) technique is essentially based on the classification of the brain signals. The sparse representation classification (SRC) method has been studied for classifying EEG signals of the motor imagery based BCI. The dictionary used in the SRC method is the simple combination of feature vectors which are extracted from the EEG signal-trials by common spatial pattern (CSP) algorithm. In this paper, we propose a method to learn a new dictionary with smaller size and more discriminative ability for the classification. The proposed method, discriminative dictionary learning (DDL), is based on minimizing an objective function containing a reconstructive term and a discriminative term. We apply an iterative scheme to the optimization and transform it to a series of mixed ℓ1-ℓ2 optimizations, which are solved based on separable surrogate functions (SSF) technique. We evaluate the proposed method using the dataset from BCI competition III. The experimental results show that the proposed method outperforms the SRC method.