Enhancing Few-Shot Learning in Lightweight Models via Dual-Faceted Knowledge Distillation.

In recent computer vision research, the pursuit of improved classification performance often leads to the adoption of complex, large-scale models. However, the actual deployment of such extensive models poses significant challenges in environments constrained by limited computing power and storage capacity. Consequently, this study is dedicated to addressing these challenges by focusing on innovative methods that enhance the classification performance of lightweight models. We propose a novel method to compress the knowledge learned by a large model into a lightweight one so that the latter can also achieve good performance in few-shot classification tasks. Specifically, we propose a dual-faceted knowledge distillation strategy that combines output-based and intermediate feature-based methods. The output-based method concentrates on distilling knowledge related to base class labels, while the intermediate feature-based approach, augmented by feature error distribution calibration, tackles the potential non-Gaussian nature of feature deviations, thereby boosting the effectiveness of knowledge transfer. Experiments conducted on MiniImageNet, CIFAR-FS, and CUB datasets demonstrate the superior performance of our method over state-of-the-art lightweight models, particularly in five-way one-shot and five-way five-shot tasks.

Multi-Type Missing Imputation of Time-Series Power Equipment Monitoring Data Based on Moving Average Filter-Asymmetric Denoising Autoencoder.

Supervisory control and data acquisition (SCADA) systems are widely utilized in power equipment for condition monitoring. For the collected data, there generally exists a problem-missing data of different types and patterns. This leads to the poor quality and utilization difficulties of the collected data. To address this problem, this paper customizes methodology that combines an asymmetric denoising autoencoder (ADAE) and moving average filter (MAF) to perform accurate missing data imputation. First, convolution and gated recurrent unit (GRU) are applied to the encoder of the ADAE, while the decoder still utilizes the fully connected layers to form an asymmetric network structure. The ADAE extracts the local periodic and temporal features from monitoring data and then decodes the features to realize the imputation of the multi-type missing. On this basis, according to the continuity of power data in the time domain, the MAF is utilized to fuse the prior knowledge of the neighborhood of missing data to secondarily optimize the imputed data. Case studies reveal that the developed method achieves greater accuracy compared to existing models. This paper adopts experiments under different scenarios to justify that the MAF-ADAE method applies to actual power equipment monitoring data imputation.

X 2-VLM: All-in-One Pre-Trained Model for Vision-Language Tasks.

Vision language pre-training aims to learn alignments between vision and language from a large amount of data. Most existing methods only learn image-text alignments. Some others utilize pre-trained object detectors to leverage vision language alignments at the object level. In this paper, we propose to learn multi-grained vision language alignments by a unified pre-training framework that learns multi-grained aligning and multi-grained localization simultaneously. Based on it, we present X 2-VLM, an all-in-one model with a flexible modular architecture, in which we further unify image-text pre-training and video-text pre-training in one model. X 2-VLM is able to learn unlimited visual concepts associated with diverse text descriptions. Experiment results show that X 2-VLM performs the best on base and large scale for both image-text and video-text tasks, making a good trade-off between performance and model scale. Moreover, we show that the modular design of X 2-VLM results in high transferability for it to be utilized in any language or domain. For example, by simply replacing the text encoder with XLM-R, X 2-VLM outperforms state-of-the-art multilingual multi-modal pre-trained models without any multilingual pre-training.

Research on optimization algorithms for localization and capacity determination of chargers considering the spatiotemporal distribution of electric vehicles.

The optimized layout of electric vehicle (EV) chargers is not only crucial for users' convenience but also a key element in urban sustainable development, energy transition, and the promotion of new energy vehicles. In order to provide a basis for the problem of localization and capacity determination of chargers and compare the merits of several mainstream algorithms, this paper first establishes an optimization model with the objective of minimizing the total investment cost of all the chargers and the constraint of meeting the charging demands of all electric vehicles. Optimizations were performed using genetic algorithm (GA), surrogate optimization algorithm (SOA), and mixed integer linear programming (MILP) algorithm, respectively. In the case of using MILP, the original nonlinear optimization problem was transformed into a linear problem. In the planning of city-level EV chargers, MILP took 14182.57 s to calculate the minimum cost of 34.62 million yuan. After retaining only 10% of the original data amount, SOA took 87651.34 s to calculate the minimum cost of 3.01 million yuan. The results indicate that GA is prone to falling into local optima and is not suitable for large-scale optimization problems. SOA, on the other hand, requires significant memory consumption, so the issue of memory usage needs to be carefully considered when using it directly. Although MILP is only applicable to linear programming problems, it has the advantages of lower memory usage and higher reliability if the problem can be transformed into a linear one.

Cross-Linking Methods of the Silk Protein Hydrogel in Oral and Craniomaxillofacial Tissue Regeneration.

BACKGROUND: Craniomaxillofacial tissue defects are clinical defects involving craniomaxillofacial and oral soft and hard tissues. They are characterized by defect-shaped irregularities, bacterial and inflammatory environments, and the need for functional recovery. Conventional clinical treatments are currently unable to achieve regeneration of high-quality oral craniomaxillofacial tissue. As a natural biomaterial, silk fibroin (SF) has been widely studied in biomedicine and has broad prospects for use in tissue regeneration. Hydrogels made of SF showed excellent water retention, biocompatibility, safety and the ability to combine with other materials. METHODS: To gain an in-depth understanding of the current development of SF, this article reviews the structure, preparation and application prospects in oral and craniomaxillofacial tissue regenerative medicine. It first briefly introduces the structure of SF and then summarizes the principles, advantages and disadvantages of the different cross-linking methods (physical cross-linking, chemical cross-linking and double network structure) of SF. Finally, the existing research on the use of SF in tissue engineering and the prospects of using SF with different cross-linking methods in oral and craniomaxillofacial tissue regeneration are also discussed. CONCLUSIONS: This review is intended to show the advantages of SF hydrogels in tissue engineering and provides theoretical support for establishing novel and viable silk protein hydrogels for regeneration.

Cell-free chitosan/silk fibroin/bioactive glass scaffolds with radial pore for in situ inductive regeneration of critical-size bone defects.

Tissue-engineered is an effective method for repairing critical-size bone defects. The application of bioactive scaffold provides artificial matrix and suitable microenvironment for cell recruitment and extracellular matrix deposition, which can effectively accelerate the process of tissue regeneration. Among various scaffold properties, appropriate pore structure and distribution have been proven to play a crucial role in inducing cell infiltration differentiation and in-situ tissue regeneration. In this study, a chitosan (CS) /silk fibroin (SF) /bioactive glass (BG) composite scaffold with distinctive radially oriented pore structure was constructed. The composite scaffolds had stable physical and chemical properties, a unique pore structure of radial arrangement from the center to the periphery and excellent mechanical properties. In vitro biological studies indicated that the CS/SF/BG scaffold could promote osteogenic differentiation of bone marrow mesenchymal stem cells (BMSCs) and the expression of related genes due to the wide range of connected pore structures and released active elements. Furthermore, in vivo study showed CS/SF/BG scaffold with radial pores was more conducive to the repair of skull defects in rats with accelerated healing speed during the bone tissue remodeling process. These results demonstrated the developed CS/SF/BG scaffold would be a promising therapeutic strategy for the repair of bone defects regeneration.

Photocross-linked silk fibroin/hyaluronic acid hydrogel loaded with hDPSC for pulp regeneration.

The construction of suitable biomaterials for pulp regeneration has always been a major challenge in the field of stomatology. Considering the complex and irregular anatomy of the root canal system, injectable hydrogels have received extensive attention as cell carriers in dental pulp regeneration. Here, we developed an injectable photocrosslinked methacrylylated silk fibroin (RSFMA)/methacrylylated hyaluronic acid (MeHA) composite hydrogel and characterized its physicochemical properties. The biocompatibility of encapsulated human dental pulp stem cells (hDPSCs) was subsequently investigated. With the addition of RSFMA, the pore size of the scaffolds became more regular with negligible change in porosity and exhibited excellent mechanical properties. Furthermore, the low concentration of RSFMA hydrogel in the composite hydrogel had higher cross-linking efficiency. In contrast to MeHA hydrogels, hDPSCs were encapsulated in hydrogels either in the absence or presence of high concentrations of RSFMA. The results indicated that cells in low-concentration RSFMA composite gel presented better growth ability, proliferation ability and osteogenic differentiation ability. This injectable photocrosslinked silk fibroin/hyaluronic acid hydrogel shows great potential in the field of dental pulp tissue engineering.

Experimental studies on ultralow frequency pulsed gradient magnetic field inducing apoptosis of cancer cell and inhibiting growth of cancer cell.

The morphology characteristics of cell apoptosis of the malignant tumour cells in magnetic field-treated mouse was observed for the first time. The apoptotic cancer cell contracted, became rounder and divorced from adjacent cells; the heterochromatin condensed and coagulated together along the inner side of the nuclear membrane; the endoplasmic reticulums (ER) expanded and fused with the cellular membrane; many apoptotic bodies which were packed by the cellular membrane appeared and were devoured by some lymphocytes and plasma. Apoptosis of cancer cells was detected by terminal deoxynucleotidyl transferase mediatedin situ nick end labeling (TUNEL). It was found that the number of apoptosis cancer cells of the sample treated by the magnetic field is more than that of the control sample. The growth of malignant tumour in mice was inhibited and the ability of immune cell to dissolve cancer cells was improved by ultralow frequency (ULF) pulsed gradient magnetic field; the nuclei DNA contents decreased, indicating that magnetic field can block DNA replication and inhibit mitosis of cancer cells. It was suggested that magnetic field could inhibit the metabolism of cancer cell, lower its malignancy, and restrain its rapid and heteromorphic growth. Since ULF pulsed gradient magnetic field can induce apoptosis of cancer cells and inhibit the growth of malignant tumour, it could be used as a new method to treat cancer.

Extremely low frequency (ELF) pulsed-gradient magnetic fields inhibit malignant tumour growth at different biological levels.

Extremely low frequency (ELF) pulsed-gradient magnetic field (with the maximum intensity of 0.6-2.0 T, gradient of 10-100 T.M(-1), pulse width of 20-200 ms and frequency of 0.16-1.34 Hz treatment of mice can inhibit murine malignant tumour growth, as seen from analyses at different hierarchical levels, from organism, organ, to tissue, and down to cell and macromolecules. Such magnetic fields induce apoptosis of cancer cells, and arrest neoangiogenesis, preventing a supply developing to the tumour. The growth of sarcomas might be amenable to such new method of treatment.