Incremental learning model inspired in Rehearsal for deep convolutional networks

Incremental learning for detection in X-ray luggage perspective images

Convolutional neural networks have achieved remarkable results in the detection of X-ray luggage contraband. However, with an increase in contraband classes and substantial artificial transformation, the offline network training method has been unable to accurately detect the rapidly growing new classes of contraband. The current model cannot incrementally learn the newly appearing classes in real time without retraining the model. When the quantity of different types of contraband is not evenly distributed in the real-time detection process, the convolution neural network that is optimized by the gradient descent method will produce catastrophic forgetting, which means learning new knowledge and forgetting old knowledge, and the detection effect on the old classes will suddenly decline. To overcome this problem, this paper proposes an incremental learning method for online continuous learning of models and incrementally learns and detects new classes in the absence of old classes in the new classes. First, we perform parameter compression on the original network by distillation to ensure stable identification of the old classes. Second, the area proposal subnetwork and object detection subnetwork are incrementally learned to obtain the recognition ability of the new classes. In addition, this paper designs a new loss function, which causes the network to avoid catastrophic forgetting and stably detect the object of the new contraband classes. To reliably verify the model, this paper produces a multi-angle dataset for security perspective images. A total of 10 classes of contraband are tested, and the interference between two object detections is analyzed by model parameters. The experimental results show that the model can stably perform new contraband object learning even when there is an uneven distribution of data types.

Few-shot contrastive learning for image classification and its application to insulator identification

This paper presents a novel discriminative Few-shot learning architecture based on batch compact loss. Currently, Convolutional Neural Network (CNN) has achieved reasonably good performance in image recognition. Most existing CNN methods facilitate classifiers to learn discriminating patterns to identify existing categories trained with large samples. However, learning to recognize novel categories from a few examples is a challenging task. To address this, we propose the Residual Compact Network to train a deep neural network to learn hierarchical nonlinear transformations to project image pairs into the same latent feature space, under which the distance of each positive pair is reduced. To better use the commonality of class-level features for category recognition, we develop a batch compact loss to form robust feature representations relevant to a category. The proposed methods are evaluated on several datasets. Experimental evaluations show that our proposed method achieves acceptable results in Few-shot learning.

Decoding Imagined Speech Based on Deep Metric Learning for Intuitive BCI Communication

Imagined speech is a highly promising paradigm due to its intuitive application and multiclass scalability in the field of brain-computer interfaces. However, optimal feature extraction and classifiers have not yet been established. Furthermore, retraining still requires a large number of trials when new classes are added. The aim of this study is (i) to increase the classification performance for imagined speech and (ii) to apply a new class using a pretrained classifier with a small number of trials. We propose a novel framework based on deep metric learning that learns the distance by comparing the similarity between samples. We also applied the instantaneous frequency and spectral entropy used for speech signals to electroencephalography signals during imagined speech. The method was evaluated on two public datasets (6-class Coretto DB and 5-class BCI Competition DB). We achieved a 6-class accuracy of 45.00 ± 3.13% and a 5-class accuracy of 48.10 ± 3.68% using the proposed method, which significantly outperformed state-of-the-art methods. Additionally, we verified that the new class could be detected through incremental learning with a small number of trials. As a result, the average accuracy is 44.50 ± 0.26% for Coretto DB and 47.12 ± 0.27% for BCI Competition DB, which shows similar accuracy to baseline accuracy without incremental learning. Our results have shown that the accuracy can be greatly improved even with a small number of trials by selecting appropriate features from imagined speech. The proposed framework could be directly used to help construct an extensible intuitive communication system based on brain-computer interfaces.