Abstract
Automatic recognition of visual objects using a deep learning approach has been successfully applied to multiple areas. However, deep learning techniques require a large amount of labeled data, which is usually expensive to obtain. An alternative is to use semi-supervised models, such as co-training, where multiple complementary views are combined using a small amount of labeled data. A simple way to associate views to visual objects is through the application of a degree of rotation or a type of filter. In this work, we propose a co-training model for visual object recognition using deep neural networks by adding layers of self-supervised neural networks as intermediate inputs to the views, where the views are diversified through the cross-entropy regularization of their outputs. Since the model merges the concepts of co-training and self-supervised learning by considering the differentiation of outputs, we called it Differential Self-Supervised Co-Training (DSSCo-Training). This paper presents some experiments using the DSSCo-Training model to well-known image datasets such as MNIST, CIFAR-100, and SVHN. The results indicate that the proposed model is competitive with the state-of-art models and shows an average relative improvement of 5% in accuracy for several datasets, despite its greater simplicity with respect to more recent approaches.
Highlights
We propose a co-training model based on classifiers given by deep neural networks to the views of the main algorithm
We focus on the comparison of our method with respect to classical co-training and several state-of-art methods, an interesting future topic is the study of the view representations, e.g., how different are the resulting views considering the iterations of the main algorithm? Overall, this analysis suggests that DSSCo-Training performance can be improved by making a careful choice of the regularization hyperparameter, as well as the number of internal iterations of training views
This work introduces DSSCo-Training, a semi-supervised learning method for visual object recognition based on the co-training algorithm using deep neural networks supported by self-supervised auxiliary networks considering rotation and filter models
Summary
The vast majority of computer vision tasks are primarily based on the ability to recognize categories as well as specific objects. Visual category recognition seeks to recognize different instances of a generic category that belong to the same conceptual class, for example, people, houses, or cars. Self-supervised learning is based on surrogate tasks that can be formulated using only unsupervised data in such a way that to reach their goal they require learning characteristics and representations of the original object [14]. The advantage of this type of learning is to avoid the costly and time-consuming process of labeling the data.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.