Radial Basis Function Networks for Convolutional Neural Networks with Mohammadreza Amirian
Radial basis function neural networks (RBFs) are prime candidates for pattern classification and regression and have been used extensively in classical machine learning applications. However, RBFs have not been integrated into contemporary deep learning research and computer vision using conventional convolutional neural networks (CNNs) due to their lack of adaptability with modern architectures. In this paper, we adapt RBF networks as a classifier on top of CNNs by modifying the training process and introducing a new activation function to train modern vision architectures end-to-end for image classification. The specific architecture of RBFs enables the learning of a similarity distance metric to compare and find similar and dissimilar images. Furthermore, we demonstrate that using an RBF classifier on top of any CNN architecture provides new human-interpretable insights about the decision-making process of the models. Finally, we successfully apply RBFs to a range of CNN architectures and evaluate the results on benchmark computer vision datasets. I
Mohammadreza Amirian received the M.Sc. degree in communications technology from Ulm University, Ulm, Germany, in 2017, where he is currently pursuing the Ph.D. degree. He is currently working as a Researcher with the Institute of Applied Information Technology (InIT), Zurich University of Applied Sciences (ZHAW), Winterthur, Switzerland. His research interests include biophysiological signal processing for person-centered medical and affective pattern recognition. His current research interests include interpretable deep-learning algorithms for industrial applications and automated deep learning.
This presentation was a part of our Virtual Deep Learning Salon, to sign up for updates on these join our meetup and slack community!
http://wandb.me/meetups
http://wandb.me/fs
👩🏼🚀Weights and Biases:
Record and visualize every detail of your research, collaborate easily, advance the state of the art - we’re always free for academics and open source projects. Email [email protected] with any questions or feature suggestions.
Blog: https://www.wandb.com/articles
Gallery: See what you can create with W&B -https://app.wandb.ai/gallery
Continue the conversation on our slack community - http://wandb.me/fs