The Rochester Signal Processing Society hosted a virtual zoom session on Thurs, April 30th entitled “Fast Deep Learning Prototypes with Tensorflow and Keras Tutorial”, by RIT PhD student Miguel Dominguez. 63 people registered for this hour-long session which included the distribution of iPython notebook sample code and deep learning software installation instructions. This tutorial was very well attended, in spite of being held, non-traditionally, via Zoom.

Deep Neural Networks (DNN) are a powerful tool for computer vision, signal processing, and natural language processing tasks. The last few years have seen the development of a plethora of software tools for the development of DNNs. This tutorial discussed using Tensorflow 2.0 with the Keras API to enable rapid prototyping of DNNs with a minimum of code. All software is free open-source code. The tutorial demonstrated how to get several variants of Convolutional Networks up and running, for training, prediction, and logging.

Miguel Dominguez is a PhD candidate in Engineering at Rochester Institute of Technology, set to graduate in the summer of 2020. His research interests include graph and point cloud neural networks as well as speech processing.Image of Miguel Dominguez

The TensorBoard dashboard pictured here shows the performance of models as they are being trained.

TensorBoard dashboard showing training performance of Deep Learning Models