Talk Timings: May 1, 2021 (Saturday) 11.00 AM – 11.55 AM IST
Abstract:
As engineers working on signal processing, we typically went from analyzing signal measurements, to developing DSP algorithms, and finally creating whole pipelines and products. Developments in machine learning and deep learning methods over the years have provided new tools to tackle signal processing problems across a broad range of industries, from predictive maintenance and health monitoring to financial portfolio forecasting and advanced driver assistance systems.
This talk will focus on fundamental ideas around application of deep learning to signals and time-series data. We will discuss the use of established pre-trained deep learning models to solve a few complex but standard problems. We will cover how you can go from building your signal datasets, to labelling your signals using apps, and preprocess your data. We will also explore various feature extraction techniques that help to create robust and accurate AI models. Finally, we will look at how to design, train, and deploy deep learning models for signal processing from scratch using MATLAB & Simulink.
Highlights:
- Easily manage signal datasets using datastores
- Using Signal Labeler and Signal Analyzer Apps for AI workflows
- Extracting signal features with established algorithms (e.g. spectrograms, scalograms, Mel Spectrograms, MFCC)
- Feature extraction techniques including AutoML techniques such as wavelet scattering and time-frequency representations
- Using application-specific pre-trained networks with a single line of MATLAB code
- Designing and training deep networks (including with CNN, LSTM, and GRU layers) using GPUs
- Deployment on embedded hardware like Raspberry Pi for prototyping real-time deep learning projects
Bio:
Praful is associated with the Education Team at MathWorks India, where he primarily works with academia in India towards making STEM education engaging and accelerating research using MATLAB & Simulink. He completed his undergraduate studies in Biomedical Engineering from Manipal Institute of Technology, and his MS and PhD from the Department of Electronics and Electrical Communication Engineering at Indian Institute of Technology Kharagpur, focusing on non-invasive blood glucose monitoring using photoacoustics. Prior to joining MathWorks, he was a R&D Engineer and later a Research Scientist at National Brain Research Centre, Gurgaon, where he worked on developing an MRI brain template for the Indian population. He is passionate about learning, teaching, and research in multi-disciplinary domains involving instrumentation, signal processing, statistics, and machine Learning.