Embedded Systems

Neural Network Framework

Bachelor’s Thesis / Master’s Thesis / Student Research Project

Abstract

The chair has its own framework for audio processing tasks (voice activity detection, keyword spotting) using different neural networks (TC-Resnet, SincNet, Branchynet, Wavenet, LSTMs,…). It has the ability to extract different features (Spectrogram, MFCC, Mel Features), quantization (weights, bias, activation) using Nervana Distiller, advanced noise handling and many more. We have many different topics for a thesis: Integration of new datasets from new sensors (accelerometer, electrocardiogram, ….), implementation of new neural networks, optimization and mapping of neural networks for our hardware accelerator UltraTrail, implementation of new quantization methods, improvement of our training loop for neural networks with early exits, hyperparameter search and many more. The framework is built on PyTorch, PyTorch Lightning and Nervana Distiller. For the training of the neural networks we have a cluster with 160 Geforce GTX1080Ti or some local machines equipped with Tesla P100.

Requirements

  • You should have basic knowledge of neural networks and Python
  • Knowledge of PyTorch, quantization and signal processing is beneficial but not necessary.

Contact

Bringmann, Oliver

Frischknecht, Adrian