StackUp BMT (Build More Things) Workshop - Power up your Raspberry Pi with PyTorch & Deep Learning

With millions sold, the Raspberry Pi is one of the most popular single-board computers for IoT applications. Its key strengths are affordability, availability of hardware and broad community support.

Developed by Facebook, PyTorch is an open source machine learning library in Python that is growing in popularity with the deep learning community due to its flexibility, ease of programming and strong GPU support compared to other machine learning libraries.

Running Deep Learning models locally on the Raspberry Pi can enable IoT scenarios such as predictive maintenance, robotics, and supply chain monitoring to run without cloud connectivity.

In this one-day workshop, you will learn how to train a deep learning recurrent neural network (RNN) using PyTorch and deploy the trained model on a Raspberry Pi. The model will be used to predict equipment failure from IoT sensor data. By the end of the workshop, you will have a Raspberry Pi 3 with source code and a working RNN model to take home for your own projects.

If you have been thinking about transitioning to a tech-related career, let us help you get started. For more on NUS-ISS StackUp's course offerings and tracks, click here.

• Describe the machine learning workflow
• Describe the application of deep learning on sequential data
• Describe the overview and application of deep neural network layers such as fully-connected and Long Short Term Memory (LSTM)
• Understand how the machine learning workflow is performed using Python libraries such as Pandas, Scikit-Learn, and PyTorch
• Use PyTorch to train an RNN model using multivariate sensor data to predict equipment failure
• Deploy the trained RNN model on a Raspberry Pi to get predictions

Pre-Workshop Set-up
Due to time limitation, it is important that you perform the following installation on your notebook before coming to the workshop

Windows

1. Download and install Git for Windows: git-scm.com/download/win


2. Download and install the Putty SSH client: www.chiark.greenend.org.uk/~sgtatham/putty/latest.html


3. Download and install WinSCP: winscp.net/eng/download.php


4. Download and install latest Anaconda for Python 3.7: www.anaconda.com/download/


5. Install PyTorch:
1. Start Menu -> Anaconda Prompt:
conda create -n torch python=3
conda activate torch
conda install pytorch -c pytorch
conda install pandas scikit-learn


6. Download and install a code editor such as Visual Studio Code: code.visualstudio.com/download

1. Download and install Git for Mac: git-scm.com/download/mac


2. Download and install latest Anaconda for Python 3.7: www.anaconda.com/download/


3. Install PyTorch:
1. Launch a new Terminal:
conda create -n torch python=3
conda activate torch
conda install pytorch -c pytorch
conda install pandas scikit-learn


4. Download and install a code editor such as Visual Studio Code: code.visualstudio.com/download