EMBO-DL4MIA Pre-course exercises and materials

Schedule:

• Webinar 1 (April 15th 14:00-17:00 CEST): Q&A session mainly focusing on notebooks 1 and 2, but happy to answer any questions.
• Webinar 2 (April 29th 14:00-17:00 CEST): Q&A session mainly focusing on notebooks 3 and 4, but happy to answer any questions.
• Webinar 3 (May 3rd 14:00-17:00 CEST): Optional additional Q&A session if required, can answer anything left unclear from the previous two sessions.

setup

The notebooks have been setup to run on Google Collab

Intro to Image Analysis Notebook 1

The exercises in the first notebook will refresh your image analysis knowledge and teach basic image analysis methods in python with scikit-image.

• 1_intro_to_Image_Analysis.ipynb: image analysis refresher on data from the Kaggle Data Science Bowl for nucleus segmentation.

If you are unfamiliar with image analysis or python in general, please check out the additional materials below.

Additional materials:

Introductions to image analysis for biology (ibiology videos)

• Introduction to Bioimage Analysis
• Bioimage Analysis

Bioimage analysis with python and jupyter (neubias webinar)

• part 1
• part 2
• Q&A
• slides
• notebooks

Introductions to python and data science in python

• Tutorial: Basic Python
• Python for Data Science and AI - basic
• Python 3 Programming Specialization - different levels
• Python for science - lecture notes, useful to walk through

Image analysis with python and scikit-image

• Tutorial: Image processing with Python
• Skimage examples
• Python image manipulation tools

Image Classification with pytorch Notebooks 2-4:

We provide exercises for pytorch.

Exercises for Image Classification with pytorch. Contains the following exercises:

• 2_Data_Preparation.ipynb: pytorch dataloader for classification.
• 3_Logistic_Regression.ipynb: image classification with a simple logistic regression model.
• 4_Multi_Layer_Perceptron.ipynb: image classification with a multi-layer perceptron.

All the exercises use the CIFAR10 Dataset.

Content

These exercises will cover basic machine learning and computer vision concepts relevant for image classification. You can find lectute notes recapping these contents (here), covering:

• Machine Learning Basics
  • Supervised Learning
  • Artificial Neural Networks
  • Stochastic Gradient Descent
• Computer Vision Basics
  • Tasks: Image Classification, Object Detection and Segmentation
  • Computer vision with preset features and shallow learning
  • Preview of feature learning and CNNs
• Deep Learning Frameworks
  • Why do we need deep learning frameworks?
  • Overview of the popular frameworks
  • Introduction to pytorch

There are also video recordings from an EMBL course on these topics:

• Video 1
• Video 2
• Video 3
• Video 4

And there are lots of other related materials available online, for example:

Pytorch

• Building networks from torch primitives