## QBI 2019

**Machine and deep learning in bioimage analysis and microscopy: Tutorial introduction for microscopists [conference link]**

- Lecture slides [PDF part 1, PDF part 2]
- Request DataCamp access: click here
- Recommended textbooks for self-study:
*Machine Learning: A Probabilistic Perspective*[Murphy 2012]*The Elements of Statistical Learning: Data Mining, Inference, and Prediction*[Hastie, Tibshirani, Friedman 2009]*Deep Learning*[Courville, Goodfellow, Bengio 2015]*A First Course in Machine Learning*[Rogers and Girolami 2016]*Information Theory, Inference, and Learning Algorithms*[Mackay, 2005].

- Agenda:

11:00 – 12:30 **What is machine learning? **(Seth Flaxman, Imperial College)

- Supervised vs unsupervised learning: k-nearest neighbors, k-means clustering
- From linear to nonlinear methods: transformations, feature engineering, and the kernel trick
- Bias vs variance, and how to deal with the problem of overfitting.
- Regularization, lasso, ridge regression, and the elasticnet

12:30 – 13:30 Lunch

13:30 – 14:30 **Advanced machine learning **(Seth Flaxman, Imperial College)

- Ensemble methods
- Gaussian processes

14:30 – 15:45 **Optimization **(Julien Mairal, INRIA)

In this lecture, we will discuss a few techniques recently introduced in machine learning and optimization to deal with large amounts of data. We will focus on regularized empirical risk minimization problems, which consists of minimizing a large sum of functions, and cover also stochastic optimization techniques for minimizing expectations. Concepts we are planning to cover include

- Stochastic gradient descent techniques with variance reduction
- Nesterov’s acceleration
- Quasi-Newton techniques.

We will also consider variants that allow dealing with nonsmooth regularization such as the l1-norm, which is useful for sparse estimation in high dimension.

15:45 – 16:45 **Deep Learning **(Christophe Zimmer, Institut Pasteur)

- what deep learning can do
- image classification
- other applications

- how deep learning works
- forward propagation
- convnets
- training
- babysitting neural nets

DIY deep learning with Keras