Machine Learning for Scientists
Lecture
Introduction
Structuring Data without Neural Networks
Principle Component Analysis
Kernel PCA
t-SNE as a Nonlinear Visualization Technique
Clustering Algorithms: the example of
\(k\)
-means
Exercise: Principle Component Analysis
Exercise: Dimensionality Reduction
Supervised Learning without Neural Networks
Linear Regression
Binary Classification and Support Vector Machines
More than two classes: Logistic Regression
Exercise: Linear Regression
Exercise: Classification without Neural Networks
Supervised Learning with Neural Networks
Computational neurons
Training
Advanced Layers
Recurrent neural networks
Exercise: Dense Neural Networks
Exercise: Machine Learning Optimizers
Exercise: Learning Rate Scheduling
Exercise: Regularizing Neural Networks
Exercise: Convolutional Neural Networks
Exercise: Discovery of Exoplanets with RNNs and CNNs
Unsupervised Learning
Restricted Boltzmann Machine
Training an RNN without Supervision
Autoencoders
Generative Adversarial Networks
Exercise: Denoising with Restricted Boltzmann Machines
Exercise: Molecule Generation with an RNN
Exercise: Anomaly Detection
Interpretability of Neural Networks
Dreaming and the Problem of Extrapolation
Adversarial Attacks
Interpreting Autoencoders
Exercise: Transfer Learning and Adversarial Attacks
Reinforcement Learning
Exploration versus Exploitation
Finite Markov Decision Process
Policies and Value Functions
Temporal-difference Learning
Function Approximation
Concluding Remarks
About us
Who we are
Index
