skip to primary navigationskip to content


Wednesday 9 September 2015

Registration from 8.30 am

9 am Introduction – Course Overview – Machine Learning Packages (AF)

10 am Linear Classification, the art of separating data into different categories where these categories are separated by (for example in 2 D) a straight line. Particular topics are: Features – Fisher’s Discriminant Analysis –Linear Discriminant Analysis – Multiple Classes – Online Learning – Perceptron – Support Vector Machine – Passive Aggressive Algorithm (AF)

11.00 am Coffee

11.30 am Clustering, where latent variable models are learned through the identification of latent variables in objects and group them accordingly. Particular topics are: k-Means - Mixture Models – Dirichlet Processes  (GP)

12.30 pm Lunch

1.30 pm Non-linear Classification, where the separation between classes cannot be a straight line, but is a curve. Particular topics are: QDA – Kernel Trick – k-Nearest-Neighbours – Boosting – Cascades – Logistic Regression – Neural Networks (AF)

2.30 pm Practical Session

3.30 pm Tea

4.00 pm Classification: Pedestrian Detection (GP)

5.00 pm Practical Session

7.00 pm Formal Dinner


Thursday 10 September 2015

9.00 am Regression, the art of inferring information about unseen data after training with samples: Linear Regression – Polynomial Regression – Partial Least Squares – Over-/Under-fitting – Regularization (AF)

10.00 am Feature Extraction and Dimensionality reduction, which crystallizes the important information contained in the data: Particular topics are: Principal Component Analysis – Independent Component Analysis (GP)

11 am Coffee

11.30 am Regression in a probabilistic framework. Particular topics are: Generalized Linear Model – Expectation Maximization  (AF)

12.30 pm Lunch

1.30 pm Feature Learning part 1:, where the algorithm answers the question which common patterns are contained in the data. Particular topics are: Auto encoders – Restricted Bolzmann Machines (GP)

2.30 pm Data Assimilation, where knowledge of the underlying system is used (AON)

3.30 pm Tea

4.00 pm Practical Session

5.00 pm Concluding Session: Regression: Image Restoration (AF) and Earth System Science (AON)

Friday 11 September 2015

9.00 am Regression, making the most of as little data as possible. Particular topics are: Bayesian Regression – RVM – Compressive Sensing (AF)

10.00 am Feature Learning part 2: where the algorithm identifies sparse representations of the available data through optimization and infinite dimensional spaces. Particular topics are: Sparse Coding – Beta Processes (GP)

11.00 am Coffee

11.30 am Concluding Session: Deep learning: Dictionary Learning (GP)

12.30 pm Lunch

Anita Faul - AF (Cambridge),
Georgios Pilikos - GP (Cambridge)
Alan O'Neill - AON (Reading)