We will see with simple examples on how to leverage a computational model of cancer to learn from experimental data by using Bayesian inference. By going through simple examples and step-by-step concepts we will build a machine learning algorithm from scratch.
The course is organized into 5 separate modules, each complete with an auto-evaluation test and teaching material, videos and complementary notes. The modules are:
-
Statement of the problem: why it is interesting to understand population dynamics in cancer
-
Galton-Watson branching processes for cell population models
-
Numerical simulations of a Galton Watson process in Python
-
A shallow and operative introduction to Bayesian inference
-
Putting it all together: inferring from cell population data
At the end of the learning process, students will be able to:
Knowledge and Understanding:
-
identify and describe elementary building blocks of machine learning through examples of a simple algorithm.
-
explain the role and relevance of computational models in ensuring the explainability of a process.
-
define Bayesian inference and stochastic branching processes, and explain how these can be integrated in a model.
Applying Knowledge and Understanding:
-
apply concepts to solve new problems by adapting examples from lectures and proposing small variations to presented solutions.
Making Judgements:
-
assess the level of transparency and interpretability of simple machine learning algorithms and justify the importance of explainable AI in scientific contexts.
Communication Skills:
-
present examples of simple algorithms and clearly describe their conceptual components in accessible language.
Learning Skills:
-
apply a step-by-step approach to break down and analyze the structure of a complex algorithm.
At the end of each module the student will be able to try a self-evaluation test, and to try the proposed additional exercises with solutions.
Within each module there will be submodules which are conceptually self-consistent and standing complete with additional resources, the slides used for learning and the written content in the form of a pdf booklet. Where possible there will be exercises either theoretical (for example a small calculation to perform) or computational (a script to write). Script used for teaching will also be available.
Basic calculus (integrals and derivatives in one or two variables), basic programming (suggested Python).
Alberto Puliafito
For any queries or technical issues relating to the course and the platform, please contact:
edvancedeh@unito.it
Contact details of the MOOC lecturer: