An introduction to fundamental concepts in deep learning with an emphasis on applications in the life sciences. Topics include basics of machine learning, introduction to neural networks, optimization, convolutional neural networks, and recurrent neural networks. We will use the Python programming language to implement deep learning models with an emphasis in core areas of bioinformatics.
Athena Title
Intro to Deep Learning Biology
Prerequisite
(MATH 2250 or MATH 2250E) and (MATH 3300 or MATH 3300E) or permission of department
Semester Course Offered
Offered fall
Grading System
A - F (Traditional)
Student learning Outcomes
Students will be able to describe fundamental concepts in deep learning and their application to a variety of problems in the biological sciences.
Students will be able to distinguish the principals of good machine learning: select a suitable model for a given application and demonstrate how to monitor and respond to feedback.
Students will be able to select and implement deep learning models and algorithms using Python.
Students will be able to design and execute a project that entails choosing a biological data set and using deep learning approaches on a problem of interest.
Topical Outline
Review of important material from mathematics and statistics
A. Linear algebra (including linear transformations, norms, decompositions, trace and determinant)
B. Probability (including random variables, distributions, conditional probability, expectation)
Basics of Machine Learning
A. Learning algorithms
B. Estimation, bias, variance, overfitting and underfitting
C. Maximum likelihood and Bayesian inference
D. Supervised and unsupervised learning
E. Stochastic gradient descent
Deep Feedforward Networks
A. Training and evaluation
B. Regularization and optimization
Convolutional Neural Networks
A. Convolution operation and motivation
B. Pooling
C. Variants of basic convolution function
D. Data types
E. Efficient convolution algorithms
Recurrent Neural Networks
A. Graph unfolding, parameter sharing and graphical models
B. Gradient computation
C. Recursive neural networks
D. Long-short-term memory
