Mathematical Aspects of Machine Learning

Our class is made up of people with diverse backgrounds and interests. As a first activity, we need to get to know each other so that we can form groups with complementary skills and overlapping interests.

To do this assessment, we will divide up the whiteboards in Bousefield A105 by skills and interests and people will write their names in places that match their profile.

Two comments:

**I know a few of the things on the list, but there are lots of things here I don’t know,
though I wish I did, so don’t be intimidated!**

**The verb “to know” in this context means – I have done this, I’m familiar with it. It does not
mean “I am prepared
to take an exam on the topic right now!”**

- I know how to look at basic elements of a dataset in a jupyter notebook using python or Rstudio
- I know how to use R or python libraries for machine learning
- I know how to write numerical algorithms in a high performance language like C++
- I know how to use UConn’s cluster computer
- I know how to use cloud computing resources like AWS or google cloud
- I know how to read documentation for an API and use it extract data from a website such as twitter
- I know how to package a python program or R program for distribution
- I know how to use tensorflow, keras, or pytorch to do neural network computations
- I know how to use pymc3, stan, or other software to do monte-carlo simulations
- I know how to use git and github (or similar repository) to manage a project

- I know about basic statistical inference (hypothesis testing, p-values)
- I know about standard probability distributions (Normal, t-distribution, Poisson, Binomial, multinomial,…)
- I know about linear regression (* overlaps with math)
- I know about linear models generally (* overlaps with math)
- I know about regularization of regression such as the lasso
- I know about Bayesian methods in settings such as regression or hypothesis testing

- I know about linear algebra, especially the theory of quadratic forms and singular value decompositions
- I know what principal component analysis is
- I know about the theory of gradient descent
- I know about measure theory and theoretical probability such as the proof of the central limit theorem
- I know about analysis on graphs, such as the graph laplacian and the spectra of graphs
- I know something about information theory and its connection to probability
- I know the theory of monte-carlo sampling such as the metropolis method or hamiltonian sampling

- I know how to create a website using gh-pages or another hosted site
- I can do design and make things look cool
- I know how to develop web templates in CSS and templating languages like jekyll
- I know how to create documentation for code using resources provided by github or other similar sites

- I am interested in financial applications
- I am interested in sports analytics
- I am interested in image classification
- I am interested in natural language processing
- I have a specific interest from my own research
- I am interested in competitions