ACM-IKDD Summer School

on Data Science

July 4th – 16th, @ IIT Gandhinagar, Sponsored by ShareChat

 

This school is about the algorithmic, statistical, and engineering challenges associated with various stages of data analysis. Each of the sub-topics will include both theoretical and hands-on aspects. We will cover how to collect and clean up data, probabilistic models for data, and various algorithmic challenges that arise when scaling these models to large data. We will also do deep dives for data-driven modeling in three different scientific domains  – natural language processing, computer vision, and earth and climate sciences. Lastly, we will learn how to deploy a machine learning model in production and keep it up-to-date. There will be multiple lectures on each subtopic, and participants will be taken from the basics to some of the cutting-edge questions in these areas. 

Topics

  • introduction to data collection pipeline, tools and techniques for data processing (e.g., normalization, outlier removal), descriptive statistics, visualization
  • models for supervised learning –  MLE, MAP, and fully Bayesian modeling, clustering, matrix factorization, spatio-temporal data modeling
  • algorithms for data and dimension reduction
  • experiment design and model evaluation, A/B testing etc.
  • modern models for NLP, computer vision, data-driven modeling for earth and climate sciences
  • data-science lifecycle, standard practices of MLOps

 Background / prior courses recommended

The following background is expected from the participants. The links curated contain material that can be used to revise/pick up the necessary material.  

  • Any specific software (Matlab, Python, etc. ) to be used: Python

Speakers

Academic coordinators

IIT Gandhinagar: – Anirban Dasgupta (anirbandg@iitgn.ac.in)

 

Schedule

Resources

Applications of ML and ethics: Video

Formal introduction to ML : Note|Video

Bayesian Machine Learning : NoteNotebook |  PDF

Probabilistic View of Linear Regression: PDF

Logistic Regression: Video|NotePDF

Naive Bayes : PDF

 

Sponsors