From dataset understanding to result presentation and how to employ deep acyclic graph models in Predictive Analytics
Welcome to 2025-2026 Data Science with Deep Learning This is the graduate course in Data Science - a 11 lecture & hands-on labs journey in the hot field that combines programming, math, statistics, bio/physics in order to provide the today world with the most advanced information technologies: from recommender systems to chatbots, from predictive analytics to visual scene understanding and many others.
- we meet each Monday 16h00-20h00 on Teams!
- Graded project - real-life problem: 70% (minimum 30% for grading)
- Research paper (important) presentation individual: 30%
- Secondary project (optional extra-credit): 15%
- 2nd Research paper (optional research highlight pres.): 15%
"root"all the slidesmandatory_papersall the available papers to chose from for mandatory research presentationscriptsscripts/notebooks from each interactive lectureresourcesimportant materials such as linear algebra, stats, etcdatafolder with toy and maybe-not-so-toy datasets
- IMPORTANT: You ARE ALLOWED to use Generative AI for you projects and presentations. BUT you SHOULD have full 100% understanding of you presentation and your generated code.
- 1-3 students per project
- end-to-end small application that will have the following components:
- script or Jupyter notebook model definition and training - using scrapped / real life data and some basic model architecture searching techniques -> will generate a serving candidate model
- simple straightforward model serving backend - using Flask or FastAPI
- simple frontend - using Bokeh, React or Angular or anything else - for example just a simple HTML page with some data entry controls within a submittion form and a result section below the form
- simple PPT/Slides/PDF presentation
- focus on the parts of the paper that you found out to be the most interesting and you liked the most
- value and quality over quantity