This lecture will present a side project focused on automating a daily task I have been struggling with. I hate choosing what to eat, so I combined my coding hobby with some data I structured to build a dinner-idea suggestion system. I designed the features I needed and learned how to structure a code in a readable way. By structuring the code in an extensible way I made it easier to extend the program to new interfaces, right now having both a CLI and a Streamlit web app that use the same core functions.
Introducing a technique from the DevOps world for configuring pipelines for processing data of various (and different) formats, characteristics and noise in a generic and easy way.
Distant Annotation (aka Distant Supervision) is an annotation method that allows training data to be labeled automatically. It has become the standard method for relation extraction tasks. The method utilizes an existing database, such as Freebase, Wikipedia, or a domain-specific database, to collect examples for the relations we want to extract.
**Causal ML** brings ML practitioners exciting opportunities to go beyond correlation, with success stories in multiple big tech companies. However, being rooted in the theory of traditional Causal Inference, Causal ML is still less accessible to data scientists who are already juggling to master numerous sub-fields. In this talk we will walk a first step to bridge this gap, by exploring the rich landscape of Causal ML from the data scientist perspective, focusing on “what’s in it for us” and which practical tools can be used.
Wordle is a daily game that’s gone viral in 2022. In this game, each day there’s a common 5 letter word you need to discover with the least number of tries. We’ll learn how to use simple Python to build an agent to play the game for us.
ספוקן וורד משעשע ומעורר מחשבה על הורות, הייטק, דאטה ורווחים.