Johann West: Reimagining ML-Dataset with Artificial Intelligence скачать в хорошем качестве

Johann West: Reimagining ML-Dataset with Artificial Intelligence

Description: Data is the new oil. Like oil, data is valuable, but if unrefined it cannot really be used. For any data scientist, a large part of the process of building a useful machine learning model is preprocessing data. Traditionally, there’s many methods to this madness. You might choose an imputation strategy for your missing values. You might write a script to do some string matching. You might even have some poor intern go row by row filling in data based on their own knowledge. What if instead of using these semi-manual and manual methods for dataset preprocessing, you let an AI do it for you? We’d like to explore the possibility of shifting the paradigm here. In this talk, we’ll walk through 3 potential strategies for utilizing agentic systems, powered by RAG, to automate and elevate the ML dataset preprocessing workflow. You’ll learn how AI powered null/missing value handling, data consolidation, and synthetic feature generation can take machine learning datasets to the next level. Bio: Johann West is a Machine Learning Engineer at Hagerty, and co-founder of PureML. He received his BS in Computer Science from Vanderbilt University in Nashville, TN, and is now located in the SF Bay area. Johann is passionate about the possibilities AI provides in benefiting the ML model building process, and is pursuing this through the PureML project.