To start, head over to [DataRes_Research_Coding_Assessment.ipynb] and download the notebook. Open it in either Google Colab or Kaggle notebook (I prefer Kaggle because the GPUs are faster) and start coding! This assessment will be done using PyTorch, so apologies if you're more familiar with TensorFlow (although the concepts will still carry over). After you're done, send me the file at larrylu0302@g.ucla.edu with the subject "DataRes Research Application S24".
Hey! I'm Larry, and I'll be the research head this year. This is a quick assessment to just test basic proficiency with deep learning and libraries.
Deep learning is the hottest branch of artificial intelligence right now thanks to how powerful it is, but anyone can copy a repo and change some code around to get a working model without actually understanding the rationale or math behind it. We won't be approaching these deep learning projects from the perspective of a data analyst or programmer; we are approaching them with the mindset of a researcher. Just like any researcher, we are going to dive deeply into novel literature regarding our field, learning from them and extend them into applications of our liking. In research, we will be focusing on reproducing papers and turning theory into our own code implementations, as this is the best way to learn what’s going on underneath the hood. If this still sounds interesting to you, then please apply! :D
That is what makes the DataRes Research team different from other data science organizations. We are scientists first, and then programmers.
S22: Graph Deep Learning: Integration of GCN Pipeline with Neo4j DBMS for Social Analysis
F23: Representation Learning with Transformer-based Language Models with Temporal Data
S24: Large Language Models to Solve Math Word Problems