Overview
Syllabus
– Introduction
– Overview & Summary of the Challenge
– 2-Stage Recommender Systems Pipeline
– Stage 1 for Tasks 1 & 2: Candidate Generation & Co-Visitation Matrices
– Stage 1: Training Product Embedding
– Stage 2 for Tasks 1 & 2: Pipeline Summary
– Stage 2: Reranker model - Feature Selection & Engineering
– Multilingual Transfer Learning
– Task 3: Next Product Title Generation Solution
– Q&A Session
– Q&A: Challenges of Using Embedding
– Q&A: Unexpected Findings
– Q&A: Critical Model Improvements
– Q&A: Differences from OTTO Multi-Objective RecSys Competition
Taught by
NVIDIA Developer