Machine Learning Specialization · Unsupervised Learning, Recommenders, Reinforcement Learning

Deep Learning for Content-Based Filtering

A two-tower neural recommender that combines user features and movie features into learned embeddings, then predicts ratings and similar items.

Two-tower neural networkUser embeddingsItem embeddingsFeature scalingSimilarity search

GitHub

What this project is about

This project moves recommendation from pure rating-matrix structure toward a neural two-tower design. One tower learns user representation, the other learns item representation, and the final similarity/dot-product signal drives predictions.

Implementation focus

The important lesson was that recommendation can become representation learning: once item and user embeddings exist, they are useful both for rating prediction and for finding similar movies.

Key results and artifacts

397 users, 847 movies, 25,521 ratings
Training vectors: 50,884
Test loss around 0.3734

What I learned

This project showed how recommender systems can move from raw collaborative signals to neural embeddings that support both prediction and retrieval.

Milestone deep dive

What this project proves in my learning path.

A two-tower neural recommender that combines user features and movie features into learned embeddings, then predicts ratings and similar items. The milestone is included because it shows a complete learning unit: a problem, a data representation, an implementation path, measurable outputs, and a lesson that later connects to larger AI systems.

Input layerusers, items, ratings, and feature vectors

Model layerlatent vectors or neural embedding towers

Evaluation layertests, metrics, plots, or qualitative output

Artifact layerpersonalized predictions and similarity scores

Problem framing

What I was trying to solve

This page is not just a lab description. It explains how I transformed the assignment idea into an engineering story: what the model receives, what it optimizes, what output it should produce, and how the result can be inspected.

Implementation

What I coded or assembled

The implementation focus was the mechanism behind the result: functions, model blocks, training logic, preprocessing, evaluation, and the small technical details that make the notebook work rather than only look correct.

Concepts

Core concepts

Two-tower neural networkUser embeddingsItem embeddingsFeature scalingSimilarity search

These tags show the technical ideas this milestone added to my toolkit.

Learning result

What changed after it

This project showed how recommender systems can move from raw collaborative signals to neural embeddings that support both prediction and retrieval.

Important outputs and evidence

397 users, 847 movies, 25,521 ratings
Training vectors: 50,884
Test loss around 0.3734

How I would explain it in an interview

I would describe this milestone by starting from the data representation, then explain the model or algorithm, then show what was measured. The important part is not only the final number; it is the ability to reason about why the method works, what assumptions it makes, and what limitations should be improved next.