Stanford CS329H: Machine Learning from Human Preferences
Course Content
Stanford CS329H: Machine Learning from Human Preferences | Autumn 2024 | Introduction
This lecture introduces decision trees, a classic machine learning method that makes choices by asking a series of yes/no or small set-of-value questions. Each internal node is a test on a feature (like 'Is it Friday or Saturday?'), each branch is the outcome, and each leaf is a predicted label. You classify by starting at the root, following the test outcomes, and stopping at a leaf.
Stanford CS329H: Machine Learning from Human Preferences | Autumn 2024 | Preference Models
Decision trees are models that turn feature values into decisions like Yes/No. You follow a path from the root question down through branches until you reach a leaf decision. They are great for problems where the target is discrete, like classifying outcomes. They also tolerate noisy data and can work even when some feature values are missing.
Stanford CS329H: ML from Human Preferences | Autumn 2024 | Model-based Preference Optimization
Decision trees are flowchart-like models used to predict a class (like yes/no) by asking a series of questions about features. You start at the root and follow branches based on answers until you reach a leaf with a class label. Each internal node tests one attribute, each branch is an outcome of that test, and each leaf gives the prediction.
Stanford CS329H: Machine Learning from Human Preferences | Autumn 2024 | Mechanism Design
This lesson explains the core pieces of machine learning: data (X and Y), models f(x;ΞΈ), loss functions that measure mistakes, and optimizers that adjust ΞΈ to reduce the loss. It divides learning into supervised (with labels), unsupervised (without labels), and reinforcement learning (with rewards). The focus here is on supervised learning, especially regression and classification, plus a short intro to k-means clustering.
Stanford CS329H: Machine Learning from Human Preferences | Autumn 2024 | Human-centered Design
Machine learning means building computer systems that learn patterns from data to make predictions or decisions. Instead of hand-coding rules like βif you see whiskers and triangle ears, itβs a cat,β we feed many labeled examples and let algorithms learn the rules. The goal is to generalize from past data to new, unseen cases in the future.
Stanford CS329H: Machine Learning from Human Preferences | Autumn 2024 | Ethics
The lecture explains regularization, a method to reduce overfitting by adding a penalty to the cost (loss) function that discourages overly complex models. Overfitting is when a model memorizes noise in the training data and fails to generalize. Regularization keeps model parameters (weights) from growing too large, which helps models generalize better to new data.
Stanford CS329H: Machine Learning from Human Preferences | Autumn 2024 | Voting
This lecture kicks off an Introduction to Machine Learning course by explaining how the class will run and what you will learn. The instructor, Byron Wallace, introduces the TAs (Max and Zohair), office hours, and where to find all materials (Piazza/Canvas). The course has weekly graded homework, a midterm, and a group final project with a proposal, report, and presentation. Lectures are mostly theory and recorded; hands-on coding happens in homework.
Stanford CS329H: Machine Learning from Human Preferences I Guest Lecture: Joseph Jay Williams
Machine learning is about computers learning patterns from data instead of being told each rule by a programmer. Rather than hardcoding how to spot a cat in a photo, you show many labeled cat and non-cat images, and the computer learns what features matter. This approach shines when problems are too complex to describe with step-by-step rules.