USC Information Sciences Institute Courses

AI in Structural Biology: Roots of Success

Explore the history and impact of AI in protein structure prediction, focusing on AlphaFold 2's breakthrough and its implications for structural and cell biology research.

• YouTube
• 57 minutes
• On-Demand
• Free Video

DeLLMa: A Framework for Decision Making Under Uncertainty with Large Language Models

Enhance decision-making accuracy in uncertain environments using DeLLMa, a multi-step scaffolding framework based on decision and utility theory principles. Improve LLM performance by up to 40% in complex scenarios.

• YouTube
• 1 hour 5 minutes
• On-Demand
• Free Video

Multimodal Machine Intelligence and Health

Explore multimodal machine intelligence applications in medical sciences and health research, addressing challenges in creating trustworthy, inclusive AI for clinical and quality of life domains.

• YouTube
• 1 hour 16 minutes
• On-Demand
• Free Video

Informative Example Selection for In-Context Learning

Explore efficient approaches for selecting informative examples in In-Context Learning, enhancing large language model performance without additional training.

• YouTube
• 59 minutes
• On-Demand
• Free Video

The Data Provenance Initiative: A Large Scale Audit of Dataset Licensing and Attribution in AI

Explore a comprehensive audit of 1800+ text datasets, examining legal and ethical risks in AI data practices. Gain insights into dataset lineage, licensing, and the divide between open and closed datasets.

• YouTube
• 56 minutes
• On-Demand
• Free Video

Do Androids Know They're Only Dreaming of Electric Sheep?

Explore transformer language models' internal representations to detect and analyze hallucinatory behavior in AI-generated content, with insights on task-dependent patterns and evaluation methods.

• YouTube
• 59 minutes
• On-Demand
• Free Video

A Gateway to Trustworthy AI: Using Visual Analytics to Unmask Coincidental Correlations

Explore visual analytics techniques to identify and address spurious correlations in AI systems, enhancing trustworthiness and reliability in machine learning and data-driven decision-making.

• YouTube
• 1 hour 1 minute
• On-Demand
• Free Video

Self-Play Fine-Tuning Converts Weak Language Models to Strong Language Models

Explore a novel fine-tuning method for Large Language Models that improves performance without additional human-annotated data, using self-play mechanisms and theoretical guarantees.

• YouTube
• 1 hour 3 minutes
• On-Demand
• Free Video

Red Teaming Language Model Detectors with Language Models

Explore strategies to compromise LLM detectors using synonym replacement and instructional prompts, revealing vulnerabilities in text detection systems and emphasizing the need for robust safeguards.

• YouTube
• 48 minutes
• On-Demand
• Free Video

Harnessing Black-Box Control to Boost Commonsense in Language Models' Generation

Enhance language models' commonsense reasoning without fine-tuning. Learn about BOOST framework for steering frozen models towards more reasonable outputs using an interpretable evaluator and auxiliary head training.

• YouTube
• 55 minutes
• On-Demand
• Free Video

An Introduction to Egyptian from Hieroglyphs to Coptic

Explore the fascinating 5,000-year history of Egyptian language, from hieroglyphs to Coptic. Discover ancient scripts, decryption methods, and linguistic intricacies with a computational linguist.

• YouTube
• 1 hour 2 minutes
• On-Demand
• Free Video

Developmental Science Insights for Artificial General Intelligence Creation

Explore how insights from infant development studies inform AI research, focusing on evaluating AI systems using tools from developmental psychology to advance AGI creation.

• YouTube
• 1 hour 2 minutes
• On-Demand
• Free Video

Visualizing and Analyzing Event Sequence Data

Explore techniques for analyzing event sequence data, including causal inference, deep learning, and generative AI, to extract valuable insights from high-dimensional business process data.

• YouTube
• 58 minutes
• On-Demand
• Free Video

Analyzing the Multilingual Performance of Text-to-Image Models

Explore multilingual capabilities of text-to-image models, examining performance variations across languages and introducing the CoCo-CroLa benchmark for measuring conceptual coverage and identifying biases.

• YouTube
• 1 hour 1 minute
• On-Demand
• Free Video

What We Learned from 570K ChatGPT Interaction Logs In The Wild

Explore insights from 570K ChatGPT conversations, revealing diverse user prompts and potential toxic use-cases. Learn about fine-tuning instruction-following models using this rich dataset.

• YouTube
• 59 minutes
• On-Demand
• Free Video