Simons Institute Courses

Bridging the Data Gap Between LLMs and Children - Understanding Higher-Level Intelligence

Explores the disparity in data requirements between AI models and human learning, examining potential explanations and presenting new insights on multimodal input and evaluation methods.

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

How Grounded Are Ungrounded LLM Models

Explore the groundedness of large language models, examining their capabilities and limitations in understanding and interacting with the real world.

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

Streaming Algorithms for Connectivity Augmentation

Explore cutting-edge streaming algorithms for enhancing network connectivity, focusing on sublinear approaches and their applications in graph theory and data structures.

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

A Bi-metric Framework for Fast Similarity Search

Explore a novel bi-metric framework for efficient nearest neighbor search, combining accurate but costly ground-truth metrics with faster proxy metrics to optimize data structures and query procedures.

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

Hypothesis Selection with Computational Constraints

Explore optimal tradeoffs between memory, sample complexity, and time in hypothesis selection algorithms for distribution matching, with focus on achieving near-optimal accuracy efficiently.

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

Applications of Sub-linear Algorithms in Other Fields

Explore how sub-linear algorithms impact diverse fields, enhancing efficiency and scalability in data analysis and problem-solving across disciplines.

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

Extroverted Sublinear Algorithms

Explore extroverted sublinear algorithms with Google Brain researcher Ilya Mironov, delving into innovative computational techniques for efficient data processing and analysis.

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

Sublinear Time Algorithms for Max-Cut Approximation on Expander Graphs

Explore random walks on graphs to achieve better-than-1/2 approximation for max-cut on expanding and clusterable graphs, using sublinear time algorithms.

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

Sublinear Algorithms in Social Networks via Core-Periphery Decomposition

Explore sublinear algorithms for large networks using core-periphery structure and degree distribution. Learn about accelerating tasks like node sampling and shortest path computations.

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

Linear and Sublinear Algorithms for Graphlet Sampling

Explore efficient algorithms for graphlet sampling, including linear and sublinear preprocessing techniques, with applications to semi-streaming and MPC settings.

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

Sublinear Insights: A Faster Classical Algorithm for Edge Coloring

Explores a novel randomized algorithm for near-linear time graph edge coloring, improving on decades-old methods. Discusses implications for Vizing's theorem and dense graph coloring efficiency.

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

Large Scale Private Learning on Data Streams and the Buffered Linear Toeplitz Operators

Innovative algorithm for privacy-preserving machine learning on data streams, introducing Buffered Linear Toeplitz operators for efficient noise generation in differentially private training.

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

Long-context Attention in Near-Linear Time

Explores HyperAttention, an innovative approach to tackle computational challenges in long-context LLMs, offering significant speed improvements while maintaining performance across various datasets.

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

Sublinear-Time Algorithms in Learning

Explore cutting-edge sublinear algorithms for efficient machine learning, focusing on extroverted approaches and their applications in large-scale data analysis.

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

Randomized Least Squares Optimization and its Incredible Utility for Large-Scale Tensor Decomposition

Explore randomized least squares optimization techniques for large-scale tensor decomposition, focusing on sublinear algorithms and their applications in data analysis.

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