Simons Institute Courses

Extractors - Building a Quantum Computer with QLDPC Codes

Discover how extractors enable fault-tolerant quantum computing with QLDPC codes, transforming memory blocks into computational units for scalable quantum architectures.

• YouTube
• 56 minutes
• Self-Paced
• Free Video

Learning Stabilizers with Noise

Explore quantum error correction through the Learning Stabilizers with Noise problem, examining random stabilizer codes and their decoding challenges in noisy environments.

• YouTube
• 30 minutes
• Self-Paced
• Free Video

Quantum Fault Tolerance with Constant-Space and Logarithmic-Time Overheads

Discover groundbreaking quantum fault tolerance protocols achieving constant-space overhead and logarithmic-time complexity using quantum locally testable codes and novel distillation techniques.

• YouTube
• 38 minutes
• Self-Paced
• Free Video

Quantum Computational Advantage with Polyloglog-Depth Noisy Circuits

Discover how fully-quantum fault tolerance protocols achieve superpolynomial quantum advantage using polyloglog-depth noisy circuits under standard local stochastic noise models.

• YouTube
• 37 minutes
• Self-Paced
• Free Video

Quantum Versions of the Karp-Lipton Theorem

Explore quantum analogues of the Karp-Lipton Theorem, examining how quantum advice affects computational complexity hierarchies and the collapse conditions for NP and counting classes.

• YouTube
• 1 hour 6 minutes
• Self-Paced
• Free Video

One-Shot Signatures

Discover how one-shot signatures from quantum cryptography provide surprising solutions to secure communication challenges and quantum state transmission problems.

• YouTube
• 1 hour 4 minutes
• Self-Paced
• Free Video

Beyond Preferences in AI Alignment: Towards Richer Models of Human Reasons and Decisions

Explore the limitations of preference-based AI alignment and discover a framework for modeling human goals and norms that enables AI assistants to adapt to context-specific values while maintaining safety standards.

• YouTube
• 46 minutes
• Self-Paced
• Free Video

Adam and Friends via Duality

Explore how optimization methods can be derived and analyzed through duality frameworks, highlighting unique similarities in the dual space compared to the primal.

• YouTube
• 1 hour
• Self-Paced
• Free Video

Detection and Steering in LLMs using Feature Learning

Explore a method for detecting semantic concepts in LLM activations and steering models toward desired outputs, with applications in hallucination detection, harmfulness identification, and concept-specific generation.

• YouTube
• 1 hour 3 minutes
• Self-Paced
• Free Video

Weak-to-Strong Generalization Even in Random Feature Networks, Provably

Explore the phenomenon of Weak-to-Strong Generalization in random feature networks, where a student model outperforms its teacher despite being trained only on teacher-labeled data.

• YouTube
• 1 hour
• Self-Paced
• Free Video

Generalization and Overfitting in Two-Layer Neural Networks

Delve into the theoretical understanding of generalization in overparametrized neural networks, exploring concepts like implicit bias, benign overfitting, and feature learning through statistical physics approaches.

• YouTube
• 59 minutes
• Self-Paced
• Free Video

Local Minima of the Empirical Risk in High Dimension

Explore the mathematical analysis of high-dimensional empirical risk minimization, focusing on Gaussian data models and the Kac-Rice formula to characterize local minima and derive sharp asymptotics for estimation errors.

• YouTube
• 1 hour 1 minute
• Self-Paced
• Free Video

A Statistical Theory of Contrastive Pre-training and Multimodal Generative AI

Delve into the statistical foundations of contrastive pre-training and multimodal generative AI with Song Mei from UC Berkeley in this deep learning theory lecture.

• YouTube
• 1 hour 6 minutes
• Self-Paced
• Free Video

Emergence and Scaling Laws for SGD Learning and Learning Compositional Functions with Transformers

Explore the theoretical foundations of deep learning, focusing on SGD scaling laws and transformers' ability to learn compositional functions with Princeton's Jason Lee.

• YouTube
• 54 minutes
• Self-Paced
• Free Video

Theory of Scaling: A General Framework for Scale-aware Training

Explore a theoretical framework for efficient learning at large scale in deep neural networks, with insights on deriving scale-aware training rules that ensure stability and optimal performance.

• YouTube
• 1 hour 4 minutes
• Self-Paced
• Free Video