How to Compress Garbled Circuit Input Labels, Efficiently
Hanjun Lee discusses improvements in garbled circuits for efficient label compression, enhancing computational efficiency and reducing communication overhead.
Differentially Private Synthetic Data without Training
The talk discusses the concept of private evolution, a method for generating differentially private synthetic data using foundation models without training, ensuring privacy while maintaining data utility.
Celebrating Susan Dumais: Reflections on a Legacy of Research and Collaboration | Plenary Session
The session celebrates Sue's contributions to research, mentorship, and interdisciplinary collaboration, highlighting her impact on AI, HCI, and information retrieval.
The Assistant: Situated Interaction Project (2012)
The transcript involves multiple interactions about scheduling and meeting updates, focusing on coordinating with Eric and managing his appointments.
The AI Revolution in Medicine, Revisited: An Introduction
The discussion explores the transformative impact of AI in medicine, emphasizing its potential to improve healthcare and challenge personal relationships.
World and Human Action Models towards gameplay ideation (Supplementary Video 1)
The video discusses the use of a generative model, Wham, to enhance creative ideation in game development through consistency, diversity, and persistency.
LLMs vs. Torch 1.5: Why Your Code Assistant Can't Keep Up
Language models struggle with frequent version changes in code libraries, impacting their ability to generate accurate code.
Using LLMs for safe low-level programming | Microsoft Research Forum
The talk discusses two projects using large language models (LLMs) to enhance memory safety in C and Rust programming, focusing on automating code annotations and fixing compilation errors.
AutoGen v0.4: Reimagining the foundation of agentic AI for scale and more | Microsoft Research Forum
AutoGen 0.4 is a redesigned open-source framework for multi-agent AI applications, offering a flexible, scalable architecture and enhanced developer tools.
Belief state transformers | Microsoft Research Forum
The Belief State Transformer architecture enhances language models by combining forward and backward encoders, improving self-evaluation and prediction capabilities.
Magma: A foundation model for multimodal AI Agents | Microsoft Research Forum
Magma is a foundation model designed for multimodal AI agents, capable of understanding and interacting with both digital and physical environments.
Chimera: Accurate synthesis prediction by ensembling models with... | Microsoft Research Forum
Microsoft and Novartis are using AI to accelerate drug discovery by improving retrosynthesis, reducing time and cost.
AI for Precision Health: Learning the language of nature and patients | Microsoft Research Forum
The panel discusses the transformative potential of generative AI in healthcare, focusing on precision medicine, continuous health monitoring, and the integration of AI in clinical settings.
Keynote: Multimodal Generative AI for Precision Health | Microsoft Research Forum
The video discusses the use of generative AI in precision healthcare, focusing on improving drug development and patient care through AI-driven insights from real-world data.
Attestations over TLS 1.3 and ZKP
Sophia Chile discusses creating privacy attestations over TLS 1.3 using zero-knowledge proofs, focusing on privacy-preserving technologies and ethical considerations.
Accelerating Multilingual RAG Systems
The talk focuses on accelerating multilingual retrieval, relevance, and generation evaluation systems, emphasizing the need for research beyond English to include more languages.
Pronouns in the Workplace: Learning Inclusive Software Design from Real-World Experiences
The video discusses the complexities of pronoun sharing in the workplace and offers recommendations for software design to support diverse gender identities.
Culturally Aware Machines: Why and when are they useful?
The talk discusses the cultural biases in AI models and the importance of cultural awareness in AI applications.
Embodied AI Workshop at CVPR 2024
The panel discusses advancements in embodied AI, focusing on integrating perception and action in robotics.
GASP: Gaussian Avatars with Synthetic Priors
GASP is a novel method for creating realistic, animatable avatars using single-camera data and synthetic priors.
A Closer Look at Falcon
The presentation discusses the Falcon signature scheme, a post-quantum cryptography algorithm, and its security challenges and modifications for provable security.
MSR Cryptography Talk Series: Quantum Lattice Enumeration in Limited Depth, Fernando Virdia
The talk by Fernando Veria focuses on assessing the concrete threat posed by a specific quantum algorithm on new standardized cryptography, particularly in the context of post-quantum cryptography and the challenges of quantum enumeration under limited depth constraints.
Enhancing Security of Bluetooth Secure Connections via Deferrable Authentication
The video discusses the security and vulnerabilities of the Bluetooth protocol stack, focusing on various attacks and potential mitigations.
Improving the Security of United States Elections with Robust Optimization
The presentation by Brad Stur focuses on using mathematical optimization to improve the security and public confidence in vote counting, specifically through enhancing logic and accuracy testing for voting machines.