Investigating vulnerabilities in LLMs; A novel total-duration-aware (TDA) duration model for text-to-speech (TTS); Generative expert metric system through iterative prompt priming; Integrity protection in 5G fronthaul networks. https://msft.it/6040m3pW8
Microsoft Research
智库
Redmond，Washington 291,880 位关注者
We advance science and technology to benefit humanity.
关于我们
At Microsoft Research, we accelerate scientific discovery and technology innovation to empower every person and organization on the planet to achieve more. We do this by bringing together the best minds across diverse disciplines and backgrounds to take on the most pressing research challenges for Microsoft and for society. Our Research Lens We consider research directions through the lens of the positive impact we aspire to create with and for customers, communities, and all of society.
- 网站
-
https://www.microsoft.com/research
Microsoft Research的外部链接
- 所属行业
- 智库
- 规模
- 1,001-5,000 人
- 总部
- Redmond，Washington
- 创立
- 1991
动态
-
Medfuzz tests LLMs by breaking benchmark assumptions, exposing weaknesses vulnerabilities to bolster real-world accuracy. https://msft.it/6042mzAFC
-
-
We’re advancing the Azure Quantum platform to tackle the world’s most pressing challenges. We're announcing the best performing logical qubits on record with Quantinuum and will provide priority access to reliable quantum hardware in Azure Quantum with Atom Computing https://msft.it/6040mzafo. #AzureQuantum #Quantum
-
-
Peter Lee has been recognized by FierceHealthcare for his commitment to innovation, equity and improving lives. Read about his accomplishments at the Fierce 50 list. https://msft.it/6040mzzXQ
-
Measuring AI risk is both an art and a science, especially when adding context to the often fuzzy realm of human concepts. Explore how our expert team of researchers and technologists navigate complex ideas from linguistics to social sciences to create safe AI: https://msft.it/6003mG5lV
-
-
GraphRAG uses LLM-generated knowledge graphs to substantially improve complex Q&A over retrieval-augmented generation (RAG). Discover automatic tuning of GraphRAG for new datasets, making it more accurate and relevant. https://msft.it/6048mKmEM
-
-
The ML with New Compute Paradigms workshop at NeurIPS 2024—which will explore how emerging compute tech can sustainably scale amid demand for GPU-based AI—has extended its call for papers through September 11. Learn more about the submission guidelines. https://lnkd.in/g2tgTE8q
The future of compute hardware for AI is more uncertain than ever. How should AI models change if compute hardware changes? What sort of models could be build with different, for example, analog or neuromorphic hardware? How should we train models that run on analog or photonic chips? How can models deal with noise or very narrow data types? We will discuss this and more at our NeurIPS 2024 workshop in Vancouver! Call for Papers: ML with New Compute Paradigms (MLNCP) Workshop at NeurIPS 2024 We call for workshop papers including but not limited to the following directions: ·???????Advances in machine-learning that benefit from compute paradigms beyond standard digital compute, for example analog, photonic, in-memory, neuromorphic, or quantum compute. ·???????Advances in machine learning methods that can handle challenges such as low bit precision, variability or noise induced by new hardware. Examples include inherently noise-robust models, algorithms to reduce errors in ML arithmetic, or ways to share and distribute models across unique analog hardware. ·???????Advances in machine-learning paradigms that facilitate training and/or inference on new hardware paradigms. This can be general or specific to, for example, generative models for modalities such as image or sequence data. ·???????Surveys and position papers for machine-learning with new compute paradigms. The submission deadline is Aug 29 '24 (Anywhere on Earth)? Please check the workshop website for updates on format and possible deadline extensions. Workshop Website:?www.mlwithnewcompute.com
ML with New Compute Paradigms (MLNCP) at NeurIPS 2023
mlwithnewcompute.com
-
What’s an effective way to store & send digital content into the cosmos? Glass! College student Dexter Greene & research manager Richard Black discuss how tech for storing data in silica is supporting efforts to communicate humanity to extraterrestrials. https://msft.it/6040m1ExC
-
-
Quantum computing uses qubits to store and process information, but current qubits are error-prone. Learn how Microsoft's qubit virtualization system acts like "noise-cancelling headphones," reducing errors and enhancing qubit reliability: https://msft.it/6046l7sJC #AzureQuantum
-
Join us! In this episode, you'll learn about the latest multimodal AI models, advanced benchmarks for AI evaluation and model self-improvement, and an entirely new kind of computer for AI inference and hard optimization. Discover how these research breakthroughs and more can help advance everything from weather prediction to materials design. If you register to watch on our official website, you can: - Access all episodes in the series - Participate in Q&A with the speakers during the episode - Receive the latest news and event updates directly from Microsoft Research Register here: https://msft.it/6005lNYbl Speakers: Jianfeng Gao, Katja Hofmann, Jianwei Yang, Hoifung Poon, John Langford, Francesca Parmigiani, Jiaqi Chu, Corby R., Megan Stanley, Kevin Yang, Mihaela Vorvoreanu
Microsoft Research Forum: Episode 4
www.dhirubhai.net