Microsoft Research Forum

This talk introduced Phi-3-Vision, an advanced and economical open-source multimodal model. As a member of the Phi-3 model family, Phi-3-Vision enhances language models by integrating multi-sensory skills, seamlessly combining language and vision capabilities.

This panel discussion explored the transformative potential of multimodal models across various domains, including precision health, game intelligence, and foundational models, addressing the challenges and opportunities in integrating diverse data sources for innovative solutions. They delve into the importance of multimodal models, the core challenges faced by researchers, and the future directions for bridging gaps and fostering synergies within the field.

This talk discussed a new kind of computer – an analog optical computer – that has the potential to accelerate AI inference and hard optimization workloads by 100x, leveraging hardware-software co-design to improve the efficiency and sustainability of real-world applications.

This talk discussed teaching language models to self-improve using a preference oracle like GPT-4, framing it as a two-player game to find an optimal policy at a Nash equilibrium, and achieving state-of-the-art win rates against GPT-4 Turbo on benchmarks such as Alpaca-Eval and MT-Bench.

This talk discussed Aurora, a cutting-edge foundation model that offers a new approach to weather forecasting that could transform our ability to predict and mitigate the impacts of extreme events, air pollution, and the changing climate.

This talk discussed how deep learning is enabling us to generate novel and useful biomolecules, enabling researchers and practitioners to better understand biology.

This talk will address the probabilistic nature of AI systems and the challenge of fostering appropriate reliance in human-AI interactions. Discover how Microsoft's initiatives are empowering users to discern when to trust AI recommendations, ensuring accuracy and reliability in AI outputs.

In this keynote, Jacki O'Neill discussed the importance of creating globally equitable generative AI. She addressed the technical and socio-technical challenges that must be tackled to positively transform work futures worldwide.

In this panel discussion, Microsoft researchers discussed the challenges and opportunities of making AI more inclusive and impactful for everyone – from data that represents a broader range of communities and cultures, to novel use cases for AI that are globally relevant.

In this talk, Daniela Massiceti delved into the transformative potential of multi-modal models such as CLIP for assistive technologies. Specifically focusing on the blind/low vision community, the talk explored the current distance from realizing this potential and the advancements needed to bridge this gap.

In this talk, Jiang Bian discussed how Generative AI transforms industries by bridging gaps between AI capabilities and sector needs. He showcased domain-specific foundation models and versatile AI agents, setting new industry standards.

In this talk, Tian Xie introduced MatterGen, a generative model that creates new inorganic materials based on a broad range of property conditions required by the application, aiming to shift the traditional paradigm of materials design with generative AI.

In this talk, Adam Fourney discussed the effectiveness of using multiple agents, working together, to complete complex multi-step tasks. He showcased their capability to outperform previous single-agent solutions on benchmarks like GAIA, utilizing customizable arrangements of agents that collaborate, reason, and utilize tools to achieve complex outcomes.

Chris Bishop shared the vision for how AI for science will leverage AI to model and predict natural phenomena, including the exciting real-world progress being made by the team.

Microsoft researchers shared their advancements in the fields of foundations models, drug discovery, material design and machine learning. They will highlight how deep learning is transforming the natural sciences.

This talk discussed the latest updates on AutoGen – the multi-agent framework for next generation AI applications. This includes milestones achieved, community feedback, new exciting features, and ongoing research and challenges.

This talk explored how metacognition—the psychological capacity to monitor and regulate one's cognitive processes—provides a valuable perspective for comprehending and addressing the usability challenges of generative AI systems around prompting, assessing and relying on outputs, and workflow optimization.

This talk proposed GigaPath, a novel approach for training large vision transformers for gigapixel pathology images, utilizing a diverse real-world cancer patient dataset, with the goal of laying a foundation for cancer pathology AI.

This talk shared recent efforts on building and re-using large collections of expert language models to improve zero-shot and few-shot generalization to unseen tasks.

This talk discussed the potential impacts and challenges of generative AI in a year with over 70 major global elections, with the aim of highlighting strategies to promote societal resilience to malign influence.

Peter Lee discussed how recent developments in AI have transformed the way Microsoft approaches research.

Microsoft AI researchers discussed frontiers in small language models and where AI research and capabilities are headed next.

Dipendra Misra presented a surprising discovery that by merely replacing selected weight matrices in an LLM with their suitable low-rank approximation, you can significantly improve the performance of the LLM, at times by 20-30 percentage points.

This talk discussed timely challenges and ongoing work on evaluating and in-depth understanding of large language foundation models as well as agent platforms built upon such models.

This talk discussed recent reserach in buildng and evaluating AI tools for helping people make better decisions and improve their own capabilities.

Shuxin Zheng presented how his team uses Generative AI to solve a long-standing challenge in structural biology and molecular science - predicting equilibrium distribution for molecular systems.

This talk discussed Kahani, a research prototype that allows the user to create visually stunning and culturally nuanced images just by describing them in their local language.