Skip to content
The Chat Archive
Perplexitysonar· Perplexity LMArena Search Arena 24K

Willie Neiswanger research

From a Chatbot Arena battleThe same person sent this prompt to two anonymous models at once. Compare it with Geminigemini-2.5-pro-exp-03-25-wo-search’s answer.
  1. User

    Willie Neiswanger research

  2. Assistant

    Willie Neiswanger is an Assistant Professor of Computer Science at the University of Southern California (USC), affiliated with the Viterbi School of Engineering and the School of Advanced Computing. His research lies at the intersection of machine learning, decision making, generative AI, and AI-for-science, with a focus on developing methods for efficient optimization and experimental design in costly real-world settings where resources are limited[2][3][5].

    Research Focus and Contributions

    • Efficient Optimization and Experimental Design: Neiswanger develops machine learning methods to optimize and design experiments efficiently, particularly in scenarios where data collection or experimentation is expensive or resource-constrained. This includes work on active learning, Bayesian decision making, and reinforcement learning[2][3].

    • Uncertainty Quantification: He works on methods to quantify uncertainty in machine learning models, which is crucial for reliable decision making and scientific applications. His research includes uncertainty quantification using martingales for misspecified Gaussian processes and calibration techniques[2][6].

    • Bayesian Optimization and AutoML: Neiswanger has contributed to scalable Bayesian optimization techniques, including applications to neural architecture search and automated machine learning (AutoML). He has developed and maintained software libraries such as Dragonfly for scalable Bayesian optimization and other tools for uncertainty quantification and multilevel optimization[2][6][7][8].

    • Distributed Algorithms and Scalable Machine Learning: He has worked on distributed inference and scalable probabilistic modeling to handle large datasets and complex models efficiently[2][4][6].

    • Applications in Science and Engineering: His methods are applied to problems in physical sciences, biology, materials science, high energy physics, and machine learning systems, demonstrating a strong interdisciplinary approach[2][3][4].

    Academic Background

    • PhD in Machine Learning from Carnegie Mellon University, advised by Eric Xing, with collaborations involving Jeff Schneider and Barnabas Póczos[2][3][5].
    • Postdoctoral research in Computer Science at Stanford University, working with Stefano Ermon and affiliated with the Stanford AI Lab, StatsML Group, and SLAC National Accelerator Laboratory[2][3][4].
    • Undergraduate degree in Applied Mathematics and Computer Science from Columbia University, where he worked with Chris Wiggins and Frank Wood[2][3][5].

    Selected Publications and Projects

    • "Neural Architecture Search with Bayesian Optimisation and Optimal Transport" (NeurIPS 2018)
    • "Myopic Posterior Sampling for Adaptive Goal Oriented Design of Experiments" (ICML 2019)
    • "BANANAS: Bayesian Optimization with Neural Architectures for Neural Architecture Search" (AAAI 2021)
    • Research on uncertainty quantification, Bayesian optimization, and scalable probabilistic modeling published in top machine learning conferences[6][7].

    Software and Tools

    • Maintains several open-source libraries on GitHub related to his research, including:
      • Dragonfly: A Python library for scalable Bayesian optimization.
      • Uncertainty Toolbox: Tools for predictive uncertainty quantification.
      • Naszilla: A library for neural architecture search.
      • TorchUQ: A PyTorch-based library for uncertainty quantification[8].

    Contact and Recruitment

    Neiswanger is actively recruiting PhD students and postdoctoral researchers interested in machine learning, decision making, generative AI, and AI-for-science. He can be contacted via his USC email [EMAIL_ADDRESS][2][3].


    In summary, Willie Neiswanger is a prominent researcher specializing in machine learning methods for optimization, uncertainty quantification, and decision making, with significant contributions to Bayesian optimization and scalable machine learning, applied across scientific and engineering domains.

Responses were produced in different, uncontrolled contexts (system prompts, settings, prior turns). This is not a benchmark.