Research

Research initiatives and papers are listed below. Please see my Google Scholar and arXiv profiles for all my research papers.

Theoretical neuroscience ↔ general AI

  • George, Lavin, et al. A generative vision model that trains with high data efficiency and breaks text-based CAPTCHAs. Science 2017. /// paper, blog, code

  • George, Lavin, et al. Cortical Microcircuits from a Generative Vision Model. CCN 2018. /// paper

  • Lavin, et al. Explaining Visual Cortex Phenomena using Recursive Cortical Network. CCN 2018. /// paper

  • Ahmad, Lavin, et al. Unsupervised real-time anomaly detection for streaming data. Neurocomputing 2017. /// paper, code

  • Hawkins, Lavin, et al. Biological and Machine Intelligence. 2016. /// book

Physics-infused ML for science

  • Lavin et al (2021). Simulation Intelligence: Towards a New Generation of Scientific Methods. (in press) /// preprint

  • Peterson & Lavin. Physical Computing for Materials Acceleration Platforms. Matter 2022. /// paper, Accelerate Conference award talk

  • Lavin et al (2021). Multi-scale Neural Operators for flexible spatiotemporal physics. (in press)

  • Lavin (2022). Uncertainty-Aware Human-Machine Teaming in Scientific AI and Simulation. SIAM 2022 Conference on Uncertainty Quantification.

  • Jiang, Lavin, et al (2021). Digital Twin Earth - Coasts: Developing a fast and physics-informed surrogate model for coastal floods via neural operators. /// paper

  • Lutjens, Lavin, et al (2020). Physics-informed GANs for Coastal Flood Visualization. /// journal paper, preprint, demo

Robust, data-efficient computer vision

  • George,… Lavin, et al. Science, 2017

  • Naud & Lavin (2020), Manifolds for Unsupervised Visual Anomaly Detection. /// paper

  • Haney & Lavin. Fine-Grain Few-Shot Vision via Domain Knowledge as Hyperspherical Priors. CVPR Fine-grain Vision Workshop 2020. /// paper, video

  • Lavin. Integration of human-like covert-overt attention with probabilistic neural networks. Neuroscience to Artificially Intelligent Systems (NAISys) 2020. /// abs

  • Lavin (2020), Accelerating Gaussian Processes and Deep Kernel Networks on GPUs. /// video

Systems engineering for AI/ML

  • Lavin et al. Technology Readiness Levels for Machine Learning Systems. Nature Communications, 2022 /// nature open access, preprint

  • Lavin. Towards Systems AI & Decisions Intelligence. 2021 Spring AAAI Symposium. /// preprint

  • Ganju, Lavin, et al (2020). Learnings from Frontier Development Lab and SpaceML – AI Accelerators for NASA and ESA. /// paper

  • Lavin & Renard. Technology Readiness Levels for Artificial Intelligence & Machine Learning. ICML 2020 Workshop on Challenges in Deploying ML Systems. /// paper

Medical ML

  • Lavin. Neuro-symbolic Neurodegenerative Disease Modeling as Probabilistic Programmed Deep Kernels. 2021 International Workshop on Health Intelligence /// paper

  • Naud & Lavin (2020), Manifolds for Unsupervised Visual Anomaly Detection. /// paper

  • Lippoldt & Lavin. Attention-Sampling Graph Convolutional Networks. 2020 BayLearn Symposium.

Others in robotics, optimization, etc.

  • Meinert & Lavin. Multivariate Deep Evidential Regression. /// ‘21 paper, ‘22 paper

  • Lavin. A Pareto Optimal D* Search Algorithm for Multiobjective Path Planning. ICRA Workshop on Planning, Perception, and Navigation for Intelligent Vehicles 2014. /// paper

  • Lavin (2019). Probabilistic programmed Bayesian optimization. /// paper

  • Lavin. Probabilistic programming for data-efficient robotics. PROBPROG 2018.

  • Lavin & Ahmad. Evaluating Real-time Anomaly Detection Algorithms - the Numenta Anomaly Benchmark. ICML 2015. /// paper

  • Lavin (2014). Optimized Mission Planning for Planetary Exploration Rovers. /// paper