I’m a software engineer specializing in artificial intelligence and machine learning, with a background in computational mechanics and spacecraft engineering. I'm data driven, passionate about learning, and strive to build a better world with technology.
We're building general artificial intelligence. My focuses:
Numenta aims to reverse-engineer the neocortex for machine intelligence. My role was mainly to prototype algorithms from raw theory, test and validate them through experiments, and develop them into production quality code.
Applied new approach to designing a re-entry system for on-demand return of scientific payloads from the International Space Station.
Provided energy company with a new approach (k-NN clustering) to better classify customers based on true energy usage patterns and investigate energy efficiency traits (see Publications below).
Designed experiments and testbed to investigate the use of nitrous oxide in hybrid propulsion engines. Ran rocket engine firing tests and analyzed results across a variety of performance metrics.
I dropped out. Why? I found most of the curriculum unfulfilling, and dedicating myself to learning engineering management skills through experience to be far more valuable.
Main languages: Python and C++. Main tools: NumPy, Eigen, gtsam, git, ROS. Ping me for more granularity.
Implemented varieties of CNN + LSTM architectures to investigate temporal dependencies in video data, specifically for predicting speed and steering for a self-driving car.
Integrated Pareto-optimality into traditional A* and D* search algorithms to provide improved methods for multi-objective path planning. I showed strong improvements over traditional methods across all optimization metrics in simulated environments, including in lunar and Mars terrain case studies.
Lunar rover project manager and lead of systems engineering.
For the task of classifying particle collision events as either signal (Higgs boson decay) or background noise, I implemented a gradient boosted classifier that performed only 4.63% behind the winning score, and almost 2x the score of the baseline naive Bayes model.
"Finite Element-based Structural Optimization of Large System Models Under Buckling Constraints"
Drawing inspiration from systems neuroscience, we introduce a probabilistic generative model for vision in which message-passing based inference handles recognition, segmentation, and reasoning in a unified way. The Recursive Cortical Network (RCN) outperforms state-of-the-art deep learning models on challenging text recognition benchmarks, while being 300x more data efficient. See our "Common sense, cortex, and CAPTCHA" blog post for more. Also, I was lead dev for the reference implementation.
Abstract: Many probabilistic programming languages decouple modeling and inference: the user specifies a model, and inference is delegated to an inference engine based on black-box MCMC or variational techniques. This approach has two limitations that inhibit its use in robotics. First, many robotics applications require real-time performance. Inference algorithms thus need to be optimized for each application. Second, many applications of robotics learning require combining probabilistic inference with optimization, in settings where evaluating the objective function requires costly real-world interactions. Modeling and inference must thus be integrated with Bayesian optimization techniques. We propose a design for a domain-specific probabilistic programming language for data-efficient robotics. The design prominently features (i) programmable inference, (ii) modular and structured optimization, and (iii) explicit uncertainty quantification. We argue that such a DSL could be useful for tasks such as Bayesian optimization for grasping, and learning policies for control.
Abstract: Understanding the information processing roles of cortical circuits is an outstanding problem in neuroscience and artificial intelligence. The theoretical setting of Bayesian inference has been suggested as a framework for understanding cortical computation. Based on a recently published generative model for visual inference (George et al., 2017), we derive a family of anatomically instantiated and functional cortical circuit models. In contrast to simplistic models of Bayesian inference, the underlying generative model's representational choices are validated with real-world tasks that required efficient inference and strong generalization. The cortical circuit model is derived by systematically comparing the computational requirements of this model with known anatomical constraints. The derived model suggests precise functional roles for the feedforward, feedback and lateral connections observed in different laminae and columns, and assigns a computational role for the path through the thalamus.
Abstract: The connectivity and information pathways of visual cortex are well studied, as are observed physiological phenomena, yet a cohesive model for explaining visual cortex processes remains an open problem. For a comprehensive understanding, we need to build models of the visual cortex that are capable of robust real-world performance, while also being able to explain psychophysical and physiological observations. To this end, we demonstrate how the Recursive Cortical Network (George et al., 2017) can be used as a computational model to reproduce and explain subjective contours, neon color spreading, occlusion vs. deletion, and the border-ownership competition phenomena observed in the visual cortex.
Models of neocortex largely consider the ventral pathway of the vision system, yet integration with the parallel dorsal pathway is necessary for sensorimotor processing. I present a ventral + dorsal computational model with an updated canonical microcircuit.
Propose a novel anomaly detection algorithm based on Hierarchical Temporal Memory (HTM), an online sequence memory algorithm derive from the neocortex.
An online textbook on Hierarchical Temporal Memory with Numenta.
Paper and oral presentation at 2015 IEEE International Conference on Machine Learning Applications
Journal paper in Energy Efficiency