I'm a PhD student in computational neuroscience at Columbia. I'm interested in understanding how high-dimensional neural representations support complex behaviors and how these capabilities are learned. I graduated from Stanford in 2019 with an undergraduate major in math and a master's in computer science. At Stanford, I worked with Shaul Druckmann on modeling neural dynamics in mouse anterior lateral motor cortex under optogenetic perturbations, and with Surya Ganguli on developing models of representations in the retina and visual cortex. I've also spent time in Jay McClelland's lab at Stanford, using deep learning models to study memory and visual attention, and at Cerebras Systems, a machine learning hardware startup.
Publications
F. Li, J. Lindsey, E. Marin, N. Otto, M. Dreher, G. Dempsey, I. Stark, A. S. Bates, M. W. Pleijzier, P. Schlegel, A. Nern, S. Takemura, T. Yang, A. Francis, A. Braun, R. Parekh, M. Costa, L. Scheffer, Y. Aso, G. S. X. E. Jefferis, L. F. Abbott, A. Litwin-Kumar, S. Waddell & G. M. Rubin (2020). The connectome of the adult Drosophila mushroom body: implications for function. bioRxiv: 2020.08.29.273276.
Link
J. Lindsey & A. Litwin-Kumar. Learning to Learn with Feedback and Local Plasticity (2020). arXiv: 2006.09549. Link
J. Lindsey, S. Ocko, S. Ganguli, & S. Deny (2019). A Unified Theory of Early Visual Representations from Retina to Cortex through Anatomically Constrained Deep CNNs. International Conference on Learning Representations (oral presentation). Link
S. Ocko, J. Lindsey, S. Ganguli, & S. Deny (2018). The Emergence of Multiple Retinal Cell types through Efficient Coding of Natural Movies. Neural Informational Processing Systems. Link
M. Jain & J. Lindsey (2018). A Neural Network Model of Complementary Learning Systems. CogSci 2018 proceedings (oral presentation). Link
M. Jain & J. Lindsey (2018). Semiparametric Reinforcement Learning. ICLR, Workshop Track. Link
J. Lindsey (2017). Pre-Training Attention Mechanisms. NeurIPS 2017 Workshop on Cognitively Informed Artificial Intelligence. Link
A computer chess player. The difficulty can be adjusted by setting the time the computer takes for each move. Uses standard techniques for chess AIs (e.g. minimax algorithm with pruning, piece-square tables, etc.) and has been optimized through a repeated genetic algorithm simulation. Rough sense of its playing level: it beats me most games, but I know people who can beat it most games.
2D Physics Engine
A 2D physics simulation program written in Java. Supports collisions between convex polygons and circles, remote attractive/repulsive forces (e.g. universal gravitation), "Earth-style" downward gravitation, springs (damped and simple), and friction. Comes with 6 preconfigured demos of the engine's various capabilities.
Conductors Simulation
Simulates the flow of charges within a conductor. Users can set the charge of the conductor, adjust its shape, and subject it to electric fields produced by external charges. Users can also toggle between viewing the charges particles within the conductor and viewing the charge density along its surface. Can be used to demonstrate various concepts in electromagnetism, including Faraday cages and the fact that charge tends to accumulate at sharp points.
Trendify KnowledgeBase
A web application build using Telescope / Meteor that displays and ranks user-inputted entries to the Trendify "Knowldgebase" (essentially a list of relevant resources for Trendify's clients).
3D Combat Billiards
A multiplayer 3D flying/combat game written in java. Graphics courtesy of LWJGL. Can be played by computers connected to the same network. Incorporates a simple 3D physics engine, multi-thread programming on the client side, some basic networking, and quaternion-based flight mechanics.
Logic Circuits
A GUI application that enables logic circuit simulation. By default, supports 'and,' 'or,' and 'not' gates. Allows user-defined gates to be saved and loaded as "black boxes," which enables arbitrarily complex simulations. Entire circuits can also be saved to and loaded from a file.
Flashcard Revolution
A simple GUI application that allows users to create digital "flashcards" and use them to study. Flashcards can be tagged and saved to / loaded from files. Provides some simple statistics on the user's historical performance.
Great Expectations Quiz
My first ever programming project (from back in 9th grade)! A program that presents the user with a series of scenarios and then purports to tell him/her which character from Great Expectations he/she most closely resembles. Warning: the program is more frank than it is accurate. Be prepared to have your feelings hurt.
Chat Bot
A limited and unfinished/unpolished experiment with natural language processing, powered by OpenNLP and WordNet. It can process simple statements (e.g. "Helen likes big dogs") and then respond to some simple, related questions (e.g. "What does Jim like?"). Warning: only understands stilted (and sometimes incorrect) grammar!
InfoSnap Automated Tests
Proprietary automated testing software developed for InfoSnap, a company that facilitates the process of registration and enrollment management for schools. Used a Python scripts, Selenium, and SauceLabs to automatically test updates to InfoSnap's web forms on all major browser/OS combinations.
Recurrent Attention Model
Performs image classification by sequentially selecting and processing small "glimpse" windows from the image. Attention mechanism is non-differentiable and thus must be trained using reinforcement learning -- specifically, the REINFORCE algorithm. Based on the architecture described in "Recurrent Models of Visual Attention" (Mnih et al., 2014). Model implemented in TensorFlow.
Autoencoding with Adversarial Networks
Variational autoencoder using a similarity metric learned by a generative adversarial network. Based on the architecture described in "Autoencoding beyond pixels using a learned similarity metric" (Larsen et al., 2016). Model implemented in TensorFlow.
Semantic Cognition
Trains a feedforward network to answer questions about whether certain subjects (e.g. "canary") have certain properties (e.g. "can fly"). After this training, performs hierarchical clustering on the subjects based on their similarity in network's feature space, revealing the formation of abstract categorizations. Written in MATLAB, from scratch. Based on the model described in "The Parallel Distributed Processing Approach to Semantic Cognition" (McClelland and Rogers, 2003).
Recurrent Backpropagation
Implements backpropagation in a recurrent neural network. Written in MATLAB, from scratch. For educational purposes!
Continuous Q Learning
General-purpose algorithm for reinforcement learning in environments with continuous action spaces. Compatible with the MuJoCo environments in the OpenAI Gym. Based on the "Normalized Advantage Functions" algorithm", described in "Continuous Deep Q-Learning with Model-based Acceleration" (Gu et al., 2016).
Pre-training Attention Mechanisms
Presented at the NIPS 2017 Workshop on Cognitively Informed Artificial Intelligence.
Recurrent neural networks with differentiable attention mechanisms have had
success in generative and classification tasks. We show that the classification
performance of such models can be enhanced by guiding a randomly initialized
model to attend to salient regions of the input in early training iterations. We further
show that, if explicit heuristics for guidance are unavailable, a model that is pretrained
on an unsupervised reconstruction task can discover good attention policies
without supervision. We demonstrate that increased efficiency of the attention
mechanism itself contributes to these performance improvements. Based on these
insights, we introduce bootstrapped glimpse mimicking, a simple, theoretically
task-general method of more effectively training attention models. Our work draws
inspiration from and parallels results on human learning of attention.
Pre-training Attention Mechanisms
Recurrent neural networks with differentiable attention mechanisms have had
success in generative and classification tasks. We show that the classification
performance of such models can be enhanced by guiding a randomly initialized
model to attend to salient regions of the input in early training iterations. We further
show that, if explicit heuristics for guidance are unavailable, a model that is pretrained
on an unsupervised reconstruction task can discover good attention policies
without supervision. We demonstrate that increased efficiency of the attention
mechanism itself contributes to these performance improvements. Based on these
insights, we introduce bootstrapped glimpse mimicking, a simple, theoretically
task-general method of more effectively training attention models. Our work draws
inspiration from and parallels results on human learning of attention.
Models of the Retina
Analysis of the computational constraints that give rise to the distinctive receptive field patterns in the retina. In progress.
3D Printed Rover Bot
A 3D printed robot created using the MakerBot printer. Contains an
ultrasonic sensor on the front that allows it to detect nearby objects.
I've programmed it to find the nearest object, approach it, and snap
menacingly with its claws. See it in action here.
