Name: Brandon Trabucco
Type: User
Company: Machine Learning Department @ CMU
Bio: Musician & Machine Learning PhD Student @ CMU, previously an Undergraduate @ BerkeleyAI
Twitter: brandontrabucco
Location: Pittsburgh, Pennsylvania
Blog: btrabuc.co
Brandon Trabucco's Projects
This is an adversarial attack against the inception v3 network on an ImageNet example.
🔀 Visual Room Rearrangement
An example flask alexa app that reads freeform variable text inputs.
An open source framework for research in Embodied-AI from AI2.
Source for Neural RPS, an android app to play rock paper scissors against a Neural Network trained in real time using tensorflow.
Runs inference and training on a TensorFlow model built in Python on a Java Android App.
This is a c++ implementation of a basic neural network
An experiment using a novel attention mechanism with multiple pathways.
Model using Inception v3 to predict attributes from the deep fashion dataset.
Audio generation using diffusion models, in PyTorch.
A simple package for automatically calculating high order jacobian tensors.
A lightweight package I developed for my own use in homeworks for implementing neural nets.
Code for "Discovering Non-monotonic Autoregressive Orderings with Variational Inference" (paper and code updated from ICLR 2021)
This is a prototype for the attention mechanism used in my research with image captioning.
Non-Sequential Decoding Strategies for Image Captioning
Implements a novel algorithm that I am proposing.
Implements big gan in tensorflow using keras.
Birkhoff-von Neumann Decomposition with a Greedy Birkhoff Heuristic
A collection of helpful image captioning classes.
This is an implementation of dqn training to solve the cartpole control problem.
A pure python cider scorer using numpy, see: https://arxiv.org/abs/1411.5726.
Code for the blog tutorial at https://www.btrabucco.com/blog/training-a-classifier.
Adds SPICE metric to coco-caption evaluation server codes
evaluation scripts to be used with the detailed_captioning repository. Forked from tylin/coco-caption
Implements a serialization script for the Google Conceptual Captions dataset for CVPR 2019.
A tensorflow implementation of constrained beam search that supports arbitrary constraint definitions.
This is an implementation of a convolutional neural network in python using numpy.
A minimalist reinforcement learning package for TensorFlow 2.0.
Effective Data Augmentation With Diffusion Models