Hi, I'm
a computer scientist from Ecuador with experience in AI/ML.
I recently completed a master's degree in AI/ML at the University of Birmingham, fully funded by the Google DeepMind International Scholarship.
My current research interests include how to create more robust AI
systems capable of adapting, focusing, or reusing knowledge. I'm intrigued by the question of why current AI struggles to preserve and
reuse knowledge. Currently, I'm keen on deepening my understanding exploring topics such as:
I have experience in computer vision and deep learning for
performing tasks such as classification, detection,
segmentation, and image generation (using StyleGAN).
Bisimulation Prioritized Experience Replay (BPER) introduces a novel approach to enhance value-based reinforcement
learning by incorporating a bisimulation metric to prioritize behaviorally relevant transitions.
By balancing traditional TD-error with state behavior similarities, BPER improves data diversity
and learning efficiency, as demonstrated in a 31-state Grid World and pixel-based environments.
During a single step forward, we compute the covariance tensor
and factorize it by Tucker decomposition to generate kernels
for a CNN. The kernels were then plugged into the CNN
(CovarianceNet) for classification.
We reproduced StyleGAN embedding algorithms, adding further
experimentation such as inpainting, super-resolution,
colorization, morphing, style transfer, and expression transfer.
This project aims to fine-tune a BERT-based model, called HangmanNet, for character-level masked language modeling to
play the Hangman game. The model predicts hidden characters by leveraging bidirectional context from both sides of a
word. Additional techniques, such as incorporating prior character frequency and custom data collators, were explored to
improve accuracy, particularly for shorter words.
This project compares and implements key model-free Reinforcement Learning (RL)
algorithms, focusing on clarity over complexity. Using online resources for
guidance, it aims to provide accessible, easy-to-understand implementations,
serving as an educational tool for those exploring RL's fundamental concepts.
We developed a multistream Automatic License Plate Recognition
(ALPR) system by assembling an Nvidia deepstream pipeline,
training 3+ models with the Nvidia TAO toolkit, and
periodically sending data streams to a Kafka Apache
for further use cases.
We developed a system to segment and count grapes bunches using
Mask RCNN and DeepSort Tracker.
After that, we extrapolate the counting information to
satellite images to generate heat maps that show the number of
grapes per parcel in a yield.
A compilation of my notes during the second semester of the master program in Artificial Intelligence and Machine Learning. I covered
the following topics: Natural Language Processing, Computational Vision and Imaging, and Current Topics of AI/ML.
A compilation of my notes during the first semester of the master program in Artificial Intelligence and Machine Learning. I covered
the following topics: Machine Learning, Deep Learning, and Mathematics for AI/ML.
Introduction to Pandas library. This short introduction includes 8 jupyter notebooks to understand the
fundamentals and minimum requirements for an job interview.