Lucy Low

transformer

About Me

Welcome to my website. I am a tech enthusiast, theoretical physicist, and software engineer.

Latest Projects


project name

Deep Neural Network Autoencoder for Data Compression in High Energy Physics at CERN ATLAs

Large Hadron Collider is the world's largest and highest-energy particle collider with the ATLAS detector generating 1 petabyte of raw data and 40 million packets of protons colliding every second. High collision rate 20 MHz means not all events can be stored. A particle physics trigger system selects specific events. The goal is to reduce the size of the data by engineering a compression algorithm for the trigger system.

Autoencoder neural networks were used for data compression and anomaly detection. The two part encoder and decoder system compresses hadron jet event data from 4 to 3 variables. It was trained over a dataset to encode the inputs into a smaller memory space with PyTorch, FastAI Library, ROOT Data Analysis Framework, and CERN ATLAS Docker images. Analysis includes plots and graphs to explain the concepts of invariant mass, purity selection, trigger efficiency, and hadron event reconstruction.

Find out more

project name

Yeezy Taught Me Text Generation - Train Next Character Prediction using Long Short Term Memory Model (LSTM)

LSTM is one of the most commercial AI achievement used in time series prediction, speech recognition, music learning, handwriting recognition, and sign language translations. Yeezy Taught Me is a web application for AI model training and text generation. Trained LSTM model to generate text based on patterns in a given text corpus. Yeezy's Model makes probability predictions of the character that follows the input sequence. Process is repeated in order to generate a character sequence of a given length hence the "text generation" part of the project.

Input file takes input text data. Model saved in IndexedDB database. User inputs Yeezy's model parameters to train model: [model layer size, number of epochs, examples per epoch, batch size, validation split, or learning rate].

Find out more

project name

Stochastic SoundCloud - Lucy’s New Mozart Mixtape. Machine Learning Generative Music using RNN LSTMs

Stochastic SoundCloud uses machine learning to generate melodies as music is an art with a temporal and hierarchical structure. The output musical state is partially determined by the preceding musical state where the concrete musical state n+2 follows after the state n+1. Piano roll representation is a music storage data type where a music piece is represented by a score-like binary valued (0 XOR 1) matrix representing music notes over different time step. Piano roll of each bar and track for generated data is represented as a fixed-size matrix for M tracks.

Music generation with three novel recurrent neural networks (basic, lookback, and attention RNNs) using Magenta from Google's Tensorflow AI. Lookback and Attention RNNs are proposed to tackle the problem of creating the melody’s long-term structure. It was fed with a chord sequence and outputs a Prediction Matrix, which was transformed into a piano roll matrix and into a melody MIDI file. 10 stochastically generated "output.mid" music files are composed and opened up on Mac's Garageband.

Find out more

project name

En Français Si Vous Plait? French-English Natural Language Processing (NLP)

Vision of connecting people through language and advancing a barrier free society for billingual speakers. As a Canadian citizen, ensure respect for English and French as the offical languages of Canada and have equality of status, rights, and privileges.

French-English translations using PyTorch’s Transformer model and paper implementation of "Attention is All You Need”. Led the development of the machine language translation transformer model uses encoder-decoder attention mechanisms in a sequence-to-sequence model. "On the WMT 2014 English-to-French translation task, model establishes a BLEU score of 41.8 after training for 3.5 days on eight GPUs, a small fraction of the training costs of the best models from the literature"

Find out more

project name

Salty Wet Man - Not Suitable for Work (NSFW) Image Classification using VGG16 Neural Network

Inspired by the #Metoo and #Timesup movements in the fight for gender equity in the workplace to create safe economic opportunities for women. Salty Wet Man is a Not Suitable for Work image classification trained on the ImageNet dataset. Defining NSFW material is subjective and the task of identifying these images is non-trivial. Salty-Wet-Man identifies images into two categories: [SFW] positively trained for neutral images that are safe for work [NSFW] negatively trained for pornographic images involving sexually explicit images.

CNN for large-scale image recognition model implementation achieves object recognition functionalities like Content-Based Retrival via Localization, Image Detection, and Segmentation, and Image Region Proposal Cropping.

Find out more

Other Projects

Amazon Developer as Low Apps In Skill Purchasing

Published developer with multiple Amazon Alexa applications on the Alexa Skills Store. Low Apps integrated with Voice User Interface (VUI) and In-Skill Purchasing (ISP). ISP allows developers to enrich in-skill experiences, drive deeper customer engagements.

Search Alexa Skills Store for Dad Jokes, Meditation Zen, Buddha365, Daily Shakespeare, or Daily Stoic.

Juypter Notebook: Data Analysis of Udacity Student Engagements

Data Analysis with Python and Jupyter Notebook on Udacity Student Engagement data with CSV data cleansing on files enrollments.csv, daily_engagement.csv, project_submissions.csv. Correlations between Udacity project completion rates and the following factors: minutes engaged, lessons completed, and days visiting the classroom.

Data analysis here.

Hack Princeton - Microsoft Word + Audio Parsing Free

Developed a Node.JS application using the Word Javascript API for Hack Princeton. User records audio from their browser and the data is automatically parsed into Microsoft Word using the JS add-on.

Record Audio

Juypter Notebook: Data Analysis of New York Subway and Weather Data. Working with 2D data with numpy and pandas.

Data Analysis with Python and Jupyter Notebook on New York Subway and Weather Data. Compared NYC subway map to find high ridership numbers in clusters from longitude from - 73.95 to -74.03 and latitude from 40.70 to 40.79.

Data analysis here.

B00m-h3adsh0t! - Neural Network Aimbot for FPS games with Custom Training Mode ︻デ═一 Free

Developed a Neural Network Aimbot for FPS games with custom training mode written in C++ providing a fast and efficient framework with scripting support. Includes customizable predictions and dynamic speed settings. Recognizes game objects in a certain range, then aims at the objects using game physics by hooking into the FPS game engine to use game data to auto-aim without altering gaming files.

Download Now.

Juypter Notebook: Gapminder Data Analysis

Data Analysis with Python and Jupyter Notebook on Gapminder data with information on employment rates (%), life expectancy (years), GDP/capita (US$ and inflation adjusted), primary school completion (% of boys), or primary school completion (% of girls) data collection.

Data analysis here.

More on Github