trdelgado Goto Github PK

ai4all

Notebook lesson plans for univariate, multivariate, and logistic regression. Notebook use is for AI4ALL students to learn the math and theory before build regression models from scratch and finally using sklearn.

ai4all-pilot-program

Pilot Program for AI4ALL's Open Learning Program

aind-american-sign-language-recognizer

Built an American Sign Language (ASL) word recognizer system using Hidden Markov Models achieving a WER score of 0.5449438202.

aind-cargo-logistic-planning

Defined a group of air cargo domain problems in classical PDDL (Planning Domain Definition Language). Experimented with various domain-independent (bfs, dfs, uniform, depth limited, etc) and informed A-star (with custom heuristics) searches to find optimal logistic scheduling for cargo.

aind-isolation

Built an adversarial search game-playing agent using AlphaBeta pruning and time-limited iterative deepening. Constructed custom isolation heuristics to beat predefined agents with the best average winning rate of 71.4%.

aind-sudoku

Implemented a diagonal Sudoku problem solver using elimination, only choice, constraint propagation, and naked twins techniques.

deeprl-grounding

Train an RL agent to execute natural language instructions in a 3D Environment (PyTorch)

deeptype

Design, evolve, and train neural type systems.

dlnd-bikeshare-prediction

Built and trained a neural network from scratch to predict the number of bike-share users on a given day.

dlnd-facial-generation

Uses a convolutional generative adversarial network to generate realistic faces.

dlnd-generate-tv-scripts

Uses deep recurrent LSTM cells to generate new Simpsons dialog from prior scripts.

dlnd-image-classification

Classify images from the CIFAR-10 dataset using a convolutional neural network.

dlnd-language-translation

Sequence-to-sequence recurrent LSTM English to French translation generator.

enron-people-of-interest

Built a predictive algorithm to identify Enron employees who may have committed fraud based on the public Enron financial and email dataset (under construction).

gym

A toolkit for developing and comparing reinforcement learning algorithms.

hadoop-and-mapreduce

Sample MapReduce python scripts using Hadoop to parallel process cvs data.

hands-on-data-analytics-for-beginners-with-google-colaboratory-video-

Hands-On Data Analytics for Beginners with Google Colaboratory [Video], published by Packt

intro-machine-intelligence

Machine Intelligence Workshop

jupyter-notebook-introduction-data-science-intro-000

learn-co-sandbox

linearalg-lib

Linear Algebra Library

machine-intelligence

Introductory course covering the landscape of Machine Learning

mlnd-boston-housing

Used aggregated Boston housing data to build an optimal model to estimate the best selling price for a clients' home. Used grid search and cross validation to find and evaluate an optimal model with an R-squared score over 0.7.

mlnd-capstone-project

Under construction

mlnd-customer-segments

Performed customer segment analysis for a wholesale distributor to make informed business decisions. Analyzed data using PCA, Gaussian Mix Models, and A/B testing to segment customers into distinct ‘Hotel/Restaurant/Cafe’ and ‘Retailer’ categories to improve delivery mechanisms.

mlnd-deep-learning-projects

Deep Learning tutorial projects for Udacity's Deep Learning course.

mlnd-digit-recognition

Under construction

mlnd-finding-donors

Employ several supervised algorithms to accurately model individuals' income using data collected from the 1994 U.S. Census. Developed an optimal model using a linear regression F1 score of 0.8877.

mlnd-smartcab

Implemented reinforcement Q-learning using Pygame to construct a simulated autonomous smartcab with A+ safety and reliability ratings. Stored state-action pairs with sensory inputs and directions to represent the environment.

mlnd-student-intervention

Used several supervised models on students’ graduation rates quantifying whether intervention is necessary. Analyzed trade-offs between F1 scores, training and prediction times to reduce costs for the local school district. Developed an optimal model using an SVM with F1 score of 0.8077.