UAB's Data Science Club
By William Monroe
UAB's Data Science ClubNov 05, 2019
Machine Learning Interpretability with Patrick Hall
This episode of the UAB Data Science Club, we are interviewing Patrick Hall. He has written the book on Machine Learning Interpretability, and is the Senior Director of Product at https://www.h2o.ai/.
Patrick guides us through a Disparate Impact Analysis, and we discuss AI security, fairness, and Asimov’s rules of robotics.
This is the notebook we looked at with Patrick Hall
https://nbviewer.jupyter.org/github/jphall663/interpretable_machine_learning_with_python/blob/master/dia.ipynb
Patrick Hall’s Machine Learning Interpretability Book
https://www.h2o.ai/oreilly-mli-booklet-2019/
Warning Signs: The Future of Privacy and Security in an Age of Machine Learning
https://fpf.org/wp-content/uploads/2019/09/FPF-Indecent-Exposure-Report-Final-digital.pdf
Fairness, Accountability, and Transparency in Machine Learning
https://www.fatml.org/
IBM AI Fairness 360 Toolkit
http://aif360.mybluemix.net/
AllenNLP Interpret: A Framework for Explaining Predictions of NLP Models
https://arxiv.org/abs/1909.09251
Data Visualization using Seaborn
This week we are looking at the first 3 notebooks in the kaggle data visualization track:
https://www.kaggle.com/learn/data-visualization
Visualization is super important to the data scientist, since these are the tools we must use to communicate findings and tell stories with the data we are analyzing.
UAB Data Science Club #24: Feature Extraction: LLE and T-SNE
Today, Ravi and I cover two more feature extraction techniques, Locally Linear Embedding (LLE) and t-distributed Stochastic Neighbor Embedding (T-SNE). We are building on the notebook we started in video #23, so check that one out if you haven't already. https://towardsdatascience.com/feature-extraction-techniques-d619b56e31be
UAB Data Science Club #23: Feature Extraction: PCA, ICA, and LDA
In this video, Ravi and I go over some basic feature extraction and dimensionality reduction techniques.
Here is the tutorial we used. https://towardsdatascience.com/feature-extraction-techniques-d619b56e31be
Next week we will do the second half of this article.
Starting, Stopping, and Resuming Training in Keras
Today, Ravi and I are using Python and the keras library to explore training convolutional neural networks with starting, stopping, and resuming training.
We are going through https://www.pyimagesearch.com/2019/09/23/keras-starting-stopping-and-resuming-training/?utm_source=facebook&utm_medium=ad-23-09-2019&utm_campaign=23+September+2019+BP+-+Traffic&utm_content=Default+name+-+Traffic&fbid_campaign=6122406376646&fbid_adset=6122407684846&utm_adset=23+September+2019+BP+-+Email+List+-+United+States+-+18%2B&fbid_ad=6122407685046
We will be using the environment we created in the first Data Science Club video: https://youtu.be/Ew6kAP_6PBI, so if you haven't already, do that one first!
Please Like and Subscribe if you would like to get these videos as we release them.
Feel free to ask any questions here or in office hours
UAB Data Science Club #21: My First Generative Adversarial Network
Today, Ravi and I are using Python and the keras library to explore generative Adversarial networks. This was a head scratcher for sure. Since it was the first time we had played around with generative adversarial networks there were a number of hard concepts we waddled through. We are going through https://machinelearningmastery.com/how-to-develop-a-generative-adversarial-network-for-a-1-dimensional-function-from-scratch-in-keras/
Jason Brownlee (the author of the post) has a whole book on using GANs, so check that out too if you are interested. We will be using the environment we created in the first Data Science Club video, so if you haven't already, do that one first! Please Like and Subscribe if you would like to get these videos as we release them. Feel free to ask any questions here or in office hours
My First Convolutional Network
Today, Ravi and William are using Python and the keras library to explore convolutional neural networks for image classification. We are going through https://github.com/keras-team/keras/blob/master/examples/mnist_cnn.py We will be using the environment we created in the first Data Science Club video, so if you haven't already, do that one first! Please Like and Subscribe if you would like to get these videos as we release them. Feel free to ask any questions here or in office hours
Data Science Club Intro Music
Hey Y'all,
This is just some synths we through together for some bumper music at the beginning of the episode while we wait for streaming to get going :).