Data Skeptic

Interpretability Tooling

Kyle and Linhda discuss how Shapley Values might be a good tool for determining what makes the cut for a home renovation.

Kyle and Linhda discuss how Shapley Values might be a good tool for determining what makes the cut for a home renovation.

Shapley Values

We welcome back <a href= "https://homes.cs.washington.edu/~marcotcr/" rel="nofollow">Marco Tulio Ribeiro</a> to discuss research he has done since our original discussion on <a href= "https://dataskeptic.com/blog/episodes/2016/trusting-machine-learning-models-with-lime" rel="nofollow">LIME</a>. In particular, we ask the question <a href= "https://homes.cs.washington.edu/~marcotcr/acl19_implication.pdf" rel="nofollow">Are Red Roses Red?</a> and discuss how <a href= "https://homes.cs.washington.edu/~marcotcr/aaai18.pdf" rel= "nofollow">Anchors</a> provide high precision model-agnostic explanations. <hr /> Please take our <a href= "https://docs.google.com/forms/d/e/1FAIpQLSc7SbmG04zJFxrDsMH0uIm1geqKwDSJ6P3gq3oGl_9T251Pww/viewform" rel="nofollow">listener survey</a>.

We welcome back Marco Tulio Ribeiro to discuss research he has done since our original discussion on LIME. In particular, we ask the question Are Red Roses Red? and discuss how Anchors provide high precision model-agnostic explanations. Please take our listener survey.

Anchors as Explanations

Mathematical Models of Ecological Systems

Walt Woods joins us to discuss his paper <a href= "https://arxiv.org/abs/1906.02896" rel="nofollow">Adversarial Explanations for Understanding Image Classification Decisions and Improved Neural Network Robustness</a> with co-authors Jack Chen and Christof Teuscher.

Walt Woods joins us to discuss his paper Adversarial Explanations for Understanding Image Classification Decisions and Improved Neural Network Robustness with co-authors Jack Chen and Christof Teuscher.

Adversarial Explanations

<a href="https://twitter.com/_abarbu_" rel="nofollow">Andrei Barbu</a> joins us to discuss <a href= "https://objectnet.dev/" rel="nofollow">ObjectNet</a> - a new kind of vision dataset. In contrast to ImageNet, ObjectNet seeks to provide images that are more representative of the types of images an autonomous machine is likely to encounter in the real world. Collecting a dataset in this way required careful use of Mechanical Turk to get Turkers to provide a corpus of images that removes some of the bias found in ImageNet. <a href="http://0xab.com/" rel= "nofollow">http://0xab.com/</a>

Andrei Barbu joins us to discuss ObjectNet - a new kind of vision dataset. In contrast to ImageNet, ObjectNet seeks to provide images that are more representative of the types of images an autonomous machine is likely to encounter in the real world. Collecting a dataset in this way required careful use of Mechanical Turk to get Turkers to provide a corpus of images that removes some of the bias found in ImageNet. http://0xab.com/

ObjectNet

<a href="https://twitter.com/filwd">Enrico Bertini</a> joins us to discuss how data visualization can be used to help make machine learning more interpretable and explainable. Find out more about Enrico at <a href= "http://enrico.bertini.io/">http://enrico.bertini.io/</a>. More from Enrico with co-host Moritz Stefaner on the <a href="https://datastori.es/">Data Stories</a> podcast!

Enrico Bertini joins us to discuss how data visualization can be used to help make machine learning more interpretable and explainable. Find out more about Enrico at http://enrico.bertini.io/. More from Enrico with co-host Moritz Stefaner on the Data Stories podcast!

Visualization and Interpretability

We welcome <a href= "https://scholar.google.com/citations?hl=en&user=bJZV7r4AAAAJ">Su Wang</a> back to Data Skeptic to discuss the paper <a href= "https://arxiv.org/pdf/1704.04550.pdf">Distributional modeling on a diet: One-shot word learning from text only</a>.

We welcome Su Wang back to Data Skeptic to discuss the paper Distributional modeling on a diet: One-shot word learning from text only.

Interpretable One Shot Learning

<a href= "https://twitter.com/wiebevr">Wiebe van Ranst</a> joins us to talk about a project in which specially designed printed images can fool a computer vision system, preventing it from identifying a person.  Their attack targets the popular YOLO2 pre-trained image recognition model, and thus, is likely to be widely applicable.

Wiebe van Ranst joins us to talk about a project in which specially designed printed images can fool a computer vision system, preventing it from identifying a person.  Their attack targets the popular YOLO2 pre-trained image recognition model, and thus, is likely to be widely applicable.

Fooling Computer Vision

This episode includes an interview with Aaron Roth author of <a href= "https://www.amazon.com/Ethical-Algorithm-Science-Socially-Design/dp/0190948205"> The Ethical Algorithm</a>.

This episode includes an interview with Aaron Roth author of The Ethical Algorithm.

Algorithmic Fairness

<h2>Interpretability</h2> Machine learning has shown a rapid expansion into every sector and industry. With increasing reliance on models and increasing stakes for the decisions of models, questions of how models actually work are becoming increasingly important to ask. Welcome to Data Skeptic Interpretability. In this episode, Kyle interviews <a href= "https://twitter.com/christophmolnar">Christoph Molnar</a> about his book <a href= "https://christophm.github.io/interpretable-ml-book/">Interpretable Machine Learning</a>. Thanks to our sponsor, the <a href= "gartner.com/us/data">Gartner Data & Analytics Summit</a> going on in Grapevine, TX on March 23 – 26, 2020. Use discount code: dataskeptic. <h3>Music</h3> Our new theme song is <a href= "https://bigdandthekidstableofficial.bandcamp.com/track/5">#5</a> by <a href="https://www.bigdandthekidstable.com/">Big D and the Kids Table</a>. Incidental music by <a href= "https://tanukisuitriot.bandcamp.com/album/the-edo-sessions">Tanuki Suit Riot</a>.

Interpretability Machine learning has shown a rapid expansion into every sector and industry. With increasing reliance on models and increasing stakes for the decisions of models, questions of how models actually work are becoming increasingly important to ask. Welcome to Data Skeptic Interpretability. In this episode, Kyle interviews Christoph Molnar about his book Interpretable Machine Learning. Thanks to our sponsor, the Gartner Data & Analytics Summit going on in Grapevine, TX on March 23 – 26, 2020. Use discount code: dataskeptic. Music Our new theme song is #5 by Big D and the Kids Table. Incidental music by Tanuki Suit Riot.

Interpretability

NLP in 2019

We are joined by Colin Raffel to discuss the paper "Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer".

We are joined by Colin Raffel to discuss the paper "Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer".

The Limits of NLP

Seth Juarez joins us to discuss the toolbox of options available to a data scientist to jumpstart or extend their machine learning efforts.

Seth Juarez joins us to discuss the toolbox of options available to a data scientist to jumpstart or extend their machine learning efforts.

Jumpstart Your ML Project

Alex Reeves joins us to discuss some of the challenges around building a serverless, scalable, generic machine learning pipeline.  The is a technical deep dive on architecting solutions and a discussion of some of the design choices made.

Alex Reeves joins us to discuss some of the challenges around building a serverless, scalable, generic machine learning pipeline.  The is a technical deep dive on architecting solutions and a discussion of some of the design choices made.

Serverless NLP Model Training

Data Skeptic is a data science podcast exploring machine learning, statistics, artificial intelligence, and other data topics through short tutorials and interviews with domain experts.

hubhopper

Data Skeptic

hubhopper