Global Data Science Forum

 View Only

Data Science Community News | Volume 1, Issue 5

By Christina Howell posted Mon December 09, 2019 01:36 PM

IBM Data Science Community Newsletter
December 2019 | Volume 1, Issue 5

IBM Data Science Community Nominations > We're looking for passionate IBM Data Science Community members to share their stories with us. Nominate yourself or someone you recommend to be interviewed and profiled for our community.


Giving Algorithms a Sense of Uncertainty Could Make Them More Ethical

The ethical impact of Machine Learning has been hotly debated—from bias in training data to trolly problem dilemmas for self-driving cars. Can an algorithm trained to optimize a singular objective be designed to consider multiple (potentially mutually inconsistent) objectives? Perhaps data sampling of comparative choices and complex loss functions are the key.

Using ML-Generated Text to Combat Hate Speech with Thoughtful Responses

The internet can sometimes feel like a haggard and unpleasant landscape to navigate. Learn about how researchers at UCSB created a system to lower the tensions in digital forums using NLP in 4 essential steps.

AI Skills

IBM's AI Workflow Specialization Course Has 4.8 Stars on Coursera

The course was built by data scientists for experienced data scientists who want to grow their skills and find new opportunities to lead AI in large enterprises. Learn more and ask questions on this topic thread or enroll here.

Tools & Libraries

A Visual Guide to Using BERT for the First Time

This is a particularly good explainer for using BERT and understanding the concepts behind it. Check out the explainer and library from HuggingFace, in action. Read more

AVA: The Art and Science of Image Discovery at Netflix

Fun article about the unique approach Netflix is using to identify frames that will capture an audience's attention for digital merchandising. Read more

Solutions & Products

A Cautionary Tale in Using Black Boxes: Apple Card and Potential Bias

New York Department of Financial services is investigating Apple's treatment of one woman's Apple Card credit limit, and her husband supposes a black-box algorithm is to blame. Read more

Reconstructing Brain Thoughts by Plugging a Neural Network into a Brain

A remarkable experiment: researchers have used brainwaves in the form of EEG data to build an image classifier that can (kind of) reconstruct what the subject is seeing.Read more


Effective Implementation of an Active Learning System in the Actual Wild

An ML group set out to help wildlife researchers track target animals. This paper details each step they took to implement an active learning system. Read more

Taxonomy of Failures in Deep Learning Systems

Deep learning systems can fail for many reasons. This paper is the first in-depth review of those causes, and a catalog of the primary fault categories. Read more


See all upcoming community events here.

December 10, 6–9 pm | In-person | Los Angeles, CA