Tumbld thoughts

Has pushing the envelope of the traditional computers expanded what we can fundamentally do with computers beyond where we started out with? Or simply put, have we changed our traditional model of computing? But suppose if we did change our model, then what? Ather Fawaz takes us on a tour of Quantum Computing and its potential in the future with his new article "Knocking on Turing’s door: Quantum Computing and Machine Learning". Now live at The Gradient!

Earthrise is a Christmas image!

In light of Dr. Timnit Gebru’s recent firing from Google, Dr. Rachel Thomas from the University of San Francisco Data Institute gives an overview of Timnit's most important contributions to the machine learning community and beyond. Read the full article at The Gradient now!

How do we know that an Artificial intelligence system is doing what it is supposed to be doing? Interpretability of AI systems is crucial for successful adoption in a wide range of domains. a discussion of how this can be achieved and the current state of research can be found in Owen Shea's new article “Interpretability in Machine Learning: An Overview”, now live at The Gradient.

There are many cases where you need something more powerful than basic statistical analysis, yet not as complex or compute-intensive as a deep neural network. In the latest article ("Don't Forget About Associative Memories") from The Gradient, Robert Bates of Veloxiti provides an comprehensive overview of Associative Memories.

Frustration with peer review in AI research is all too common, but what is its source? "How Can We Improve Peer Review in NLP?", now live on The Gradient, helps us through this topic by highlighting a recent paper by Anna Rogers and Isabelle Augenstein. Join in on a discussion of why reviewing is hard to do well, and how we might improve this process using Natural Language Processing.

Our four-year long national nightmare is over!

Interested in learning more about transformer models? Check out "Transformers are Graph Neural Networks", now live at The Gradient. Chaitanya Joshi describes how the popular Transformer architecture is actually a Graph Neural Network in disguise! This suggests many potential areas of research where NLP can learn from graphs!

A great number of online courses has emerged claiming to teach skills for building a career in AI. In "The Gap: Where Machine Learning Education Falls Short", now live at The Gradient, are these classes actually helping students achieve their goals?

[2] DevoWormAI (including OpenDevoCell)

Neural Nets can often succeed on datasets, while failing to actually do the intended task. But how do this occur? In "Shortcuts: How Neural Networks Love to Cheat", now live at The Gradient, Jorn Jacobsen, Robert Geirhos, and Claudio Michaelis expand on the concept of "Shortcuts" as a unifying way of thinking about such failures.

Are human and machine perception comparable? To know more, check out "Challenges of Comparing Human and Machine Perception", now live at The Gradient. Following up on their arXiv preprint “The Notorious Difficulty of Comparing Human and Machine Perception”, Judy Borowski and Christina Funke discuss the pitfalls of making the analogy between these very different forms of intelligence, and misconceptions about machine intelligence more generally. They conclude with the observation that comparisons between natural and artificial perception must be rigorously standardized. Only then can we observe meaningful similarities and understand their common mechanisms.

Greetings from this year’s csv,conf! This version (5) was held virtually on Crowdcast, but still featured many interesting talks and llamas. Videos coming soon to the csv,conf YouTube channel. Also featured was the Orthogonal Research and Education Laboratory, with a talk on Epistemological Directories!

Interested in the intersection of Theoretical Biology and Open Data? I just presented a talk called “Developing the (Virtual) Worm Via Data-theoretical Synthesis” at Virtual Worm 2020. Virtual Worm is an online lecture series hosted by Dr. Adrian Assie, and serves as an excellent resource for C. elegans researchers homebound during the COVID pandemic. Check out their YouTube channel for past talks.

May the Fourth be with you! Here are some fun spreadsheet-based Star Wars items (see image for attributions and links). 

Open Science is great, but institutional inertia and societal challenges can still hinder research. In "Building a Distributed Virtual Laboratory Adjacent to Academia", live now on MetaArXiv, learn how to develop an alternative academic institution.

.....and you want to be my latex salesman (on Mars)!