Excited to share my research tomorrow #AAAI23 Doctoral Consortium UMBC Graduate School Edward Raff

Came across this interesting SF book written in the 70s. Considering ChatGPT rave, thought to share. Amazon reviewer wrote 'Its theme is not a traditional sf topic...language as the means to bridge the gap between human consciousness and the otherness of the objective world.'

Me giving Goodbye speech to the 1 follower who always likes my tweet
😬
#Twittershutdown
#TwitterOFF
#تھپڑ_شریف

Coolest paper I read this week: 'Translation between Molecules and Natural Language'

Uses self-supervised language modeling + fine-tuning on a dataset of molecule descriptions to:

1) Generate text descriptions of molecules
2) Generate a molecule from a text description!

13/18

This network solved 10 International Math Olympiad problems. That is 10 more than what I can solve 😁

Apart from being ground-breaking, it's also quite accessible -- available as a VSCode plugin!

Congrats Guillaume Lample and team!

5 best Raspberry Pi alternatives zdnet.com/article/best-r…

Grad school applicants: my Computational Robotics group at Harvard SEAS is hiring PhD students in Fall 2023!

If you are interested in robotics, computer vision, machine learning, applied math, and their application in safe autonomy, you are the person I am looking for!

My Home Assistant staging instance. Raspberry Pi 4 with four Raspberry Pi Zero 2W. PoE powered
#100DaysOfHomeLab #raspberrypi

It turns out that if you let neural nets *think* long enough, they can perform logical extrapolation and solve *hard OOD* examples during test time!

Hoping one day to share my thesis like him.

I'm so excited to share my PhD thesis publicly. It's 88 pages long, I wrote it from scratch, and I tried to create a useful document for anyone wanting to gain a rigorous but unfussy understanding of the mathematics of deep learning.

It's here: arxiv.org/abs/2210.10101

(1/7)

The 7 approaches for dealing with limited labeled data in supervised machine learning settings:

1 Transfer learning
2 Self-supervised learning
3 Semi-supervised learning
4 Active learning
5 Few/zero-shot learning
6 Weakly-supervised learning
7 Label more data!

Anything missing?

This morning is paper implementation Wednesday. Actually getting pretty good at this now... 😎 Edward Raff UMBC Department of Information Systems Booz Allen Hamilton

Can mathematical obfuscation help to get papers accepted?

Very excited and honored to get accepted to the AAAI 2023 Doctoral Consortium! Edward Raff UMBC Department of Information Systems Booz Allen Hamilton

Great news! I passed my proposal defense UMBC Department of Information Systems and am now finally a PhD candidate 😎 grateful for an incredible set of advisors and committee Edward Raff and Sanjay Purushotham UMBC Department of Information Systems Mohammad Osman Gani and Profs. Oates, Roy, and Foulds.

Organizing old papers I saved over the years... That time when I was into capsule networks.

Jupyter notebook if someone wants to toy around with it: github.com/rasbt/machine-…