ML News - DeepMind does Nowcasting | The Guardian's shady reporting | AI finishes Beethoven's 10th (Video)

Your holy update on what’s new in the Machine Learning world.

OUTLINE:
0:00 - Intro
0:30 - DeepMind tackles Nowcasting
3:30 - The Guardian’s shady reporting on TruthfulQA
6:15 - Stochastic training not necessary for generalization
7:35 - Google AI’s efficient partitioning of road networks
9:15 - MiniHack Reinforcement Learning Environment
10:45 - Plato XL 11B dialog model
11:35 - AI finishes Beethoven’s 10th Symphony
13:10 - AI casts doubt on painting authenticity
15:55 - ShadowDragon social media surveillance
18:45 - Helpful Libraries
25:20 - Samsung to copy-paste brains onto chips

References:
DeepMind improves Nowcasting
https://deepmind.com/blog/article/now
https://www.nature.com/articles/s4158
https://github.com/deepmind/deepmind-
https://colab.research.google.com/git

The Guardian’s shady reporting on TruthfulQA
https://www.theguardian.com/commentis

Stochastic Training is Not Necessary for Generalization

Google AI - Efficient Partitioning of Road Networks
https://ai.googleblog.com/2021/09/eff

MiniHack Reinforcement Learning Environment
Facebook

Baidu PLATO-XL 11B Dialog Model
http://research.baidu.com/Blog/index-

AI finishes Beethoven’s 10th Symphony
https://thenextweb.com/news/computer-

AI casts doubt on paining authenticity
https://www.smithsonianmag.com/smart-
https://art-recognition.com/
https://art-recognition.com/case-stud
FAQ - Art Recognition

ShadowDragon Social Media Surveillance
https://www.rt.com/usa/535630-ai-surv
https://theintercept.com/2021/09/21/s

Helpful Libraries / Datasets

The Text-Based NP Enrichment Task (TNE) | TNE
[2109.10282] TrOCR: Transformer-based Optical Character Recognition with Pre-trained Models
https://github.com/microsoft/unilm/tr
People + AI Research – Medium
https://raft.elicit.org/
https://huggingface.co/spaces/ought/r
https://huggingface.co/spaces/ought/r
https://arxiv.org/pdf/2109.14076.pdf
https://arxiv.org/pdf/2109.14394.pdf
https://www.robots.ox.ac.uk/~vgg/rese
PASS: An ImageNet replacement for self-supervised pretraining without humans | Zenodo
https://github.com/yukimasano/PASS/
https://openreview.net/pdf?id=BwzYI-K
https://github.com/pytorch/data?utm_s

Samsung Method to copy paste brain onto chip
https://www.engadget.com/samsung-copy