Paper-Title: COBRA: Data-Efficient Model-Based RL through Unsupervised Object Discovery and Curiosity Driven Exploration
#DeepmindAI
Link to the paper: https://arxiv.org/pdf/1905.09275.pdf
The three specific tasks performed in this paper using SpriteWorld(https://github.com/deepmind/spriteworld) are:-
Goal-finding task. The agent must bring the target sprites (squares) to the centre of the arena.
Clustering task. The agent must arrange the sprites into clusters according to their colour.
Sorting task. The agent must sort the sprites into goal locations according to their colour (each colour is associated with a different goal location).
The main technical contributions of the paper are:-
A method for learning action-conditioned dynamics over slot-structured object-centric representations that require no supervision and is trained from raw pixels.
A method for learning a distribution over a multi-dimensional continuous action space. This learned distribution can be sampled efficiently.
An integrated continuous control agent architecture that combines unsupervised learning, adversarial learning through exploration, and model-based RL.

New models in 17 and 100 languages XLM/mBERT pytorch
LM supports multi-GPU and multi-node training

https://github.com/facebookresearch/XLM#pretrained-cross-lingual-language-models

The world's largest chip : Cerebras Wafer Scale Engine
A new class of computer system that accelerates artificial intelligence by orders of magnitude beyond the current state of the art - https://www.cerebras.net
#ArtificialIntelligence #DeepLearning #Hardware

Andrew ng :


I’m delighted to announce our first international office in Medellin, Colombia! Landing AI, deeplearning.ai and AI Fund have teams there. AI needs to expand beyond a small handful of hubs like Silicon Valley and Beijing, and I’m bullish about Latin America. See why:

https://medium.com/@andrewng/ai-in-latin-america-announcing-our-first-international-office-in-colombia-ac4e203c1564?source=friends_link&sk=22f7f197049bbeb72489ac25dd6e18ed

Transformers from scratch
Modern transformers are super simple, so they can be explained in a really straightforward manner
Blog by Peter Bloem, with pytorch code : https://peterbloem.nl/blog/transformers
#MachineLearning #PyTorch #Transformers

"Mathematics for Machine Learning"
Book by Marc Peter Deisenroth, A Aldo Faisal, and Cheng Soon Ong : https://mml-book.github.io
#ArtificialIntelligence #MachineLearning #Mathematics

Yoshua Bengio :

Happy to share the CFP for the AI for social good workshop at NeurIPS 2019! Please refer to our site for additional details. Workshop site: https://aiforsocialgood.github.io/neurips2019/cfp.htm … Submission site: https://cmt3.research.microsoft.com/NIPSJWAISG2019 Submission deadline: Sept 6

Data Science Roles: A Classification Problem

https://towardsdatascience.com/data-science-roles-a-classification-problem-ebe6fae10169

Spriteworld: A Flexible, Configurable Reinforcement Learning Environment
Watters et al., 2019, DeepMind : https://github.com/deepmind/spriteworld
#deeplearning #artificialintelligence #reinforcementlearning

Generative Adversarial Networks (GANs) in 50 lines of code (PyTorch)
tl;dr: GANs are simpler to set up than you think
Blog by Dev Nag : https://medium.com/@devnag/generative-adversarial-networks-gans-in-50-lines-of-code-pytorch-e81b79659e3f
#deeplearning #generativeadversarialnetworks #pytorch

Boundless: Generative Adversarial Networks for Image Extension. https://arxiv.org/abs/1908.07007

Joint Embedding of 3D Scan and CAD Objects. https://arxiv.org/abs/1908.06989

Reinforcement Learning Applications. https://arxiv.org/abs/1908.06973

Learning Fixed Points in Generative Adversarial Networks: From Image-to-Image Translation... https://arxiv.org/abs/1908.06965

Learning Representations and Agents for Information Retrieval. https://arxiv.org/abs/1908.06132

Robustness package
A library for experimenting with, training and evaluating neural networks, with a focus on adversarial robustness
Engstrom et al. : https://github.com/MadryLab/robustness
#artificialintelligence #deeplearning #neuralnetworks