Lightning defers training and validation loop logic to you. It guarantees correct, modern best practices for the core training logic.
Rapid research framework for Pytorch. The researcher's version of keras
https://github.com/williamFalcon/pytorch-lightning

A small team of researchers at Indiana University has created the first global map of labor flow in collaboration with the world's largest professional social network, LinkedIn. The work is reported in the journal Nature Communications.
The study's lead authors are Jaehyuk Park and Ian Wood, Ph.D. students working with Yong Yeol "Y.Y." Ahn, a professor at the IU School of Informatics, Computing and Engineering in Bloomington.
According to the researchers, the study's result represents a powerful tool for understanding the flow of people between industries and regions in the U.S. and beyond. It could also help policymakers better understand how to address critical skill gaps in the labor market or connect workers with new opportunities in nearby communities.
The study showed some unexpected connections between economic sectors, such as the strong ties between credit card and airline industries. It also identified growing industries during the study period from 2010 to 2014, including the pharmaceutical and oil and gas industries—with in-demand skills such as team management and project management—as well as declining industries, such as retail and telecommunications.
IU researchers created the map using LinkedIn's data on 500 million people between 1990 and 2015, including about 130 million job transitions between more than 4 million companies. The researchers gained access to this rare data as one of only 11 teams selected to participate in the inaugural LinkedIn Economic Graph Research program in 2015. They later became one of only two teams—IU and MIT—selected to continue their work beyond 2017. The team worked closely with LinkedIn engineers, including Michael Conover, a graduate of the IU School of Informatics, Computing and Engineering and a senior data scientist at LinkedIn at the time of the study.
In a blog post on LinkedIn, Park compares the study to a "roadmap" to the future economy since the first step in any journey requires understanding the current landscape.
"We expect this study will provide a powerful foundation for further systematic analysis of geo-industrial clusters in the context of business strategy, urban economics, regional economics and international development fields—as well as providing useful insights for policymakers and business leaders," he said.
https://phys.org/news/2019-08-global-economy-collaboration-linkedin.html

"Anyone Can Learn Artificial Intelligence With This Blog"

A Simple, Illustrated Explanation in Colab, By David Code : https://colab.research.google.com/drive/1VdwQq8JJsonfT4SV0pfXKZ1vsoNvvxcH

#artificialintelligence #deeplearning #neuralnetworks

FairSight: Visual Analytics for Fairness in Decision Making. arxiv.org/abs/1908.00176

Curiosity-driven Reinforcement Learning for Diverse Visual Paragraph Generation. arxiv.org/abs/1908.00169

Supervised Learning of the Global Risk Network Activation from Media Event Reports. arxiv.org/abs/1908.00164

Multi-path Learning for Object Pose Estimation Across Domains. arxiv.org/abs/1908.00151

3D Virtual Garment Modeling from RGB Images. arxiv.org/abs/1908.00114

Energy-Based Adversarial Training and Video Prediction, NeurIPS 2016

By Yann LeCun, Facebook AI Research

YouTube: https://youtu.be/x4sI5qO6O2Y

#DeepLearning #EnergyBasedModels #UnsupervisedLearning

Deep TabNine: The all-language autocompleter.

It uses machine learning to provide responsive, reliable, and relevant suggestions.
Traditional autocompleters suggest one word at a time while Deep TabNine, based on GPT-2 and trained on 22 million Github files suggest way much more.

Isn't it cool ! Give it a try 🙌

Website: https://tabnine.com/
Github: https://github.com/zxqfl/tabnine

"Advanced NLP with spaCy"
By Ines Montani : https://course.spacy.io
#machinelearning #nlp #naturallanguageprocessing

"Rapid research framework for PyTorch. The researcher's version of Keras"

GitHub, by William Falcon : https://github.com/williamFalcon/pytorch-lightning

#deeplearning #pytorch #research

Graph Neural Networks for Small Graph and Giant Network Representation Learning: An Overview

Jiawei Zhang : https://arxiv.org/abs/1908.00187

#MachineLearning #NeuralEvolutionary #MachineLearning

Functional Regularisation for Continual Learning
Titsias et al.: https://arxiv.org/abs/1901.11356
#ArtificialIntelligence #BayesianInference #MachineLearning

Disentangling Disentanglement in Variational Autoencoders

Mathieu et al.: https://proceedings.mlr.press/v97/mathieu19a.html

#DeepLearning #VariationalAutoencoders #VAE

MelNet: A Generative Model for Audio in the Frequency Domain

Sean Vasquez, Mike Lewis: https://arxiv.org/abs/1906.01083

Blog: https://sjvasquez.github.io/blog/melnet/

#ArtificialIntelligence #AudioProcessing #MachineLearning

Welcome this Chinese DeepLearning Chip called “Tianjic”.

Paper: https://www.nature.com/articles/s41586-019-1424-8

It can run traditional deep learning code and also perform "neuromorophic" operations in the same circuitry.

Ten trends in Deep learning NLP
https://blog.floydhub.com/ten-trends-in-deep-learning-nlp/

"Deep Boltzmann Machines"
Ruslan Salakhutdinov and Geoffrey Hinton : https://proceedings.mlr.press/v5/salakhutdinov09a/salakhutdinov09a.pdf
#BoltzmannMachines #DeepBoltzmannMachines #DeepLearning