Microsoft’s Free Datasets, TensorFlow’s Free Platform for Amateur Data Scientists, Facebook’s Open Source NLP Dataset, among other Game-Changing ML Developments last week!

Welcome to another jam-packed development-filled round-up of all the incredible happenings in the machine learning community last week! From TensorFlow’s Seedbank to Facebook open sourcing a huge NLP and navigation dataset, we have rounded-up the latest cutting-edge developments for you.

Other highlights from the past week: An autonomous car that teaches itself to drive in 20 minutes, NVIDIA’s awesome technique to fix bad photos in milliseconds, Samsung’s NLP competitions-winning algorithm, and more!

You can get these AVBytes articles delivered straight to your inbox on a daily basis! All you need to do is subscribe here. We’ll do everything else. :)

Click on the headline to read the full article.

• Microsoft has Released a Collection of Awesome Free Datasets: Here’s a chance for you to work with real-world industry data, polish your skills and get noticed! Microsoft has released a collection of open free datasets curated over a number of years from their research areas.

• Learn and Improve your Machine Learning Skills with TensorFlow’s Free Seedbank Platform: Starting off with applying ML concepts can be challenging. So here comes Seedbank — a free in-browser platform that has pre-loaded Python codes and pretrained models on classification, unsupervised learning, NLP and other ML tasks! And you get free GPU support!

• Facebook Open Sources Dataset on NLP and Navigation Every Data Scientist should Download: Facebook’s AI team has released an open source dataset, called “”Talk the Walk””, that combines NLP with navigation data. Get your hands on it now and start playing around with it! It also comes with a challenge — can you make the model work?

• NVIDIA’s Noise2Noise Technique Fixes Bad Images in Milliseconds: Each one of us has taken blurry or grainy photos at some point. Now, you can fix them in milliseconds with machine learning thanks to NVIDIA’s Noise2Noise technique! It doesn’t even need clean images to learn, it produces stunning quality using only corrupted images!

• Samsung’s ConZNet Algorithm just won Two Popular NLP Challenges (Dataset Links Inside): Samsung’s deep reinforcement learning algorithm, ConZNet, just won 2 hugely popular NLP challenges — TriviaQA and MS MARCO. Both these datasets are open source and you can download them NOW (links inside). Recommended for anyone interested in NLP!

• An Autonomous Car Learned how to Drive itself in 20 minutes using Reinforcement Learning: This autonomous car learns how to drive itself in just 20 minutes! Using reinforcement learning, the car is “”rewarded”” each time it learns from it’s mistakes. It took less than 20 trials to achieve 95% accuracy (DeepMind’s Atari algorithm took over a million trials)! Check out details of the algorithm + research paper and video inside.

The above AVBytes were published from 9th to 15th July, 2018.