A summary of top updates in the field of Artificial Intelligence (AI) for week 39 of 2020. For more regular updates and related discussions, join us in the Artificial Intelligence Community. As always, thanks for your support for sharing it with your friends and colleagues on social media.

Industry updates for week 39 of 2020

This section is for recent industry updates related to Artificial Intelligence.

  • DeepMind’s Agent57 shines in 57th Atari Games: This deep reinforcement learning agent sets a milestone by defeating previous DRL agents. To read more about its features and field of applications visit here.
  • Apple’s Handwriting recognition secret: You write fast or slow, apple’s new iPadOS update for with Apple’s Pencil uses machine learning to understand anything scribbled. To read out more, visit here.
  • Fairer AI needs a peep into the black box: The physicist usually builds greater models to decrypt the inner functioning of anything that is of interest. This perspective will help make AI responsible and ethical as assuring the task assigned to an AI is done reliably. To read about how physicists are planning on doing so, visit here.
  • All-new Echo: The Amazon’s echo had a maker over about its looks and a customized processor for enhanced machine learning tasks. Here are the launch details and costing.
  • Ignite 2020 Conference updates: The highlights from Microsoft’s Ignite 2020 conference, which was a virtual event mentioned Azure Cognitive and Bot services, Cosmo DB, Edge, and much more. To get the detailed highlights read here.
  • Facebook AI’s Dynabeach: A testing platform for benchmarking. Here are all the technical details.
  • AI-accelerated Alice Camera: The UK based startup, Photogram AI has developed an AI-assisted camera called Alice camera that promises better connected than the DSLR. To know more about this new camera, read here.
  • Facebook’s policy violation by GAN’s fake faces: A violation of Facebooks’ co-ordinated in-authenticate behavior, the social media giant eliminated more than 150 accounts and associated Instagram and groups. They coordinated with Graphika to discover that the Chinese network of dummy accounts used GAN’s for developing fake faces. Read the entire coverage here.
  • Microsoft to license GPT-3: The joint venture of Microsoft and OpenAI to license GPT-3 is for developing advanced AI solutions. To know more, read here.
  • Competition to GPT-3 arrives: The researchers at Ludwig Maximilian University, Munich have developed a text generator that is a bite-sized and reduced number of parameters. The whole article can be found here.
  • Tesla’s machine learning training for web-service: Tesla is planning to train all the video data using Dojo for its self-driving cars. To know more, read here

Articles for week 39 of 2020

This section is for recent Blog Posts/Tutorials related to Artificial Intelligence.

  • Google’s Projection-based model: The extension for PRADO, pQRNN which is the new state-of-the-art model with the small model size and fast computation of NLP tasks. Here is where you can read more about the parallel quasi RNN model.
  • Multi-step time series prediction for Covid-19 cases: A good article if you need a start to work on deep learning for COVID-19 case prediction. The multi-step time series for predicting COVID-19 cases along with the script is mentioned here
    .
  • Bernoulli Distribution Tutorial: The pre-requisite for any AI task needs knowledge of random variables and data distribution methods such as Bernoulli. This article explains the basics with the python script.
  • Interactive analysis of sentence embeddings: This article explores the Embedding Projector tool for visualizing high-dimensional data, such as sentence embeddings and their corresponding class labels. You may find the steps to use it, here.
  • A penguin-fish recommender: An example based study post is elaborated for predicting the food requirements of certain species affected by certain environmental conditions. The post mentions the script to address this machine learning task.
  • NAS surpassing EfficientNet: RegNet is a simplified regularization method that achieves similar performance as EfficientNet. To know more, visit here.

Softwares

This section is for recent Tools/Applications related to Artificial Intelligence.

  • Adversarial NLI: A new benchmark for NLP understanding.

Videos

This section is for recent Lectures/Conferences/Webinars related to Artificial Intelligence.

  • GANs for Good: A keynote organized by Deeplearning.ai featuring Ian Goodfellow, Alexei Efros, Andrew Ng, and many more. The event is scheduled to be held on 30th September and to register for free, visit here.
  • Time-series Analysis using STUMPY: A quick kick-start with time-series data mining task using Stumpy. Watch the video here.
  • Digit Classification: This video explains digit classification using CNN. This is done using TensorFlow, Keras, and OpenCV.
  • Efficient robots for impossible tasks: Boston Dynamics unveiled footage showcasing the new tricks that Atlas and Spot Mini managed to acquire.
  • AI plays Fall Guys and beats humans: This video surfaces AI unveiling abilities to do anything like a human, including leisure like playing games. Here is the video showing AI beating human score while learning to play FallGuys. 

Research

This section is for recent research published related to Artificial Intelligence.

  • Backdoor Attacks that threaten AI: This survey paper is based on a comprehensive study of backdoor attacks that are a potential threat to AI systems. The paper details are described here.
  • DiffWave Model for Audio Synthesis: An article based on the Diffusion probabilistic model for waveform generation is available for reading in this link and for code.  
  • Graph Normalization for Graph Neural Networks: The link to paper and code is available here. 

Miscellaneous

This section is for updates for other various related topics to Artificial Intelligence.

That’s all for week 39 of 2020. In case you think we have missed out something and that is significant to add, please free to reach out and submit to us via our Contact form or via Social media channels: Twitter and Facebook.