A summary of recent happenings in the field of Artificial Intelligence (AI) for week 36 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 36 of 2020

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

  • Hypatos raises quite a million: The German-Poland based startup has raised around $11.8 million, and is planning to invest it in R&D for processing documents. Here you can read more.
  • Predictive algorithms, interference in privacy: The Canada Police Authorities are employing machine learning algorithms to predict crimes to maintain law and order in the country. Though this step appears to be for the betterment of their citizens, it still violates certain fundamental rights through monitoring of social media posts, facial recognition across various footages. Read more about it here.
  • NVIDIA’s RTX3090: This giant GPU with 24GB of G6X memory has 8K resolution perfect for gaming with a $1499 price. To know more, click here.
  • AI for chemical research: RoboRXN, IBM’s deep learning algorithms, Cloud and robotic labs are together working towards automating research labs by lending a helping hand to the chemists without human intervention. To read more, visit here.
  • Facebook’s Superhearing Glasses: Facebook’s AR glasses tend to amplify that you are trying to hear through input from the glasses. This technique is similar to hearing aids with the ability to help both impaired and not impaired people. Read more here.
  • Heartbeat Detection for DeepFake: The researchers at Binghamton University and Intel used residual biological signals from PPG cells on a person’s face. This is the DeepFake Heartbeats that are used to find out the source of DeepFake videos. Read more on this news over here.
  • Facebook’s Opacus: A new release of a high-speed library that uses differential privacy to train Pytorch models. To learn more about Opacus, visit here.
  • Google’s new “News” feature: Google is working on adding ‘Your News Update’ feature in Google News that would club personalized news feed based on your past views into an audio feed. To get a detailed read visit here.
  • Cheap Supercomputer substitute for researchers: Not everyone can afford GPT-3 but worry not, even individual researchers can use a powerful distributed training paradigm. To know how – read this.
  • Flood forecasting: The Google AI team has devised a new approach, the Morphological inundation model for flood forecasting. This model uses machine learning to determine more accurately and scalable inundation models in real scenarios. Read more this approach over here.

Articles for week 36 of 2020

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

  • Convert PDFs to Audio Books: It’s a dream of a busy person who doesn’t find time to read out something he wishes to but cannot due to circumstances like traveling, daily chores, or any other reason. This article extracts PDFs with text and uses TTS to convert into speech using Machine Learning. To read more, visit here.
  • Pix2Pix GAN: This is a Deep Convolutional Neural Networks designed for Image to Image Translation. The tutorial considers satellite to map image as a case study. To learn more about this, visit this page.
  • DP-SGD tutorial: This is the first post out of the tutorial series by Pytorch based on the Differentially Private-Stochastic Gradient Descent algorithm (DP-SGD). It includes a detailed introduction of Differential Privacy and an extended description of SGD. To read the entire article, visit here.
  • Steps to become AI-expert: This complete guide takes you closer to becoming an AI-expert. Right from programming tips to Statistics to Jobs to Data handling and ultimate the Kaggle competition, are the essential steps to become a master of AI. Read more about this article here.
  • Why choose BERT: This leading and state of the art model for QA and language translation is now a choice of most of NLP enthusiasts. The reason behind its performance and salient features are discussed in this article.
  • Multiple Material Property prediction: For the researchers benefitting from Material Research, the Material Graph Network (MEGNet) which overcomes challenges in predicting properties of materials such as catalysts, crystals, molecules, etc. The shortcomings faced in multiple material prediction and how MEGNet is proving to be beneficial are discussed here.
  • Introduction to Linear Regression: An elegant yet simple article for beginners on Linear Regression can be found here. The blog post is well supported with scripts and visuals for a clearer first-time introduction.
  • Road Lane Lines: A focused image processing task using OpenCV is to find lane marks on the road (region of interest). This post shares steps and script to do so in a conveniently simple manner using visualization. Read the entire article here.
  • Contrastive self-supervised learning: The new framework is used to analyze examples of leading approaches such as SimCLR, CPC, and AMDIM. To read more on this post, visit here.
  • Meta-Learning: A quick introduction article on various types of meta-learning is a must-read for starters in this area.
  • Frechlet Inception Distance: An evaluation measure for GANs is explained in this article. The research paper is also available here.
  • Facebook’s BigGraph: A new framework by PyTorch that easily helps in developing graph embeddings for larger graphs. To read the entire post, visit here.


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

  • Source Code Modeling: This is a GitHub repository to train deep learning models on source code.
  • KerasTuner: A jupyter notebook to use KerasTuner for discovering the best model can be found here.


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

  • Deep Reinforcement Learning Workshop: The deadline for paper submission is 5th October 2020. To submit your article, visit here.
  • FDP on IoT with ML and AI: The ISEA sponsored Faculty Development Program based on the Internet of Things with Machine Learning and Artificial Intelligence is suppose to be held on 7th September 2020. The registration is free and can be done here.
  • Graph Convolutional Networks: This a lecture video from NYU’s Deep Learning course on Graph Convolutional Networks (GCN). The link to the video is available here
  • Where to start with Machine Learning: A nice introductory video mentioning 5 curriculum for mastering ML and 6 tips to keep in mind. 
  • Text Data Analysis and Visualization: This is a video tutorial for those who wish to analyze their text data and visualize it in python. To watch the video, visit here.
  • Robot Rickshaw: Adam Savage’s Spot robot Rickshaw is shown in this video.
  • Scope of Neuralink: Lex Fridman Podcasts presents its second video on 8 futuristic ways for Neuralink including VR, Telepathy, and Memories. To see the video, visit here
  • Age prediction from face: This video uses a small five liner script using deepfake model to predict the age of a face.


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

  • AllenAct: Allen Institute for AI-open sources has launched its flexible framework for Embodied-AI research. The GitHub script and research paper are available.
  • English Text Summarization: OpenAI’s new approach that focuses on English Text Summarization shows improvement in comparison to 10x larger trained models. These models produce summaries that are trained using Supervised learning only. Here is their entire article along with GitHub code and paper
  • The 3D posing of Transparent objects: KeyPose estimates 3D keypoints from monocular or stereo images. To know more about this Google research, read their blog here.
  • Worm Pose estimation: This WormPose package estimates difficult poses of C. The Github script and paper are available. 
  • Wholistic view of Continual Learning: The Deep Neural Networks and their not so perfect data are given a new angle for processing. The research paper can be found here


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

  • AI-Scholar: This hub of AI articles has a whole lot of content which includes NLP, Image and Voice Recognition, GAN, Reinforcement learning, and much more. Visit the website here
  • Dive into Deep Learning: Amazon’s team has developed an open sourcebook with extended updates for PDF, HTML versions. Recently, the authors added two new frameworks: PyTorch and Tensorflow to attract a larger audience. It’s an amalgam of text, maths, and runnable scripts. Read more about its features here.
  • COVID-19 analysis: Oak Ridge National Lab Tennesse has trained 40,000 genes from 17,000 genetic samples, leading to a “eureka moment”. Dr. Daniel Jacobson and his team we able to make speculations that were new and are considered to be a breakthrough in understanding COVID-19 better. Read more about this news here.
  • Relationship analysis using ML: See there is nothing that AI can’t do. You dot need a tarot card reader to you about your relationship crisis. Just follow the scientific way of machine learning rules and you would have a happy relation similar astrology. To read more about the five perfect rules for happy love life, read here.  

That’s all for the week 36 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.