A summary of recent updates in the field of Artificial Intelligence (AI) for week 01 of 2021. 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 01 of 2021

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

  • AI, the new democrat: Yet another article describing the new required trend in training ML/AI algorithms through less amount of data. Less than One short learning (LO) seems to be gaining popularity due to its ability to learn to distinguish between objects with fewer data. To read more, visit here.
  • Training NN on Mobiles: The ongoing outcome of IBM research would open doors for AI scientists to train their Neural Networks using a 4bit encoding. The research published in NeuroIPS has proved to limit losses through improvement in the accuracy. To follow more on this article, visit here.
  • Song lyrics by AI mentions Hilary Clinton: The fans of Hamilton are super happy to listen to this AI song although it has few hitches with some weird words added. To get more details, visit here.
  • AI for mental-health monitoring: The National Institutes of Health-funded researchers have released the outcome of studies conducted by using AI sensors to modify bacterial protein for improving fidelity. This is through the production of Serotonin that is also responsible for mood stabilization and physical wellness. To read more about the details of this research visit here.
  • Data memorization in GPT2 model: The team of AI researchers at Berkeley have conducted a study to evaluate language models regarding their memorization and regurgitating skills t find 0.1% text being copy-pasted from training data. Find out more from this post.
  • Need for new AI benchmarks: Turing’s theory predicted chances of a 70% reduction in the human ability to distinguish between machines from humans by the year 2000. And now 20 years later, we are ready to question, the already available models to build newer ones that require enhanced changes as overtime, newer tools and data handling schemes changed the complete perspective of building ML models. To know more, visit here.
  • Doubts over ANN models: The article describes the struggles of researchers to explain their problem-solving agenda using Neural Networks due to lack of knowledge to the clients. Moreover, the ML experts too are not sure about the working of Neural Networks and just the hypothetical assumptions that are leading the AI research. To know more, visit here.

Articles for week 01 of 2021

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

  • Visualizing NN: An amazing visualization tool for representing NN is PlotNeuralNetwork. The tool produces Latex Code for every NN built using this library To get more details, visit here.
  • Top CV human Pose estimation projects: The article covers till date best amongst the rest computer vision Pose estimation projects which might help to take off the research in this field. To find out the pointers to these projects, visit here.
  • Learn Neural Networks from Scratch: The MLP training and implementation is the theme of this article. Seems to be a good start for beginners. To learn more, visit here.
  • MODNet for background removal: The MODNet is a light-weight faster model for removal of the background using Deep Image Matting. To read more, visit here.
  • Review of MIT course on Datascience: The blog post is a review of MIT 6.0001 on CS and python and to read the full review, visit here.

Softwares

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

  • Benchmarking OpenBLAS on MacBook: The GitHub repository consists of test results of vecLib and VORTEX executed on Macbook M1. To know more, visit here.
  • Shortformer: The Github repository has a Shortformer model which is based on Staged training and Position infused attention and caching.

Videos

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

Research

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

  • PIC-GAN: Parallel imaging (PI) along with GAN (PIC-GAN) is proposed for improving MRI reconstruction in a multi-channel environment. To know more, read here.
  • Transformers for High-resolution Image Synthesis: The research published recently uses CNNs to learn rich vocabulary from the image and use transformers to build high-resolution images. To get more details, visit here.

Miscellaneous

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

  • KUHack Fest 2021: The Hackathon based on AI, IoT, Cloud, Blockchain is due to be held on Feb 19th for 48hours. To register visit here.
  • Machine Learning- from Training to Production: This is a Facebook event on 14 January 2021 and registrations are open.

That’s all for the week 01 of 2021. In case you think we have missed out on 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.