Top updates related to Artificial Intelligence during week 51 of 2020, weekly edition of This week in AI
A summary of recent updates in the field of Artificial Intelligence (AI) for week 51 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 51 of 2020
This section is for recent industry updates related to Artificial Intelligence.
Multi-Expert Learning Architecture: (MELA) The training of multiple DNN to obtain the combined benefits of individual networks forms the basis for MELA. It is developed by a combined effort by researchers from the University of Edinburgh and Zhejiang Univerisity. Read complete coverage here.
Inductive Biases for DL and Human Cognitive abilities: A mentored technique by Yoshua Bengio and of course Anirudh Goyal’s effort has led to a proposal of inductive biases for improving the qualitative performance of DL models. Here is the article in more detail.
Brain mapping through machine intelligence: The research offers powerful insights that will help to understand the brain’s wiring for interpreting neurological disorders like Parkinson’s. The research originating at Japan’s Brain Science Project extends machine intelligence by improving the accuracy and reliability of brain mapping techniques. Here is the detailed coverage of the research work.
Interpreting code versus interpreting language: The neuroscientists have concluded that code interpretation isn’t similar to reading a language as it activates different parts of the brain. Like in the case of reading computer code, a general-purpose brain network is activated while language processing centers are activated in the latter task. To know more, visit here.
Zetane’s effort to accelerate AI adoption: The Zetane Viewer is a digital tool meant to display, inspect, and share datasets and models for helping anyone looking forward to working in AI and ML. To know more, visit here.
RoboGym: OpenAI’s robogym is a framework to train robots and improve their abilities. Here is the detailed article for a quick read.
Understanding Neuromorphic Computing: The aim of introducing Neuromorphic Computing is to mimic the human brain for better interpretations of problems and improve accuracy. To know more, read the entire article here.
Memory augmented ANN architecture: The research from Technische Universität Darmstadt has developed a new memory augmented ANN architecture for problem-solving. Here are more details on the article.
AI co-pilot tested at a US base in California: The U2 spy plane was a part of the successful flying AI co-pilot plane marking in the history of the Department of Defense. Here is coverage of news for a glance.
NASA suggests deploying Robodog to Mars: The robodogs developed by Boston Dynamics are suitable for long-distance and multi-terrain similar to the Mars rover. Thus NASA is planning on sending a modified version of this robot to locate locations on Mars. To know more visit here.
Articles for week 51 of 2020
This section is for recent Blog Posts/Tutorials related to Artificial Intelligence.
PyCaret: The PyCaret is an ML library that is a low code resource for ML researchers to skip the hypothesis to insight time. This introductory article can be read here.
XLnet: This a python notebook that uses XLNet for Tunisian sentiment analysis. The model supports PyTorch and Tensorflow.
A mathematical approach for Neural Abstraction: The article describes the functional connectivity of the brain along with spectral graph theory. To know more, visit here.
Seeing AI: The developments in Computer Vision (CV) has been phenomenal with addition through Seeing AI, a camera app for the visually impaired to manage their personal and business data efficiently. So basically it employs CNN with RNN Pool operator and here is the main post by Microsoft.
This section is for recent Tools/Applications related to Artificial Intelligence.
Hypersim dataset: It is an indoor scene dataset created synthetically by professionals with more than 77000 images taken from 461 indoor places. Here is the GitHub repository in case you wish to access the dataset.
This section is for recent Lectures/Conferences/Webinars related to Artificial Intelligence.
Precision, Recall, and F1 score: The ML metrics such as precision, recall, and F1 score are explained in this video. Along with that, Binary classification, use cases, and multi-class classification are also discussed.
This section is for recent research published related to Artificial Intelligence.
Face ID Transformer for anonymization: The research proposes FaceID along with the password for anonymization, privacy, and accessibility. Additionally, a 1:1 password to face ratio is proposed in cases where the wrong password leads to the production of face image animosity leading the hackers to nowhere. To read the article, visit here.
Real-time plant assessment using DeepLens: The reduction in economic losses and conservation of plant species is possible through automatic detection and classification of leaf and plant illness using AWS DeepLenns and DCDM. The model was able to detect 98.78% of diseases accurately and can be accessed for reading.
Sketch generation: The article describes an image to pencil sketch translation along with the drawing process. To know more, visit here.
HR depth estimation (DE): The research published explains self supervising image sequences for monocular depth estimation. The Depth-Net, HR Depth are used to improve depth estimation with the GitHub code available.
Understanding NLP: The article discusses the usage of music and source code could improve accuracy against natural language. Additionally, simple AI structures might prove beneficial for transfer learning while grammatical structures might not be needed for learning the vocabulary.
This section is for updates for other various related topics to Artificial Intelligence.
Register for “All About AI”: The first episode of All About AI is set to be aired on 27 January 2021 and the registrations are open and free. To register, visit here.
MIT’s ML course for free: A 13 weeks ML course that covers principles, algorithms, and applications areas are available for free for encouraging interest in supervised and reinforcement learning. To know more, visit here.
ML in Physical Sciences: This year’s Compute Fest is due to be held on 21 January 2021 which is organized by Harvard Institute for Applied Computational Science. The session is based on workshops, labs based on deep learning and is open for registration.
That’s all for the week 51 of 2020. 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.