Trending this week: Learn Graph Transformers; Deep Learning is not enough: Deep Reasoning is the answer; Top Deep Learning books to read in 2021.
Every week we analyze the most discussed topics on Twitter by Data Science & AI influencers.
The following topics, URLs, resources, and tweets have been automatically extracted using a topic modeling technique based on Sentence BERT, which we have enhanced to fit our use case.
This week, Data Science and AI influencers on Twitter have talked about:
- Machine & Deep Learning Resources
- Must-Have in Data Science & ML in 2021
- Explained: ML, Deep Learning, And Explainable Deep Learning
The following sections provide all the details for each topic.
Machine & Deep Learning Resources
This week, the influencers have shared amazing resources on machine learning and deep learning.
Dr. Ganapathi Pulipaka has shared a blog post talking about Graph Transformer, a generalization of Transformers to graphs. This blog is based on the paper A Generalization of Transformer Networks to Graphs. This post explained how to generalize transformer neural networks to graphs so that they can learn on graphs and datasets with arbitrary structure rather than just the sequential (as can be interpreted to be done by the NLP Transformers). It provides information on their design aspects and architecture, and also some remarks from their experimentation on some known benchmark datasets.
Additionally, Dr. Ganapathi Pulipaka has also shared a link to the Github repository of an amazing toolkit for the evaluation performance and robustness of machine learning classification models called PRESC. PRESC provides insights into model performance that extend beyond standard scalar accuracy-based measures and into areas which tend to be underexplored in application, including its: ability to generalize, sensitivity to statistical error, performance evaluation localized to meaningful subsets of the feature space, and n-depth analysis of misclassifications and their distribution in the feature space.
Finally, he also shared a very insightful post that provides some concrete examples of how to leverage the full power of SHAP values for understanding populations and events from data. This post explores two popular Kaggle datasets: Will it rain in Australia? and Is the car accident fatal? Some insights are extracted from those datasets and shared from visualizations. Then, all the code is available in a provided Kaggle notebook.
Must-Have in Data Science & ML in 2021
This week, the influencers have shared lists of new data science and machine learning books to read.
Dr. Ganapathi Pulipaka has shared a link to Top Natural Language Processing (NLP) Books and Top Deep Learning Books to read in 2021 available on the Left Bank Books website. Amongst others, you can learn state-of-the-art NLP approaches based on deep learning there.
He also shared an article presenting 15 PyTorch Books You Have to Read. Pytorch is an amazing Python framework allowing the implementation of state-of-the-art deep learning models for different tasks.
On his side, Andreas Staub shared a post that presents the 7 Best Python Libraries You Shouldn’t Miss in 2021. This article makes a tour of popular libraries that help to perform: data processing, data visualization, and machine and deep learning as well.
Explained: Machine Learning, Deep Learning, and Explainable Deep Learning
This week, MIT published an article answering questions like — What is machine learning? How businesses are using machine learning? And how machine learning works: promises and challenges.
This article created a lot of buzz in the data science influencers on Twitter and it was widely shared by influencers including — Bob E. Hayes, Nige Willson, Ronald van Loon, Marcus Borba, AI, and Spiros Margaris among others.
Ipfconline shared an interesting article explaining that Deep Learning is Not Enough: Deep Reasoning is the Answer. The article explains that relation networks can be plugged into deep learning models to generate relational reasoning functionalities.
They also shared interesting articles on Keras for deep learning and using it in real world problems.
Finally, Mike Tamir shared an article by Google: “A New Lens on Understanding Generalization in Deep Learning”. The article talks about how understanding generalization is one of the fundamental unsolved problems in deep learning, and can thus be mitigated.
He also shared an article on GeoGuessing with Deep Learning