Categories
Articles

Top Twitter Topics by Data Scientists #37

Trending this week: Learn deep evolutionary reinforcement learning; 6 important questions you should ask while implementing an anomaly detection system; Know how to ace A/B testing in data science interviews.

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.

Want to know more about the methodology used? Check out this article for more details, and find the codes in this Github repository!

Overview

This post, dedicated to the technology watch based on posts from Data Science and AI influencers on Twitter, will cover various topics this week. In particular, we will take about:

  • Updates on deep learning (DL) & deep reinforcement learning (DRL)
  • Tips and tricks for machine learning projects
  • Facebook Becomes Meta

DL & DRL Updates

This week, data science and AI influencers have shared some amazing updates on deep learning and deep reinforcement learning.

Ronald Van Loon has shared an amazing article introducing Deep Evolutionary Reinforcement Learning (DERL), a new deep reinforcement learning technique that helps AI to evolve. DERL uses a complex virtual environment and reinforcement learning to create virtual agents that can evolve both in their physical structure and learning capacities. This will allow overcoming the currently limited flexibility of AI agents in some of the basic skills found in even the simplest life forms. DERL is introduced in the research paper “Embodied intelligence via learning and evolution” published in the scientific journal Nature by AI researchers at Stanford University.

The following collection has been shared by ipfconline:

A post talking about the real-life applications of reinforcement learning (RL), focusing on 10 Real Reward & Punishment examples. Here, various domains are addressed, namely the use of RL in: self-driving cars, industry automation, trading and finance, natural language processing, healthcare, engineering, news recommendation, gaming, marketing and advertising, and robotics manipulation.

This article scratches the surface as far as application areas of reinforcement learning are concerned, the main goal is to spark some curiosity that will drive you to dive in a little deeper into this area. If you want to learn more check out this awesome repo — no pun intended, and this one as well.

Finally, an article on machine learning, deep reinforcement learning, and their current limitations. This post discusses how the combination of ML with DL and DRL is bringing significant advances in AI. It also emphasizes the existing limitations to some specific applications. Here are some resources shared in this article, with associated links:

  • Triumph of Deep Reinforcement Learning: Deep Q-Network (DQN)
  • Market Success of Evolution Strategies in Reinforcement Learning
  • The Next Hurdle for DeepMind
  • Role of Deep Learning in AI: Misinformation is Possible
  • The Biggest Limitation of Deep Learning: Machines Cannot Provide Legal Explanations.

ML How-Tos

Here we provide you with some very useful resources on machine learning that will help you effectively solve various problems in your projects.

Ronald Van Loon has shared an amazing article indicating the Questions To Ask While Implementing Anomaly Detection System. This article provides the following list of six important questions to have in mind while building and deploying an anomaly detection system to ensure the deployment of the correct product for your needs:

  • What is the alert frequency (5 minutes/ 10 minutes/ 1 hour or 1 day)
  • Requirement of a scalable solution (Big data vs. regular RDBMS data)
  • On-premise or cloud-based solution (Docker vs. AWS EC instance)
  • Unsupervised vs. Semi-supervised solution
  • How to read & prioritize various anomalies in order to take appropriate action (Point based vs. Contextual vs. Collective anomalies)
  • Alert integration with systems

Taking into account these questions will help you select the appropriate configurations to obtain the best possible performances for your business.

An anomalous observation (Source: Unsplash).

Kirk Borne has shared an A-Z article on Outliers Detection and Handling in Machine Learning. This article goes through the concept of outliers in statistics and its application in the field of machine learning. It provides practical means to identify outliers in your data, and finally handle them.

This post provides the following mathematical and visual methods to identify outliers in your data:

  • Standard Deviation Method
  • InterQuartile Range Method
  • Control Charts
  • Box Plots

This post provides the following methods to deal with outliers in your data:

  • Dropping
  • Applying Mathematical Transformations
  • Applying Minkowski Error Method

This article is a good starting point to get to know how to address the problem of outliers.

On their side, KDnuggets has shared:

A mock interview that provides a step-by-step guide through how to demonstrate your mastery of the key concepts and logical considerations on A/B testing. This post will help you to understand the process of A/B testing and to learn its key concepts, and get you to know how to discuss this approach during data science job interviews.

A/B Testing (Source: Pixabay).

A post sharing some Streamlit Tips, Tricks, and Hacks that you can also use to unleash your powerful DS, AI, or ML applications. This article presents some tips and tricks on how to develop Streamlit applications. Some of these tricks may become natively available in the future versions of Streamlit, so that we may not need to do the hacks, or on the other hand, they may come with some updates to prevent the hacks given here. In the meanwhile, you can enjoy them for now, and hope for new amazing features in Streamlit.

Facebook Becomes Meta

The news of Facebook becoming Meta, created a buzz in the data science and AI influencers on Twitter.

They shared the news with their own opinion on it.

Carla Gentry tweeted an article that mentions how Zuckerberg wants to create a make-believe world in which you can hide from all the damage Facebook has done. It mentions that the name change heralds the reorganization of the Facebook empire around a single concept: the metaverse. Also explains that Meta will be made up of Facebook, WhatsApp, Instagram, Oculus, and then all the stuff that provides the metaverse.

Kirk Borne has shared an article that shares the news that Facebook is Now Meta. The article talks about the fate of Oculus — all Oculus branding is going away. Starting early next year, the Oculus Quest will be known as the Meta Quest, the Oculus App will be the Meta Quest App, etc. Meta aims to make sure everyone knows that the “brand formerly known as Oculus” is a Meta product and core to its metaverse future.