Stay Informed: The Latest in Machine Learning and AI
Written on
Chapter 1: Introduction to ML UTD #13
Welcome to the thirteenth edition of the Machine Learning Up-To-Date (ML UTD) newsletter from Life With Data! We aim to help you navigate the overwhelming landscape of software engineering and machine learning, filtering out the noise to deliver the most pertinent updates.
Life With Data is committed to providing well-curated updates on machine learning and software engineering that highlight crucial advancements without unnecessary details. This approach allows for frequent, succinct updates across the sector without causing information overload.
Section 1.1: Key Applications and Updates
The latest edition features several noteworthy topics:
FastAI Version 2: FastAI has launched its upgraded version, which is now built on PyTorch. Key enhancements include:
- A new type dispatch system for Python along with a semantic type hierarchy for tensors.
- A GPU-optimized computer vision library that can be extended using pure Python.
- An optimizer that condenses modern optimization functions into just 45 lines of code.
- An innovative two-way callback system that can access and modify any part of the data, model, or optimizer during training.
- A new data block API, among other improvements.
Easy Crawling of Government Data: Web scraping can be streamlined by utilizing sitemaps provided by large websites. The govinfo site offers easy access to a variety of data sources, simplifying the process of data collection.
GAO Report on Airport Facial Recognition: The U.S. Government Accountability Office released a report highlighting several concerns with facial recognition technologies used in airports, including outdated information, incomplete datasets, and demographic biases. These challenges underline the complexities involved in deploying machine learning applications effectively.
AI Timelines Update by Alex Irpan: Alex Irpan revisited his predictions regarding the timelines for achieving artificial general intelligence (AGI), suggesting they may be shorter than previously thought, along with explanations for these adjustments.
Self-Organizing Autonomous Vehicles: Research from Bar-Ilan University demonstrates that self-organizing autonomous vehicles can enhance traffic flow, even when human drivers dominate the roads. This raises important questions about the interaction between autonomous and human drivers.
Critique of OpenAI’s Language Model: An article by Gary Marcus and Ernest Davis effectively counters the hype surrounding GPT-3, emphasizing the importance of maintaining a grounded perspective in the machine learning community.
Section 1.2: Staying Updated
That wraps up ML UTD #13! However, the pace of change in academia and industry is rapid. To stay informed, check out the Life With Data blog, articles on Medium, and follow relevant discussions on Twitter.
Chapter 2: Video Insights
To further enhance your understanding of these topics, here are two informative videos:
How to Stay Up to Date on ML Research: This video provides strategies for keeping track of the latest advancements in machine learning research.
Project 8: Gold Price Prediction using Machine Learning with Python: This tutorial dives into a practical application of machine learning through a project focused on predicting gold prices.