Here at AssemblyAI, our team is always looking for ways to keep up with the latest Machine Learning and Deep Learning research to help fine tune our models and push our Speech-to-Text API to the next levels of accuracy and usefulness.
In fact, this research helps us ship weekly model updates that directly improve our API’s functionality for our customers.
Today, our team has compiled the top five Machine Learning blogs they reference as part of their continual research, including Distill, Machine Learning Mastery, ML CMU, Neptune Blog, and Hacker News. These blogs are also great resources for those just setting out in Machine Learning, as many offer excellent foundational content as well.
Machine Learning Blogs:
Below, we break down each Machine Learning blog and why we find them such a helpful resource.
Distill is an online research journal that presents itself as an alternative to academic publications. Content on the website is sorted by peer reviewed, thread, commentary, and editorial, depending on your interest. Its peer reviewed articles are registered with the Library of Congress and CrossRef and appear in Google Scholar.
What makes Distill particularly unique is its use of interactive media to “distill” the latest findings in Machine Learning.
Recent blog posts include:
Understanding Convolutions on Graphs
A Gentle Introduction to Graph Neural Networks
Multimodal Neurons in Artificial Neural Networks
Visualizing Neural Networks with the Grand Tour
Read more articles on Distill here.
Machine Learning Mastery
Founded by Jason Brownlee, who holds a Ph.D. in Artificial Intelligence, Machine Learning Mastery offers more practical, hands-on articles aimed at developers. Topics covered include Machine Learning, Deep Learning, Computer Vision, Neural Net Time Series, Natural Language Processing (NLP), GANs, LSTMs, Algorithms, Python, Ensemble Learning, Imbalanced Learning, Data Preparation and more.
In addition to the blog, Machine Learning Mastery offers a series of eBooks and guides--from foundational to advanced--to assist developers in their Machine Learning journey.
Recent blog posts include, written by both Brownlee and contributors, include:
Using a Singular Value Decomposition to Build a Recommender System
Principal Component Analysis for Visualization
Optimization for Machine Learning Crash Course
How to learn Python for Machine Learning
Calculus in Action: Neural Networks
Read more articles on Machine Learning Mastery here.
ML CMU is Carnegie Mellon University’s dedicated Machine Learning blog. The goal of the blog is to provide content that is accessible to a diverse audience, from the general public to advanced researchers. Its biweekly publications are written by CMU students, postdocs, and faculty, and often include findings from academic research performed at the university.
Recent blog posts include:
The Limitations of Limited Context for Constituency Parsing
Strategic Instrumental Variable Regression: Recovering Causal Relationships from Strategic Responses
A Unifying, Game-Theoretic Framework for Imitation Learning
PLAS: Latent Action Space for Offline Reinforcement Learning
A Learning Theoretic Perspective on Local Explainability
Read more articles on ML CMU here.
The Neptune Blog was created for research and production teams that run frequent experiments. In addition to blog content, Neptune helps developers perform experiment tracking and model registry to aid research and development. With Neptune, you can log and display Machine Learning metadata, organize experiments and training runs, compare experiments and models, share results, and more.
Its blog is a wealth of insightful information across Machine Learning experiment tracking, model management, MLOps, tools, and more.
Recent blog posts include:
How to Choose a Learning Rate Scheduler for Neural Networks
Depth Estimation Models with Fully Convolutional Residual Networks (FCRN)
Retraining Model During Deployment: Continuous Training and Continuous Testing
Debug and Visualize Your TensorFlow/Keras Model: Hands-on Guide
Read more articles on Neptune Blog here.
Run by Y Combinator, Hacker News is a social news website a la Reddit. While not an exclusive Machine Learning resource, Hacker News houses an abundance of user-generated or shareworthy Machine Learning content for users to sift through. Other users can upvote content and comment to rank articles and increase visibility, or simply use the search function to search for particular Machine Learning content.
Recent Machine Learning posts include:
Dendritic Computing: Branching Deeper into Machine Learning
Ask HN: Does Machine Learning Always Need Huge Data Sets?
Data Science: Future Trends in AI, Big Data, and Machine Learning
What Can Go Wrong with Machine Learning?
How to Train Large Deep Learning Models as a Startup
Read more articles on Hacker News here.
Sourcing the Best Machine Learning Content
The field of Machine Learning is constantly evolving, perhaps more rapidly now than ever before.
Whether you’re just getting started in the field of Machine Learning or are a serious developer looking to stay on top of current research trends, these five blogs can be insightful resources to reference.
Want to play around with our API instead? Sign up here!