They finally did it: They made “artificial intelligence” a buzzword. Typically, buzzwords don’t come from decades-old evolving disciplines of computer science. Making “machine learning,” “AI” and “neural nets” into buzzwords means that millions of developers are likely having their first experience with this stuff now.
In that vein, we bring you a nice long list of machine learning, deep learning, neural network and artificial intelligence how-to’s. Buzzword or not, it’s fairly obvious this stuff will be a big part of enterprise software for the next few decades. Time to get on board.
For those just starting out on the road to deep learning, GitHub user Flood Sung has compiled an excellent bibliography for taking developers from fledgling beginners to experts who can optimize an AI application.
François Maillet has taken the time to build his own blog on AI and machine learning. There’s some good stuff in there, though a lot of it has to do with his Machine Learning Database. The most recent post describes how to build an image classifier.
“Rejection Sampling Variational Inference,” a paper by Christian A. Naesseth, Francisco J.R. Ruiz, Scott W. Linderman, and David M. Blei, is super relevant for developers working on complex probabilistic models.
That previous paper goes well with this one by Soham De, Abhay Yadav, David Jacobs and Tom Goldstein. Titled “Big Batch SGD,” it seeks to describe a method of building automated inference using adaptive batch sizes.
For the tools fans out there, version 1.0 of spaCy was released on Tuesday. It’s an NLP library that’s focused on speed.
When it comes to deep learning, it’s all about the algorithm. Here’s a comparison of five algorithms for training neural networks.
In the “good tips” department, this blog entry offers some advice for processing data in a deep learning scenario. Specifically, it advises readers not to interpret linear hidden units, as they don’t exist.
If you’re too lazy or don’t have the budget to build your own neural network, perhaps you can use the Artificial Intelligence Open Network.
More research: “Learning in Implicit Generative Models,” by Shakir Mohamed and Balaji Lakshminarayanan.
And finally, for those of you working on image recognition, modification and normalization, here’s an excellent blog post by Augustus Odena, Vincent Dumoulin and Chris Olah about deconvolution and checkerboard artifacts in images.