Make sure you know the difference between all of the machine learning services out there.

Signal to Noise: Understanding the State of Machine Learning Services

Share on FacebookShare on Google+Tweet about this on TwitterShare on LinkedIn

Getting past the buzzwords to the meat of what is on offer in the machine learning services landscape isn’t easy. It’s also sometimes difficult to separate tools from APIs, algorithms from services, the plausible from just plain speculative. That’s why we’ve put together a little guide to help you understand how you can start utilizing real and true machine learning to accelerate your business.

Open-Source vs. Machine Learning Services

Lots of buzzwords these days actually reference open-source machine learning platforms and libraries, which often require knowledge of computer science, statistics, linear algebra, calculus and sometimes even more arcane math to fully utilize. These are separate from the technologies built on top of them, but can get conflated in the media landscape. Here’s a quick run down of tools versus services.

Google

Tensorflow is the (mostly) open-source library that Google maintains and presumably builds their internal and external machine learning services with. Popular wrappers for Tensorflow include Keras, Edward, and Tensorflow-Slim.

Google Vision is their cloud-based machine learning API, which offers object detection, optical character recognition, safe-for-work indicators, natural language analysis and more.

Google’s DeepMind is responsible for the Inception deep-convolutional model, the RAISR algorithm, and AlphaGO.

Facebook

Facebook does not currently offer any public services to consume, but they did author the popular open-source machine learning library Torch, whose bindings were used to create PyTorch.

Amazon

Rekognition is Amazon’s commercial machine learning service, which offers object detection, celebrity recognition, image moderation, and other products.

Microsoft

Though not seen popularly as a strong machine learning company, their Microsoft Cognitive Services provides state-of-the-art OCR, as well as many of the same image analysis tools as others on this list.

Their research team came up with ResNets, an advanced deep-learning model for image recognition.

Clarifai

Winners of the 2013 ImageNet competition, Clarifai offers a variety of models focused on image analysis and recognition, including travel, celebrity, wedding and food detection, as well as many others.

If you’re feeling adventurous, you can even check out Google’s machine learning glossary, which though highly technical, will surely make you the star of your next dinner party.

 

We hope this guide will prove useful if you are looking to incorporate machine learning into your business, and please leave any thoughts or questions below.

Share on FacebookShare on Google+Tweet about this on TwitterShare on LinkedIn

Read More →