AutoAI: The Magic of Converting Data to Models

Visual Recognition in Watson Studio

Watson Studio Recognition service uses cutting-edge deep learning algorithms to analyze and classify images and detect objects and other content. It allows users to build a collaborative environment where machine learning engineers can work together remotely. It connects the projects with Github to make it easier to push and pull the changes. It supports different image formats such as JPEG and PNG. It also supports notebooks.

Natural Language Classifier

Watson Studio natural language classifier allows users to build text classifiers based on user-defined classes. It supports CSV (.csv) data files that contain samples of each class. It supports different languages including English, Arabic, French, German, Italian, Japanese, Korean, Portuguese (Brazilian), and Spanish. Watson Studio provides multi-category classification and multi-phase classification (up to 30 separate inputs in a single API request). It also supports APIs in notebooks.

Here is a sample model built using Watson Studio with three classes (hi, problem, question.

AutoAI Examples

In the following, I list several blogs that I built to guide you throw the steps of training and deploying different models using AutoAI:

Image Classification

We will use Stanford Dogs Dataset that contains images of 120 dog breeds from around the world. The goal of our classifier is to classify a dog image based on its breed.

Use pre-trained Image Classifiers

We learn in this blog how to reuse a pre-trained model by exposing them as an API for inference tasks.

Object Detection

In this blog, we learn how to build a model to detect cancer and fluids in brain scans.

Text Classification

We build a model for classifying different StackOverflow posts into different tags.

If you are new to image classification/deep learning, check my post to learn the basics of deep learning and understand the whole training process end-to-end in the following blog:

Learn more about Embeddings and text models

If you a beginner with textual data and would like to start from scratch, take a look at the following post

Conclusion

Machine learning engineers often spend considerable amounts of time searching for a proper model (reference model). Using AutoAI drastically improves and facilitates this step. It helps in building deep learning models easily providing your data only — without a single line of code.

Please let me know if you have any questions!

Resources

• IBM Watson Studio - AutoAI overview