Leveraging machine learning to process data and workloads has proved to be significantly beneficial for diverse enterprise industries in recent years. Whether it be healthcare, BFSI or retail, machine learning systems turned out to be extremely promising to process millions of data and build complex models. Having said that, the traditional machine learning process involves humans to look after the operations, to code, and to build the models. But, with the crisis in hand, businesses are looking to reduce their workforce, some are even not equipped with resources to spend on employing an experienced data science team. And that’s when AutoML can come to rescue for many.
AutoML, aka automated machine learning, is an automated end-to-end process of deploying machine learning techniques to solve complex business problems and build models. Well, the concept of automated machine learning isn’t novel; instead, it has been the talk of the town since Google first released its AutoML product. Not only AutoML can be used in image recognition and NLP tasks, but also for speech recognition, semi-supervised learning and reinforcement learning.
With the massive potential of AutoML making a mark in the industry, it is the right time for ML practitioners, data scientists as well as non-tech professionals of the organisation to get a more comprehensive understanding of AutoML. Thus, in this article, we will share seven online resources that can help you get your hands-on automated machine learning.
Also Read: Top 5 Books On AutoML To Streamline Your Data Science Workloads
AutoML Vision API Tutorial
By Google Cloud
About: This tutorial of AutoML Vision API Tutorial is provided by Google Cloud, demonstrating the right way to create a new model with your own dataset of training images. This tutorial will also further help in evaluating the results and predicting the classification of test images using AutoML Vision. This tutorial by Google Cloud will be using an image dataset of five different kinds of flowers — sunflowers, tulips, daisy, roses and dandelions, and stages like training the custom model with the dataset, evaluating the model performance for better accuracy, and classifying new images using the custom model.
Check the tutorial here.
Also Read: What Are The Limitations Of AutoML
AutoML Natural Language API Tutorial
By Google Cloud
About: AutoML natural language API tutorial is also provided by Google Cloud, where it showcases steps to create a model that can classify content leveraging AutoML natural language. In this tutorial, it will be highlighted how the application trains the custom model with an open-source data set HappyDB. This would result in a model which would be able to classify the happy moments into specific categories stating the cause of happiness.
Check the tutorial here.
Also Read: Addressing Drawbacks Of AutoML With AutoML-Zero
Using AutoML To Predict Taxi Fares
By Microsoft Azure
About: In this tutorial, learners can get their hands-on using AutoML in Microsoft Azure machine learning to develop a regression model to predict taxi fares in New York City. This is a tutorial for a specific case study and will be using training data and configuration settings to automatically surf through normalisation/standardisation methods and hyperparameter settings to develop the best model for predicting the fares. Along with GitHub, this tutorial can also be run on learners’ local environment.
Check out the tutorial here.
Also Read: Why 2020 Will Be The Year Of AutoML
AutoML Tables Tutorial Notebook
By Kaggle
About: AutoML Tables Tutorial Notebook is a step by step tutorial of using Kaggle’s new integration with Google’s AutoML Tables, provided by Devrishi, a product manager at Google. In this tutorial, the presenter will apply Tables on one of the Kaggle Competitions — Housing Prices, where the data will be used to predict the sale price. This tutorial will showcase how leveraging AutoML Tables will help in getting an initial high performing solution without spending much on feature engineering, model selection or hyperparameter tuning.
Check out the tutorial here.
Also Read: AutoML Is Functional But Have Limitations That We Cannot Ignore
AutoML capabilities of H2O library
By Kaggle
About: AutoML capabilities of H2O library is a tutorial presented by Kaggle, where the presenter — Dmitry Burdeiny showcases an overview of AutoML capabilities of H2O library. The H2O AutoML interface has been designed with few parameters making it easier for users to point to their dataset, identify the response column and optionally specify a time constraint to train their model. With this tutorial, learners will get hands-on how to use AutoML capabilities of H2O library, the open-source ML and AI platform.
Check out the tutorial here.
Also Read: Does AutoML Work For All Data Science Stakeholders
Auto ML with Auto-Keras
By DataCamp
About: Auto ML with auto-Keras is a tutorial provided by DataCamp, which will teach learners how automated machine learning can be done with the auto-Keras library. It covers understanding the machine learning pipeline and automating that to the introduction of AutoML and auto-Keras. It also has a case study application. From initialising ML models to training, testing and evaluating the model, this tutorial will provide comprehensive knowledge on the level of abstraction auto-Keras can offer and how easy it can be to use it for machine learning.
Check out the tutorial here.
Also Read: 10 Popular AutoML Tools Developers Can Use
Creating Machine Learning Models With AutoML
By Macgyver (YouTube)
About: Lastly, as the name suggests, this YouTube tutorial is showcasing how to use AutoML to create ML models without writing any code. In this video tutorial, the presenter will teach how to generate some artificial data using Google Sheets and App Scripts, and then upload the same to a Google Cloud Storage Bucket. Post that, the tutorial will teach how to use Tables/AutoML to generate a classifier based on the tabular data. Once that’s done, the presenter will show how to deploy the model and run some prediction on new data records to evaluate the model’s accuracy.
Check out the tutorial here.