Product Feature

AutoML, short for Automated Machine Learning, is a rapidly growing field of artificial intelligence that aims to automate the end-to-end process of building, optimizing, and deploying machine learning models. With AutoML, individuals and organizations can leverage the power of machine learning without the need for extensive technical expertise, as the system takes care of many of the complex tasks involved in model building.  

AutoML tools use various techniques, such as neural architecture search, hyperparameter optimization, and automated feature engineering, to automatically generate and optimize machine learning pipelines. By enabling faster and more efficient model development, AutoML is poised to revolutionize the way we approach machine learning.

In this blog post, we will explore the AutoML feature integrated into Prederas AIQ, which provides users with comprehensive assistance in building and optimizing machine learning models. We will discuss the key aspects of AutoML in AIQ and provide a guide on how to leverage its capabilities effectively.

The initial step is to locate and access the AutoML page within the AIQ platform and click on New AutoML Experiment.

‍

Upon accessing the AutoML page, the user is prompted to enter several inputs, including  

• Name of the experiment  

• Brief description
• Type of model the user wishes to create
• Desired duration of each iteration

Finally, the user must upload the dataset to be used for model training and evaluation.

‍

‍

To initiate the process, the initial step involves assigning a name to the model and providing a description. AIQ provides users with a diverse range of model options to choose from when creating an AutoML model. The platform offers a list of various types of models that users can select according to their specific requirements.

The following is a comprehensive list of the various models that are available for users to choose from in AIQ.

‍

‍

• ‍Classification: This type of model is used to classify data into different categories or classes based on a set of input features.

• Regression: Regression models are designed to predict numerical values or continuous outputs based on a set of input features.
• Time Series: Time series models are ideal for analyzing and forecasting data that changes over time, such as stock prices, weather patterns, or economic indicators.
• Clustering: Clustering models group data into distinct clusters based on their similarity, making it useful for tasks such as customer segmentation or image recognition.

• Anomaly Detection: Anomaly detection models are used to identify and flag data points that deviate significantly from the norm, making them valuable for fraud detection and network security.

Once the modeling objective has been established, the user must then choose the iteration duration for the AutoML model.

After entering the necessary criteria, the user is required to select/choose the data file that will be utilized for model training purposes.

‍

‍

There is also an option to view a selected file as a data frame. The user needs to click on the Preview File Data to view the file selected.

‍

Once the data has been previewed, the user must click on the "Upload File" button to proceed with uploading the data.

‍

After uploading the file, the user will receive a message indicating the successful creation of the dataset.

The subsequent task involves clicking on Target Column to choose the desired column from the dataset, which will serve as the target or prediction variable.

‍

Clicking the target column will present a list of available columns, allowing the user to make a selection from a dropdown menu for the target column.

‍

Once the above criterions are decided and inputted by the user, the user needs to click on Submit for the AutoML to start functioning.

‍

After clicking the "Submit" button, a confirmation message will be displayed, indicating the successful creation of the AutoML.

Completing the input process, a summary page will display the newly created AutoML model's essential details, such as its name, type, creator, and creation date. Additionally, users can access previous versions of the model and review the input variables list, with the option to make necessary changes.

Upon accessing the AutoML project that the user has created, they will be displayed with a comprehensive page displaying summary information pertaining to the AutoML model. This page will provide an overview of various aspects, including any deployed models, registered models, and trained models associated with the current model.

‍

After training, an AutoML model is deployed and utilized in three distinct locations.

• Trained Models (Experiments)

• Registered Models
• Deployments

When a user creates an AutoML model, it generates ten versions of the model by employing various algorithms for training. The model exhibiting the highest efficacy is then stored in the Registered models and Deployments.  

Now, let's explore the multiple experiments conducted by AutoML.

‍

‍

From the provided list, it is evident that multiple versions of the model have been trained using various algorithms to identify the version with the highest efficacy. Subsequently, the model demonstrating the highest efficacy is selected and placed within the Registered models and Deployments.

The above display showcases the top-performing model from all iterations, which has been stored in the registered models. It provides detailed information about the version, algorithm, library, and other relevant details pertaining to the model.

‍

The above section presents the deployed models where we can observe the presence of the trained model within the deployed models.

Prederas AIQ's integrated AutoML feature provides users with not only efficient model building capabilities but also seamless deployment options. Once a model is trained and logged by AutoML, it can be deployed with a single click, enabling users to quickly operationalize their models.  

The REST endpoints of the deployed models can be easily accessed for inferencing purposes and seamless integration with other applications. This end-to-end workflow empowers users to not only build accurate and optimized models but also deploy them effortlessly, unlocking the full potential of their machine learning projects. Prederas AIQ's AutoML feature offers a comprehensive solution that streamlines the entire process, from model creation to deployment, making AI adoption accessible and efficient for users.

In conclusion, Prederas  AutoML feature offers users a powerful and user-friendly solution for automating the machine learning model building process. By providing an intuitive interface and a wide range of model options, AIQ simplifies and accelerates the development and optimization of models for various purposes, including classification, regression, time series analysis, clustering, and anomaly detection.  

With the ability to customize iterations and access previous model deployments, users have greater flexibility and control over their AI projects. By leveraging Prederas AIQ's AutoML capabilities, users can unlock the potential of machine learning and make informed decisions based on accurate and efficient models.  

‍