What is Create ML – The Apple Machine Learning Framework from Coding compiler. Create ML is a framework from Apple for machine learning. It allows the creation and training of machine learning models within the Apple ecosystem. The trained models can be easily integrated into own apps with Core ML. The models can be used, for example, for image, text or speech recognition.
What is Create ML – The Apple Machine Learning Framework
With Create ML, Apple provides a framework to build and train machine learning models using Apple technologies such as Swift or Xcode. The trained models are easy to integrate into your own apps and to use for Artificial Intelligence ( AI ) applications. Core ML is used to integrate the models.
[Related Article: What is Core ML ]
Machine learning becomes possible within the Apple ecosystem thanks to Create ML and Core ML, without the need for third-party software or services. Developers do not need special programming skills for machine learning. The trained models can be used for image recognition, speech recognition, text extraction or in general for finding patterns and relationships in data.
Swift scripts automate the creation and training of models. Models created with Create ML can be tested and trained on a developer’s local Mac computer. Dedicated servers are not necessary for this. A graphical user interface makes it possible to train the models, in which data files are fed via drag & drop.
[Related Article: Artificial Intelligence Trends ]
The structure of the underlying neural network and the algorithms of machine learning used need not be known to the developer. Create ML leverages Apple’s existing machine learning infrastructure, which is already used by applications such as Siri or Photos. Create ML was first released at the Apple Worldcup Worldwide Developers Conference in 2018.
[Related Article: Machine Learning vs. Deep Learning]
The functions of Create ML
The most important features of Create ML are:
- Training of machine learning models with image data, texts or numerical data
- Use Apple’s existing machine learning infrastructure
- Training and testing models on local Mac machines
- Create and train models using Apple technologies such as Swift or Xcode
- Integration of ready-made models into own apps
- Automation of creating and training with Swift scripts
- Possibility for experts to use complex algorithms
[Related Article: What is a Convolutional Neural Network?]
Process of training models with Create ML
Training a model with Create ML has a typical process. The model is supplied with representative data to learn to recognize the desired patterns. These training data can be images, texts, voice data or other data. After the first training of the model with this data, new, previously unused data are fed. The model tries to find the previously learned patterns in this data. If the results are not yet satisfactory, a new training phase with new representative data has to be carried out. The retraining and the subsequent evaluation with additional data can be repeated until the model achieves the desired recognition performance. Only then will the model be integrated into your own application.
[Related Article: The Deep Learning Framework]
The interplay of Create ML and Core ML
In addition to Create ML provides Apple Core ML. Core ML was presented at WWDC 2017 a year earlier. While Create ML is intended for creating and training, Core ML takes on the task of integrating the finished models into their own applications. Core ML can not only integrate models designed with Create ML, but also models from other frameworks such as TensorFlow or the IBM Watson Services platform.
[Related Article: What is Cognitive Computing? ]
Benefits of using Create ML and Core ML
Create ML provides application developers with the ability to integrate AI and machine learning functions into their own applications with little effort and deeper machine learning programming skills. The time to create and train a model is drastically reduced. The framework is easy to use and provides a graphical user interface that lets you perform many actions using drag & drop. The creation and training of a model with local data can be automated with just a few lines of Swift code. The training can take place directly on a local computer and does not require a powerful server.
[Related Article: Artificial Intelligence And The Blockchain ]
The entire process, from creation to training, to integration with your application, is entirely within the Apple ecosystem and does not require external or third-party software. Machine Learning and Artificial Intelligence will be available to the masses of iOS developers. If required, models created with other frameworks such as TensorFlow or Watson Services can be integrated with Core ML.
Related Technical Articles:
Artificial Intelligence Trends
Learn Walt Disney’s Approach To Data Analysis
SIEM Tools List For Security Information Management
Robotic Process Automation Data Migration Case Study
RPA – The Line Between Reality and hype