Machine Learning is Helping in a New Era
Modern machine learning methods have been around for more than 10 years, but it seems that recently there is a trend to add machine learning to a wider range of applications including medicine, agriculture, and even IoT devices. This could be triggered by the big capacity of the mobile devices, chips specialized in AI and also because the development platforms are more accessible, and the generated models are now more compact. For example one of the biggest accomplishments announced at Google I/O 2019 is that all the language processor will fit inside a phone, and not even a high-end phone. That means that our mobile devices without internet connection will have all the text to speech functionality, reducing the network latency and associated cost. So this compact models that can fit in small devices made them indeed more self-sufficient, which is triggering the possibility to create applications that make use of AI models in real-time. Moreover, we now are starting to see nice examples of that from a crop disease detection in Africa to real-time learning to dance app.
Also, the platforms are getting more accessible, TensorFlow is now Google’s open source cross-platform machine learning framework. It covers all the steps in building a learning model. There are many built-in operations to develop, train, and develop machine learning and deep learning, but if someone needs to add specific operations or customizations, it is also allowed to do so. If using Firebase, then probably ML Kit would be the best option since it is fully integrated and optimized for mobile.
So which scenarios are suitable for machine learning? As humans, if we want to learn something, we need to try it and see if it works or not, or we can learn by someone else’s experience. Same with computers, if we have a scenario where we are able to tests our inputs (the faster, the better) or if we have data already classified, and the most important we don’t know exactly how to obtain the results then machine learning could be an excellent candidate to find that function for us. Moreover, perhaps the machine learning algorithm could find relationships that for us were not evident before. For example, Google used deep learning for detection diabetic eye disease which was very successful in classifying the health and unhealthy images but not only that, after training the neural network for age and gender, they found that that information could be retrieved with those images too. So not only the network was able to perform the task correctly; it also found a relationship that was no evident even for an expert.
I think it is a great moment to revisit if a solution could make use of machine learning pretty much on any platform from ML on Firebase to Tensor Lite for embedded microcontrollers.
Related Blog Posts
Now more than ever before, businesses need to be a part of the mobile space. The Flutter framework provides one possible solution to digital transformation for a business.
Apple will release iOS 13 on September 19 2019. Learn about changes necessary to update your mobile application for iOS 13 and Xcode 11.
Kotlin has been adopted by many Android developers and companies because it offers a concise programming syntax which makes developers more comfortable writing code, prevents the common errors seen when developing in Java, and is easy to switch from iOS development given that Kotlin syntax is very similar to Swift.
TensorFlow is a Machine Learning cross-platform that has started to be adopted widely worldwide. It was released by Google in 2015 and now TensorFlow 2.0 Alpha is available.
ARCore is Google’s platform for building Augmented Reality experiences using Android, Unity, Unreal, or iOS as development environments.
GET IN TOUCH!
Copyright © 2019 Krasamo Inc. All rights reserved. All Trademarks are the property of their respective owners.