Over a million developers have joined DZone.
{{announcement.body}}
{{announcement.title}}

How Google's AutoML Takes Machine Learning A Step Further

DZone's Guide to

How Google's AutoML Takes Machine Learning A Step Further

We are looking into the capabilities of AutoML and are investigating ways to integrate with it. How do you think machine learning will help transform your business?

· AI Zone ·
Free Resource

EdgeVerve’s Business Applications built on AI platform Infosys Nia™ enables your enterprise to manage specific business areas and make the move from a deterministic to cognitive approach.

Last week, Google announced an alpha launch of AutoML, a new machine learning platform that takes the space to the next level. In the product announcement, AutoML is described as a service that allows users to train machine learning models without requiring an in-depth knowledge and proficiency in machine learning. The potential of this platform is very appealing:

  • It promises to eliminate the mental and financial barriers that smaller companies face when attempting to integrate state-of-the-art machine learning into their platform.
  • It will move machine learning from behind its theoretical walled garden, still mostly accessible only to those with academic backgrounds or the capital to hire top candidates, into the space of the real and practical.
  • It will help business align human intellectual capital to innovate and differentiate.

On a more technical level, AutoML is just as impressive:

  • AutoML utilizes transfer learning to get its users' models trained without loading large amounts of data. Transfer learning allows models to obtain knowledge from other models as opposed to gaining it from raw data. This will cut down on time required to make informed business decisions.
  • AutoML provides a platform for automating object detection and image tagging. These features are not new to the Google platform or its competitors — Clarifai, Hive AI, and Microsoft Cognitive Services also offer these services. AutoML differentiates itself from the rest by empowering users to provide the platform a sample set of images with specific tags called out. The system learns about these images and the tags therein. Once the model is trained, an unseen sample will accurately detect and tag the items based on the learned model.

The Filestack Take

Filestack loves AutoML's focus on simplicity and efficiency. However, we believe the industry needs automated, easy solutions for the more difficult use cases. For example, what if the use case requires object detection, object localization and optical character recognition (OCR) to work together? 

Filestack provides a platform where custom models utilize multiple ML services. All the integration needs is a couple of lines of code within our customer's software stack. We provide our customers with the most compelling feature set across all the leading machine learning platforms. Customers integrate Filestack's API into their software stack to ensure that edge content gets pushed to the cloud reliably, securely and intelligently. It is processed, transformed and analyzed to ensure that it is in the exact state that the application expects it.

We recently announced customized trained machine learning models for deriving relevant information from physical mail envelope scans as well as from scanned bank checks. This is what Steven Maguire, VP of Technology at Earth Class Mail, had to say about his experience with this service:

"A simple integration with Filestack's content intelligence API allowed us to offload the analysis of a large volume of mail quickly and reliably. This automation has proven critically important as our team has more time to focus on enhancing our core product."

We are looking into the capabilities of AutoML and are investigating ways to integrate with it. In the meantime, we'd love to hear your thoughts on how machine learning may help transform your business.

How do you think ML will change the game for you?

Adopting a digital strategy is just the beginning. For enterprise-wide digital transformation to truly take effect, you need an infrastructure that’s #BuiltOnAI. Click here to learn more.

Topics:
ai ,automl ,machine learning ,training data ,transfer learning ,object detection

Published at DZone with permission of

Opinions expressed by DZone contributors are their own.

{{ parent.title || parent.header.title}}

{{ parent.tldr }}

{{ parent.urlSource.name }}