Application development that incorporates advanced technologies like AI and ML is continuing to evolve — this time through integrating various deployment results into one general-purpose, no-code platform. This way, line-of-business users can simply click from analyzing data records over to natural language processing, and then to imaging and video outputs.

This kind of versatility for non-IT employees hasn’t been available in the market, since so many software makers are able to focus on only one of those outputs at a time. Market leaders such as DataRobot, Amazon Web Services, Microsoft, DataBricks, and SAS do not provide this exact functionality. H2O.ai however, has set out to solve this.

Mountain View, California-based H2O.ai today unveiled a key new addition to its open-source-based platform: H2O Hydrogen Torch. This feature is a deep-learning training engine that it claims smooths the way for companies of any size in any industry to make state-of-the-art image, video, and natural language processing (NLP) models without coding. These models can be used in production for discovering new business insights about customers, competitors, the market, and other areas of interest.

Until now, creating deep-learning models has required extensive knowledge and time to code and tune accurate models. These investments can be expensive because data scientists are among the highest-paid specialists in the IT world. H2O Hydrogen Torch was developed by the world’s best data scientists, Kaggle Grandmasters, and the challenging parts of creating world-class deep learning models are handled automatically by the product, CEO and cofounder Sri Ambati told VentureBeat. 

Through a simple, no-code user interface, Ambati said, savvy LOB employees, data scientists, and developers can rapidly make models for numerous image, video, and NLP processing use cases, including identifying or classifying objects in images and video and analyzing sentiment or finding relevant information in texts.

Video use cases, for example, would include monitoring foot traffic in public buildings, malls, and stores, noting the frequency of visitors and where they move from location to location. Retail stores can use the platform to see which sales displays attract the most attention. All of the data is compiled immediately and made available for queries in the H2O.ai analytics engine, Ambati said.

“There’s so much unstructured data out there and specifically in the area of images and text in companies,” Ambati said. “There’s a lot of untapped potential. The goal is really to allow and enable users to really build state-of-the-art models for different types of use cases. Basically, in Hydrogen Torch, we are really giving them these capabilities to tackle different types of use cases.”

According to multiple analyst estimates, 80% to 90% of data is unstructured information, yet only a small percentage of organizations can derive value from unstructured data, Ambati said. 

Deep-learning models provide the ability to unlock opportunities to transform industries including health care with computer-aided disease detection or diagnosis through the analysis of medical images; insurance with the automation of claims and damage analysis from reports and images; and manufacturing by utilizing predictive maintenance by analyzing images, video, and other sensor data, Ambati said.

Image and Video Processing

For images and videos, Hydrogen Torch can be trained for classification, regression, object detection, semantic segmentation, and metric learning, Ambati said. In a medical setting, for example, Hydrogen Torch can analyze medical X-ray images for abnormalities with a “human in the loop” to make the final decision. Other image-based use cases include object detection in a manufacturing facility to determine whether a part is missing or metric learning that alerts an online retailer to duplicate images on a website, Ambati said.

Natural Language Processing

For text-based or NLP use cases, Hydrogen Torch can be trained for text classification and regression, token classification, span prediction, sequence-to-sequence analysis, and metric learning. NLP use cases range from predicting customer satisfaction from transcribed phone calls to sequence-to-sequence analysis to summarize a large portion of text, like medical transcripts.

These models then can be packaged automatically for deployment to external Python environments or in a consumable format directly to H2O MLops for production, Ambati said.

H2O.ai’s platform, which currently offers a free trial, is used by more than 20,000 global organizations, Ambati said, including AT&T, Allergan, CapitalOne, Commonwealth Bank of Australia, GlaxoSmithKline, Hitachi, Kaiser Permanente, Procter & Gamble, PayPal, PwC, Unilever, Walgreens.