Start Arya.ai Monday presented Braid, an open source tool available free of charge for companies developing neural networks.
Braid is a flexible, customizable, and modular meta-framework that works with operating systems for deep learning, according to the company. It is designed for rapid development and to support arbitrary network designs.
It’s simple and scalable, for use with networks that need to handle many high-volume data points, Arya.ai said.
Braid allows for quick experimentation without having to worry about one-size-fits-all code components, the company said. Users can create complex architectures by drag and drop or by coding.
Developers can use Braid to customize or add existing neural layers, according to Arya.ai.
Interlace with Vega
Braid works in collaboration with Vega, Arya.ai’s end-to-end platform that simplifies complex artificial intelligence processes and enables users to build and deploy deep learning algorithms at scale.
Vega is interoperable with multiple data platforms and can work with structured, unstructured or hybrid datasets. It can integrate with SQL, MongoDB and NoSQL databases. It is compatible with S3, Drive and Box storage systems. It also integrates with Hadoop, MapReduce and Redshift HPC clusters.
Users can deploy Vega-based algorithms in their data centers using an integrated provisioning layer or on Arya’s cloud. Vega offers hardware recommendations and allows users to monitor their applications and scale as needed.
Vega’s workbench can access training data sources, ingest live data and learn from it continuously through feedback loops at set frequencies.
The work plan allows collaborative modeling among multiple members, and a dashboard allows companies to monitor all development activity and ongoing deployments through a single interface.
Deep learning use cases for Vega include word processing, image processing, predictive analytics, and the Internet of Things.
Difficult to stand out
“Neural networks are very complex and very difficult to configure,” noted Jim McGregor, senior analyst at Tirias search. “You need different kinds of network setups for every kind of learning you’re trying to do. “
Arya.ai “is trying to create a tool to help you model networks and choose the one that best suits your purpose,” he told LinuxInsider. His claims are “grandiose – but having said that, there is a lot of history with the development of network technology, and I’m sure some of their algorithms apply.”
Frameworks that help developers build smart applications are “very popular today,” noted Holger Mueller, senior analyst at Constellation research.
“The challenge is that there are almost more options to work with than there is time to build these apps,” he told LinuxInsider, “so reduce the number of vendors and find winners or losers is a tall order.
FOSS muscle flexion
The FOSS approach will likely prove to be a major asset for Arya.ai.
“Open source is the winner as a co-development model for modern infrastructure,” noted Mueller. “Even the largest providers rely on open source. “
Open source is already well represented in AI: Google offers TensorFlow; the University of Washington offers Alchemy’s open source AI; and OpenCog, the Open Cognition Project, aims to create an open source framework for general artificial intelligence.
However, “a company’s open source line of code does not make it a successful open source project,” Mueller warned. “It requires a community of developers, committed resources, etc. It is too early to judge the success of this offer from this angle.
Where are the dollars
Taming mountains of raw data is “the real problem with artificial intelligence,” suggested McGregor de Tirias. “Everyone is trying to figure out how to be the aggregator. “
Arya.ai’s FOSS offering “looks like a business model game,” he suggested, “where they’re trying – whether or not the tools are as powerful as they claim – to bring people to register and work with them. The big advantage is not necessarily your business, but being that benchmark data aggregator. “