Nvidia launches TAO, an enterprise workflow for AI development

During its GTC 2021 virtual keynote, Nvidia introduced a new product designed to help enterprises choose, adapt, and deploy machine learning models. Called TAO and available starting today in early access, it enables transfer learning as well as other machine learning techniques from a single, enterprise-focused pane of glass.

Transfer learning’s ability to store knowledge gained while solving a problem and apply it to a related problem has attracted considerable attention in the enterprise. Using it, a data scientist can take an open source model like BERT, for example, which is designed to understand generic language, and refine it at the margins to comprehend the jargon employees use to describe IT issues.

TAO integrates Nvidia’s Transfer Learning Toolkit to leverage small datasets, giving models a custom fit without the cost, time, and massive corpora required to build and train models from scratch. TAO also incorporates federated learning, which lets different machines securely collaborate to refine a model for the highest accuracy. Users can share components of models while ensuring datasets remain inside each company’s datacenter.

In machine learning, federated learning entails training algorithms across client devices that hold data samples without exchanging those samples. A centralized server might be used to orchestrate rounds of training for the algorithm and act as a reference clock, or the arrangement might be peer-to-peer. Regardless, local algorithms are trained on local data samples and the weights — the learnable parameters of the algorithms — are exchanged between the algorithms at some frequency to generate a global model.

TAO also incorporates Nvidia TensorRT, which dials a model’s mathematical coordinates to a balance of the smallest model size with the highest accuracy for the system it’ll run on. Nvidia claims that TensorRT-based apps perform up to 40 times faster than CPU-only platforms during inference.

Elements of TAO are already in use in warehouses, in retail, in hospitals, and on the factory floor, Nvidia claims. Users include companies like Accenture, BMW and Siemens Industrial.

“AI is the most powerful new technology of our time, but it’s been a force that’s hard to harness for many enterprises — until now. Many companies lack the specialized skills, access to large datasets or accelerated computing that deep learning requires. Others are realizing the benefits of AI and want to spread them quickly across more products and services,” Adel El Hallak, director of product management for NGC at Nvidia, wrote in a blog post. “TAO … can quickly tailor and deploy an application using multiple AI models.”

The benefits of AI and machine learning can feel intangible at times, but surveys show this hasn’t deterred enterprises from adopting the technology in droves. Business use of AI grew a whopping 270% from 2015 to 2019, according to Gartner, while Deloitte says 62% of respondents for its corporate October 2018 report deployed some form of AI, up from 53% a year ago. Bolstered by this growth, Grand View Research predicts that the global AI market size will reach $733.7 billion by 2027.