Home / SmartTech / Truera applies AI explicability research to expand its model assessment platform

Truera applies AI explicability research to expand its model assessment platform

As AI affects more industries and areas of society, startups are developing test tools that companies and governments can use to assess the condition of their models and identify potential compliance issues. Andrew Cloud, chief AI scientist at Google Cloud, recently predicted that the AI ​​development pipeline will one day focus on testing and validation. Truera, formerly known as AILens, however says that it builds a different kind of platform based on the work of co-founders Anupam Datta and Shayak Sen, the explainability of which AI research takes into account causality and cooperative game theory, as well as fairness and bias.

Datta, who is on leave from his professorship at Carnegie Mellon University, told VentureBeat that the research behind Truera started in 201

4 when CMU researchers uncovered forms of gender bias in online advertising but had no tools to do so understand how to measure this distortion. The Truera platform can identify and mitigate problems that occur in AI models, e.g. B. Distortions, distortions in data distribution that can affect predictions, and stability over time. According to Datta, customers are more likely to use this feature due to anomalies caused by COVID-19.

Datta, who serves as President and Chief Scientist at Truera, said a central tenet of Truera’s approach is that models that are free from heavily weighted prejudice are not only more robust and powerful, but also better for business.

“The thesis was that explainability is the key to protecting yourself from social damage and creating higher quality models. It wasn’t like either or, ”Datta explained. “In a way, the same methods that you can use to learn more about models, how they work, how … we responsibly evaluate or adopt them are also the things that help make models of better quality come into production, and therefore also be good for the world and for business. These two requirements are not in conflict. You are well aligned. “

Today Truera focuses on classification and regression models as well as deep neural networks such as Convolutional Neural Networks (CNN). The company does not currently offer services for recommendation algorithms or certain other forms of AI.

Truera raised a $ 5.1 million round of financing in 2019 to build and expand its service. Greylock Partners led the round with the participation of Wing VC and Conversion Capital. The funds will be used to create the first version of the Truera product, expand the company team and launch the platform. First-time customers use the platform to check compliance with legal requirements in highly regulated industries such as banks and insurance companies.

Truera was founded in 2019 and is based in Redwood City, California. The company employs around 15 people.

Sign up for Weekly funding Start your week with VB’s top finance stories.

Source link