AvaWatz Patents FALCON, A New Trust Assessment and Auto-Correction for AI Models and Machine Learning
AvaWatz Company ("AvaWatz"), a technology company that supports collaboration among robots, is pleased to announce the filing of another patent application with the United States Patent and Trademark Office (USPTO). The patent application, titled "Systems and Methods for Trust-Aware Error Detection, Correction, and Explainability in Machine Learning and Computer Vision," introduces the AvaWatz approach to make Machine learning and deep learning models more trustworthy and reliable.
The more widely applicable machine learning (ML) and artificial intelligence (AI) get, the more we need to ensure trustworthiness and reliability in the AI/ML model predictions. Wrong predictions by a model could be dangerous and even fatal in applications like autonomous driving and medical imaging. AvaWatz's solution called FALCON focuses on (i) detecting when a model makes a mistake, (ii) deciding what kind of mistake is made, and (iii) fixing the model mistake. Avawatz plans to support several use cases, including object tracking, object detection & classification, and natural language processing.
Earlier this year, AvaWatz applied to patent GENIE and introduced the AvaWatz approach to the difficult issue of training data required to train deep learning and machine learning models. The AvaWatz approach saves time, reduces the cost of training such systems, and makes for a more accurate model that a user can trust.
"FALCON is a flagship product for trustworthy AI performance that empowers many use cases immediately. It works seamlessly with our GENIE product to create high-performing AI models. We foresee these product lines supporting many critical applications and missions across different market sectors." said Dr. Rajini Anachi, CEO of AvaWatz. "We make FALCON and GENIE available as part of our Trusted AI services to our customers and others in the Artificial Intelligence community. We feel that adopting these novel technologies will help many who work with specialized use cases and need to build customized machine learning applications."