Deep learning can turn basic unstructured inputs into meaningful concepts. While Tech giants use deep learning technology for consumer application, we focus on specific high-value tasks. We research and develop proprietary deep learning algorithms, with a focus on image recognition and text understanding

Tractable Image Recognition

The past 3 years have seen fundamental breakthroughs in computer vision via Deep Learning. Deep learning systems now surpass human accuracy on image recognition tasks. Tractable's powerful image and video recognition technology is built on the most advanced deep learning systems and is easily accessible by API.

Tractable Text Understanding

Tractable employs deep learning technology to parse unstructured document (OCR) and understand natural language (NLP). We apply our proprietary active learning platform to improve how AI analyses text content and understand context. Tractable's powerful text understanding technology is built on the most advanced deep learning systems and is easily accessible by API.

Tractable Proprietary Active Learning Platform is the key enabler to unlock disruptive applications of Deep Learning

Deep learning is leading the revolution in Artificial Intelligence

Four years ago, image analysis was limited to simple pattern recognition cases. Today, our deep learning technology rivals human expert performance. This revolution is spreading to speech, natural language and microbiology.

But the requirement of high volumes of tagged data is holding back disruptive applications

We think the recognition and prediction tasks that matter are those which require human expertise. But gathering expert tags is slow and costly.

We believe active learning is the solution. We are developing proprietary technology to cut data volume requirements 1000X.

Our mission: starting from unlabelled data, with minimal manual intervention, reach comparable performance to supervised learning on labelled data.

We have successfully applied our technology to auto claim images (patents pending). Other use cases are on their way.