Deep learning startup Tensorleap offers explainability platform for neural networks
Backed by Angular Ventures, Sozo Ventures and Industry Ventures, the company raised a seed round of 5.2 million dollars
Deep learning startup Tensorleap has come out of stealth mode with its novel debugging and explainability platform for neural networks.
Tensorleap is led by co-founder and CEO, David Ben David, co-founder and CTO, Yotam Azriel, and co-founder and CCO, Nir Ben David.
The three have extensive experience in the software industry as well as startups in the fields of cloud computing, artificial intelligence, computer vision and big data. Backed by Angular Ventures, Sozo Ventures and Industry Ventures, the company raised a seed round of 5.2 million dollars.
The startup aims to bring transparency to deep learning models during the development phase, drastically lowering the risk of models failing in production. Despite the enormous growth and promise of Deep Neural Networks (DNN), their black box nature results in numerous challenges that limit their potential, such as long and inefficient development cycles, lack of explainability, and insufficient testing - all of which contribute to AI failures in production.
When it comes to life-affecting decisions, such as medical scan diagnostics or autonomous driving, the ability to verify and explain success and failure are critical to mission success. By providing full visibility in the development stage, Tensorleap enables organizations to unleash the full potential of deep learning.
“Tensorleap is creating a new category in neural network development. This technology will redefine how deep learning models are developed and will enable wider expansion of one of the fastest growing AI technology sectors” noted Gil Dibner, General Partner and Founder, Angular Ventures.
Current neural network development solutions do not have the capability to show data scientists how models perceive the data, and thus require numerous experiments based on trial and error to improve models and outcomes. They also lack the tools to carry out comprehensive testing before production, reducing trust in the data model due to frequent failures.
“I experienced first-hand the current limitations of neural network development as CTO in the autonomous driving space,” said David Ben David, Tensorleap co-founder and CEO. “We had zero visibility when it came to pinpointing where and why models failed and how to improve them. My partners and I founded Tensorleap to remove the blindfold.”
Tensorleap is a SaaS subscription-based platform that can be installed on the client's infrastructure or in the cloud and supports all types of data, including images, text, graphs, tabular information and more. The company is based in Ramat Gan and offers services for various fields fields, from medical products to chip manufacturers, gaming and social media platforms.