Beyond Limits: Going Beyond Conventional AI

The Achilles heel of conventional AI techniques like machine learning, deep learning, or neural networks is that they are “black boxes” that cannot explain how they came up with an answer.

Ironically, the beauty of a neural net–its greatest advantage–is also its greatest curse. You train it on data, but it is not required for a neural net to know what it's being trained on. To it, the data is just nothing but a bunch of numbers. For example, by training it with thousands of photos of chairs, a machine learning system can learn to recognize a chair. But it will never know what a chair is for, or why a human might need more than one type of chair or what makes for a good or bad chair. In regulated, compliance-oriented industries like healthcare or finance, or in high-risk, high-reward industries like logistics or oil and gas production, this is simply unacceptable.

In contrast to conventional AI approaches, Beyond Limits AI solutions are always “explainable”. Our cognitive engines deliver clear audit trails explaining the reasoning behind their recommendations, and showing the evidence, risk, confidence, and uncertainty.

How Beyond Limits Helped a Customer

We have conducted several projects in the energy sector, specifically in oil and gas exploration and extraction. For example, we built a world-first solution that involved teams of geologists, petrophysicists, seismologists, and reservoir engineers to automate data collection for reservoir management. The system dramatically accelerates the analysis of massive amounts of previously siloed data to form a true picture of subsurface assets. The core of the solution is a truly cognitive reasoning engine that can be scaled to solve problems for organizations in many industries.

Identifying Specific Challenges and Building Solutions to address them

We refer to ourselves as a cognitive intelligence company because we produce systems that are both trained and educated. The training is done through machine learning and deep learning tools that have access to a lot of data. Sometimes, the data may be limited or unreliable, which is why it is imperative for our AI systems to be educated by mechanisms such as knowledge derived from human experts.

We are working with numerous strategic partners globally, for whom data is the most important asset. Often data extracted from industrial sensors or applications can be missing or incomplete. However, on space missions you have to be ready to deal with the unknown and keep the mission on track. Our systems have the capability to handle missing or misleading data and still reason out an intelligent result.

The Future of the Company

Since 2014, we have had two rounds of funding, and we are planning to close our third round in the 2018. As we move into 2019, Beyond Limits is planning to be a player across high-value sectors of the AI industry. As we apply our capabilities to the energy, FinTech, healthcare, and logistics industries, we are also adding more application-specific IP to our portfolio. One key development is AI on a Chip–which enables intelligence to be distributed to edge devices in clinical and industrial environments.

We also plan to broaden our global footprint to Europe, Asia, and additional offices in the U.S. It is an exciting time to be a part of Beyond Limits.