By Bryan Lares & Keith Estes
What puts SparkCognition’s technology at the leading edge of machine learning?
SparkCognition has seen a meteoric rise in a crowded field of AI providers vying to persuade customers who are skeptical of its value. In the last year, we have tripled in size as a company, and recently received the largest investment the Austin market has seen this year. While there are many factors that differentiate us as a company, there are four key components behind the accomplishments of our AI: the core of our technology, our focused product suite, the explainability of our machine learning, and the success of our customers.
Our core technology
Every machine learning (ML) company will proclaim that their algorithms for model building are the best. However, the reality is that the touted algorithm is unlikely to be a differentiator, or even unique—algorithms are difficult to patent, and thus can be accessed by others in the ML space. On top of this, no amount of explaining, arguing, or evangelizing will prove that one algorithm is more effective than another, as different tests can be set up to make anything look like the best solution. What sets companies apart is how they use software engineering and data science to package their algorithms.
SparkCognition leverages patented automated model-building algorithms to develop a customized solution for each dataset—a step that must be done by humans with other AI providers. Our automated model-building solutions process information from the data, draw conclusions, and then codify instincts and experience into learning.
Based on the validated results in industries such as wind, oil and gas, and cyber security, we have been able to expand and apply our machine learning solutions to a wide variety of industries, including maritime, aerospace, and defense. This vast bank of experience grants us the aptitude to tackle diverse challenges.
Our product suite
Another aspect that defines SparkCognition is our focused approach to the problems we’re solving. We have cultivated specific verticals where our products are dominant and successful. Our products in cognitive anti-malware, prescriptive maintenance, and natural language processing are specialized to each deployment, yet can be applied in many different industries.
An important element in developing machine learning solutions is building a product for subject matter experts (SMEs) and tailoring it to their needs. Augmenting SMEs through a specific user experience to enable those customers to work with the product autonomously, as well as collaboratively, is a critical driver in the development of our UX. We provide turnkey solutions that will provide sorted, prioritized, workable data to customers on day one, to allow SMEs to use their expertise instead of getting bogged down in data science.
Explainability of our AI
The final technical aspect that propels SparkCognition to the leading edge of machine learning is the explainability we provide for our analytics. While some competitors offer predictive analytics solutions, they fall short when explaining how their model got the specified prediction and what the next step should be.
SparkCognition’s solution to predictive analytics is more than just the automated model; it is a solution that provides a predictive analysis of the problem, how it reached that prediction, and the next step toward remediation of the impending issue. Our technology is able to back up our predictions and simplify the reasoning for the user, offering in-context advisory within our solutions.
While our patented technologies and methodologies set us apart in the market, it is the customer success we’ve enabled which truly places us at the leading edge of machine learning.
Invenergy, North America’s largest independent wind power generation company, applied SparkCognition’s machine learning solution to gearbox monitoring. Our solution was able to extend predictive capabilities for gearbox failure and degradation with a 30- to 60-day notice, while also providing global fleet visibility and a component risk index to allow for actionable maintenance insight. The analysis resulted in zero false positives, and no failures were missed. The success of the application resulted in Invenergy’s commitment to enterprise-wide rollout. As Invenergy’s VP of Asset Management, John Majewski, stated in the Chicago Tribune “They’re using our data to improve their algorithms, and in turn, we benefit from the improvement of their algorithm.”
For Flowserve, one of the largest suppliers of industrial pumps and related equipment, working with SparkCognition has expanded both the depth and breadth of diagnostics the company can perform on running machinery. It has also improved the predictive power of those analytics, moving the needle on forewarning from two hours before failure to five days in advance.
These successes have continued across the numerous applications of SparkCognition’s machine learning solutions. We currently have a variety of deployments, many of which are Fortune 500 companies. Our customers are leaders in their industries and with their continued partnership, we are delivering the industry’s leading AI solutions to solve their biggest challenges.