Technology 2017-03-14T19:37:54+00:00

Technology

Compellon’s patent-pending collaborative artificial intelligence technology empowers you to make decisions faster.

 

Technology

Compellon is the only platform capable of creating custom self-aware, self-adapting predictive algorithms powered by collaborative AI. No fixed models here.

Compellon20|20 creates custom self-aware, self-adapting predictive algorithms. No fixed models here.

What makes Compellon’s breakthrough user experience possible is the industry-unique artificial intelligence technology under the hood. Compellon’s approach to AI is to extend (rather than to imitate) human expertise. The engine acts as an autonomous intelligent agent that performs the analytics based only on the user-provided goal and the evidence in the data. This empowers users to focus solely on the business challenges and the actions to achieve their objective, without extensive skills and assumptions, which are typically required to perform advanced analytics.

Compellon’s approach incorporates and builds upon elements of mainstream machine learning practice, while differing in several substantial ways:

Custom Structured

Compellon autonomously derives a unique model structure for each individual data set. This is in stark contrast to mainstream statistical and machine learning approaches that always start by assuming a type of structure (such as a decision tree, a logistic regression, or SVM) or use trial and error to choose from multiple structures.

Self-Aware & Adaptive

The Compellon engine stores the image of its own structure and has the ability to change it. By using this characteristically-AI ability, it modifies its own behavior to avoid past errors and to account for changes in environment. This is in contrast to the mainstream machine learning approaches that require human intervention to adapt models, often by using techniques that add complexity (such as “bagging” and “boosting”) to correct errors rather than recognize and remove their reasons.

Information Based

The Compellon engine consistently uses concepts of information theory (e.g., Shannon entropy and many other measures based on it) for every phase of analysis, including variable selection, discovery of relations, determining the model structure, tuning the model, forming prescriptive output, and providing adaptive capabilities when conditions are changing. Due to this approach, the engine avoids many analytical guesses and arbitrary decisions that would otherwise be necessary.

These capabilities today power the breakthrough user experience that transforms knowledge discovery and predictive modeling with the unique combination of speed, ease, and powerful insight into why and what to do. They also position Compellon with the technology to uniquely address the growing need for autonomous operation to deal with the explosion of data across a variety of use cases.

What makes Compellon’s breakthrough user experience possible is the industry-unique artificial intelligence technology under the hood. Compellon’s approach to AI is to extend (rather than to imitate) human expertise. The engine acts as an autonomous intelligent agent that performs the analytics based only on the user-provided goal and the evidence in the data. This empowers users to focus solely on the business challenges and the actions to achieve their objective, without extensive skills and assumptions, which are typically required to perform advanced analytics.

Compellon’s approach incorporates and builds upon elements of mainstream machine learning practice, while differing in several substantial ways:

Custom Structured

Compellon autonomously derives a unique model structure for each individual data set. This is in stark contrast to mainstream statistical and machine learning approaches that always start by assuming a type of structure (such as a decision tree, a logistic regression, or SVM) or use trial and error to choose from multiple structures.

Self-Aware & Adaptive

The Compellon engine stores the image of its own structure and has the ability to change it. By using this characteristically-AI ability, it modifies its own behavior to avoid past errors and to account for changes in environment. This is in contrast to the mainstream machine learning approaches that require human intervention to adapt models, often by using techniques that add complexity (such as “bagging” and “boosting”) to correct errors rather than recognize and remove their reasons.

Information Based

The Compellon engine consistently uses concepts of information theory (e.g., Shannon entropy and many other measures based on it) for every phase of analysis, including variable selection, discovery of relations, determining the model structure, tuning the model, forming prescriptive output, and providing adaptive capabilities when conditions are changing. Due to this approach, the engine avoids many analytical guesses and arbitrary decisions that would otherwise be necessary.

These capabilities today power the breakthrough user experience that transforms knowledge discovery and predictive modeling with the unique combination of speed, ease, and powerful insight into why and what to do. They also position Compellon with the technology to uniquely address the growing need for autonomous operation to deal with the explosion of data across a variety of use cases.

Want more information on our technology?

Technology Backgrounder Whitepaper

DOWNLOAD

Want to learn more?

Technology Backgrounder Whitepaper

DOWNLOAD

Read our Chief Data Scientist’s technology vision

Read Now

Read our technology vision

READ NOW
X