Exploring AI Mesh: How Forcepoint AI Classification Works

AI classification is a buzzword wrapped inside a black box. 

In other words: what exactly is AI classification – and how does it work?

AI classification is included in Forcepoint Data Security Posture Management (DSPM) and the technology that underpins it is AI Mesh. AI Mesh is a highly networked architecture of AI classifiers and data science models that provides highly precise data classification with ultimate efficiency.

How Forcepoint AI Classification Works

Forcepoint AI classification uses a mix of AI classifiers and data science models to automate the identification and cataloging of unstructured data across cloud and on-premises storage. It enables organizations to automate the discovery and protection of assets that are normally difficult to do so, such as intellectual property.

AI Mesh helps the Forcepoint AI classification engine understand relationships between words within a file to correctly classify its criticality. AI Mesh uses a mix of nodes, such as deep neural network classifiers—machine learning—to more accurately spot and define these relationships. These AI classifiers and predictive data modelling techniques work together to create highly efficient and accurate data classifications when trained properly.

AI classification is a non-negotiable tactic for organizations that work with sensitive or proprietary information. It ensures that the Data Loss Prevention (DLP) policies built to enable organizations to comply with regional and industry-specific regulations are enforceable – because without accurate regulatory classification, data security policies don’t know what to protect.

Here’s an example of what happens when unstructured data is fed into Forcepoint AI Mesh:

Example of AI Classifiers Used in AI Mesh

At the heart of Forcepoint AI Mesh sits a generative AI Small Language Model (SLM). The GenAI SLM generates vector document representations of the actual document itself, which the rest of the nodes – AI classifiers, data science models and other capabilities – analyze to determine the file’s classification.  

SLMs carry many benefits over their larger counterpart, the Large Language Model (LLM), despite being less well known. SLMs run on CPUs and require magnitudes less computing power than LLMs, but with the smaller size comes improved resource and cost efficiencies, as well as better performance. Forcepoint DSPM can SLMs can discover roughly 200 milliseconds per file – and the AI Mesh can classify at roughly one-fifth of that speed. 

The AI Mesh architecture is designed to make AI classification attainable for organizations of all sizes. The SLM is joined by other AI classifiers and data science capabilities within the networked architecture, including:

• Deep neural network classifiers for sentiment analysis.
• Light AI classifiers such as Bag of Words for determining topics.
• Bayesian Inference for predictive modeling of text.
• Regex filters to delineate what text is classifiable.

Benefits of AI Mesh for Data Classification

While AI classification is a more common practice in data security these days, the algorithm behind it varies from vendor to vendor. AI Mesh is a unique approach to AI classification that drives highly precise results in a fraction of the time.

Benefits of AI Mesh include:

• Reduce Risk: High-precision data classification mitigates risk of a data breach, leak, or non-compliance.
• Cut Cost: Reduce costs for extensive computing resources associated with LLMs, maintenance and training.
• Increase Productivity: Save time on classification without compromising on accuracy through better efficiency.
• Streamline Compliance: Easily prove compliance with auditable and explainable AI technology.

Interested in learning more about AI Mesh? Talk to an expert today to book a demo for DSPM.