Binarly Secures Patent for Machine Learning Technique to Optimize Large-Scale Binary Analysis

SANTA MONICA, Calif.--(BUSINESS WIRE)--Binarly, a leading innovator in software supply chain security, today announced that it has been granted U.S. Patent No. US 12,236,262 B1 for its groundbreaking "Machine Learning Technique for Efficiently Scheduling Tasks for Large-Scale Analysis of Binary Executables."

The patent, issued on February 25, 2025, covers a novel method for analyzing binary software efficiently by leveraging machine learning to predict peak memory usage and dynamically allocate computing resources.

The patented technology extracts features from software packages (like binary executables) using advanced preprocessing techniques and machine learning models trained on extensive datasets generated from monitoring memory usage during analysis.

By estimating the peak memory usage required for analysis, the system can determine and allocate the optimal number of memory resource units, or queue the task in a backoff queue when resources are insufficient. This approach enhances the efficiency of large-scale analysis by ensuring that each software package is processed using precisely the amount of computing resources needed, thereby reducing processing delays and lowering operational costs.

The Binarly invention, credited to researchers Alex Matrosov, Sam Thomas, Yegor Vasilenko and Lukas Seidel, is particularly significant in the context of security analysis and vulnerability assessment, where rapid and precise evaluation of binary executables is critical for identifying potential security risks within software supply chains.

“Our patented technology represents a significant advancement in the field of software security analysis,” said Matrosov, who is CEO and Head of Research at Binarly. “By intelligently predicting and managing resource allocation for binary analysis, we can dramatically increase the throughput and accuracy of vulnerability detection, ultimately contributing to a more secure digital ecosystem.”

“We’re proud to say that the techniques described in this patent has been implemented in our flagship Binarly Transparency Platform since last year,” Matrosov added.

The invention is expected to have broad applications in enterprise environments where large volumes of software need to be analyzed quickly and efficiently. Binarly anticipates that this technology will not only enhance the performance of existing analysis pipelines but also enable new approaches to automated vulnerability management and risk assessment.

For additional information about the patent and the Binarly Transparency Platform solutions, please visit www.binarly.io or contact our media relations team.

About Binarly:

Binarly is a global firmware and software supply chain security company founded in 2021. The company’s flagship Binarly Transparency Platform is an enterprise-class, AI-powered solution used by device manufacturers, OEMs, IBVs and product security teams to identify known and unknown vulnerabilities, misconfigurations and signs of malicious code implantation. Binarly’s validated remediation playbooks have significantly reduced the cost and time to respond to security exposures. Based in Los Angeles, California, Binarly brings decades of research and program analysis expertise to build solutions to protect businesses, critical infrastructure, and consumers around the world.