Launch of ML Ex-Lab Open Source

After two months of intensive research, D-Tech LLC, a technology firm specializing in applied machine learning (ML) and cybersecurity in Northern VA, released the source code of ML Ex-Lab, an ML model selection and optimization tool as open source on GitHub. This tool addresses the problems of disorganization and lack of consistency in typical ML model training processes, and drastically reduces the time required for ML model training and optimization. The ML Ex-Lab tool provides ML designers and data scientists with a streamlined method for fine-tuning ML training parameters, recording experimental data, and comparing the performance of different models. The source code, along with detailed usage instructions can be found at https://github.com/dtechspace/ml-exlab

The model development in ML is generally a tedious manual process that often results in inconsistency, inefficiency, and inaccuracy of ML models.  ML Ex-Lab is designed as a solution to help data scientists and AI/ML engineers with model training, analysis and optimization. ML Ex-Lab automatically logs experiment results based on degrees of freedom configured by the ML designer.  It provides a common reference framework that enables accurate comparison of data points. This tool is easily integrated with the hyper optimization of higher-level variables including model type and nature of data processing. An additional feature of the tool is its ability to specify ML experiments from an external customizer (e.g. network requests and graphical interfaces).  

With the publication of ML Ex-Lab, we hope to contribute to the ML community in targeting the common ML problems, and to invite ML researchers and enthusiasts to collaborate on the project.  Our goal is to make ML Ex-Labs a transformative and automated tool to support ML model construction and optimization for different ML platforms at scale. 

Other open source projects by the Company can be found on D-Tech’s GitHub page at github.com/dtechspace. Additional information about D-Tech’s products and solutions can be found at www.dtechspace.com.