AI Model Tuning

Outcomes are key

Cyber Security ML detections require tuning to deliver outcomes that are crisp, clear, and actionable. Tuning takes a period of time for the tool to learn, then a tuning phase is required to perfect the results. In many cases a COTS ML solution may require tuning, sometimes several rounds, to make the outcomes useful for a given site, architecture, or unique data set.

Tuning is key

CyberQ has a team of experts to focus on helping customers achieve positive results, and tuning is one key aspect of AI in Security.


Every customer architecture, log style, naming conventions, and application framework is different. There is no one ML model that works every time. Expert tuning will help deliver actionable outcomes for many COTS based ML tools, along with custom models that may not work properly.

Remove False Positives

The concept of using ML in the SOC is to reduce or remove false positives, but that can only happen with a properly tuned model that is tested with live customer data onsite.

Tuning methodology

Starting with an outcomes based methodology, AI models are tuned to deliver specific alerts and outcomes, while reducing false positives to the required threshold for each individual customer’s requirements, These requirements are defined at the early stage of the engagement to ensure the SOC is delivered exactly the level of detail required.

Training the Customer

Many times the concepts and implementation of ML is a new subject within the SOC team, or the insider threat teams. CyberQ works with each individual customer to help guide the teams with a customized learning path using the models and live data to help customers become self-sufficient over time.


CyberQ AI Custom Tuning is delivered using a methodology to design usable outcomes, compliance driven, with training and a workflow that can be repeated over many models and data sets.  

AI and Analytics test

Validation is a key step in the methodology that proves to the business that a given custom model is complete, and has the required outcome to make decisions. 

Engaging CyberQ

CyberQ AI Models program can be purchased as a complete program to design, implement, and rollout into production a mature customized ML solution using COTS SW that can be integrated into the main SOC workflow.  The program is focused on achieving results from the customers platform.

Quality Data Creates

Quality Results

The backbone of quality ML models is quality Data. CyberQ will work with the customer to audit and modify data ingestion to ensure quality outcomes from any Custom ML model.