SMILE4KMU

Motivation

Medium-sized companies are also increasingly using machine learning (ML) methods, which form a sub-area of artificial intelligence. Computers are trained with data and experience instead of being explicitly programmed for specific tasks. Such ML methods complement both traditional software development and the specific value chain of companies. It is particularly important that these ML processes are secure and cannot be manipulated. This must already be taken into account during the development of software and throughout its entire life cycle. Research has made considerable progress in this area in recent years under the heading of “secure software lifecycle”. However, there are still only a few comprehensive, structured and easy-to-use approaches for ML procedures and their secure integration into development processes. Such approaches are of great importance, especially for small and medium-sized enterprises (SMEs). While many SMEs want to exploit the economic potential of ML processes, they often lack the know-how and skills to use them safely due to limited resources.

Goals & Approach

The project “Secure Machine Learning Lifecycle” (SMILE4KMU) has set itself the goal of taking a holistic view of the security of ML methods in development processes, from the protection of training data to the distribution of the software. The researchers want to develop processes and procedures that also allow SMEs to use machine learning securely in their development projects. Accordingly, the project team attaches particular importance to securing theML processes used in a structured and comprehensible manner. The project “Secure Machine Learning Life-cycle” (SMILE4KMU) has therefore set itself the goal of taking a holistic view of the security of ML procedures in development processes, from the protection of training data to the distribution of the software. The researchers want to develop processes and procedures that also allow SMEs to use machine learning securely in their development projects. Accordingly, the project team attaches particular importance to securing theML processes used in a structured and comprehensible manner. In addition, the researchers ensure that the software created with ML is easy to protect and license, and that all this can be implemented at a cost that is economically feasible for SMEs.

#MachineLearningSecurity #SecureMLIntegration #AI #MLDevelopment #AIResearch #MLBestPractices #DataProtection #ITSecurity