An algorithm for problem solving in cars and the human brain

Imagine having a car accident in a parking lot, your car has some small damage and you have to get it fixed. What if your car could tell you which parts are broken and how much it will cost to fix it? During his Ph.D. at BMW, Milan Koch designed the Automated Damage Assessment service, a customer service that does just this. “It should be a nice experience for customers, even in such an awful situation.”

Time series problems

“From scratch, we have developed a service idea that is about detecting damaged parts from low speed accidents. The car itself is able to detect the parts that are broken and can estimate the costs and the time of the repair.” Koch explains. He uses data from sensors that gather data over time from different parts of the car. Therefore, this problem can be classified as a multivariate time series problem.

Koch developed and compared different multivariate time series methods, based on Machine Learning, Deep Learning and also state-of-the-art AutoML methods (automated machine learning) with different levels of complexity to find the best way to solve multivariate time series problems. Two of the AutoML methods and his hand-crafted machine learning pipeline gave the best results for the multivariate time series problems.

Domain shift

The machine learning pipelines he created are applicable not only in the automotive field, but can also be applied to other multivariate time series problems. Koch collaborated with researchers from the Leiden University Medical Center (LUMC) to use his hand-crafted pipeline to analyze Electroencephalography (EEG) data. Koch: “We predicted the cognition of patients based on EEG data, because an accurate assessment of cognitive function is required during the screening process for Deep Brain Stimulation (DBS) surgery. Patients with advanced cognitive deterioration are considered suboptimal candidates for DBS as cognitive function may deteriorate after surgery. However, cognitive function is sometimes difficult to assess accurately, and analysis of EEG patterns may provide additional biomarkers. Our machine learning pipeline was well suited to apply to this problem. We developed algorithms for the automotive domain and initially we didn’t have the intention to apply it to the medical domain, but it worked out really well.”

His pipelines are now also used in Electromyography (EMG) data, to distinguish between people with a motor disease and healthy people.

Koch will continue his work at BMW Group, where he will focus on customer-oriented services, predictive maintenance applications and optimisation of vehicle diagnostics.