Meta-learning for Automotive Sensor Systems
Master Thesis
Motivation and Goals
Multi-task learning (MTL) is an interesting subfield of Meta-Learning in which it refers to building a single model that learns across multiple related predictive modeling tasks. Theoretically, MTL models should perform as well, if not better, as models trained on each task individually as they benefit from Implicit data augmentation, eavesdropping and better regularization. One of key advantages of MTL is that it is easier to maintain, host and deploy a single model capable to generalize for many tasks over maintaining a model for each task.
With that in mind, the objective of this work is to explore MTL techniques and strategies for deterministic systems. Automotive sensor system will be used as a case study. In Automotive Sector, machine learning models are developed to reconstruct a sensor that is used to enhance user experience by studying and observing the standard sensors in the vehicle. For a single vehicle, many models are developed to monitor a specific component, like tires, tie-rod, suspension, etc. These models mostly share the same feature space but are trained for specific tasks. The objective is training a single model capable of reconstructing multiple sensors in a vehicle. The data provided for this work is collected from CAN bus of Tesla Model 3.
Requirements
- Currently studying master’s in informatics or equivalent studies. ·
- Have knowledge in Deep learning models.
- Have experience in python programming language.
- Familiar with at least one Deep learning framework, such as, TensorFlow, Keras or Pytorch.
- Your work is independent, structured, and responsible.
