Efficient Multi-Class Out-of-Distribution Reasoning for Perception Based Networks: Work-in-Progress
| Autor: | Abhishek Dubey, Arvind Easwaran, Shreyas Ramakrishna, Zahra Rahiminasab |
|---|---|
| Rok vydání: | 2020 |
| Předmět: |
Class (computer programming)
business.industry Computer science Cyber-physical system 02 engineering and technology 010501 environmental sciences Work in process Machine learning computer.software_genre 01 natural sciences Autoencoder Control theory 0202 electrical engineering electronic engineering information engineering 020201 artificial intelligence & image processing Artificial intelligence State (computer science) business computer Generative grammar 0105 earth and related environmental sciences Test data |
| Zdroj: | EMSOFT |
| DOI: | 10.1109/emsoft51651.2020.9244027 |
| Popis: | Perception-based deep neural networks used in Cyber Physical Systems are known to fail when faced with inputs that are out-of-distribution (ODD). ODD detection is a complex problem as we need to first identify the shift in the test data from the training distribution and then we need to isolate the responsible generative factor(s) (weather, lighting levels, traffic density, etc.), Unlike the state of the art that uses multi-chained one-class classifiers, we propose an efficient single monitor that uses the principle of disentanglement to train the latent space of a variational autoencoder to be sensitive to distribution shifts in different generative factors. We demonstrate our approach using an end-to-end driving controller in the CARLA simulator. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|