Towards Self-Improving Activity Recognition Systems Based on Probabilistic, Generative Models.
Martin JänickeSven TomfordeBernhard SickPublished in: ICAC (2016)
Keyphrases
- generative model
- activity recognition
- probabilistic model
- human activity recognition
- human activities
- mixture model
- posterior probability
- prior knowledge
- bayesian framework
- maximum entropy principle
- discriminative learning
- em algorithm
- topic models
- discriminative models
- object categories
- smart home
- computer vision
- action recognition
- semi supervised
- data mining
- conditional random fields
- latent variables
- supervised learning
- knn
- probability distribution
- sensory data
- learning environment
- bayesian networks
- accelerometer data