Feature-Rich Audio Model Inversion for Data-Free Knowledge Distillation Towards General Sound Classification.
Zuheng KangYayun HeJianzong WangJunqing PengXiaoyang QuJing XiaoPublished in: ICASSP (2023)
Keyphrases
- input data
- expert knowledge
- prior knowledge
- knowledge discovery
- raw data
- test data
- data mining techniques
- experimental data
- data sets
- network structure
- probabilistic model
- statistical methods
- semantically rich
- conceptual model
- background knowledge
- missing information
- domain experts
- decision trees
- feature vectors
- data analysis
- training data
- high level
- confidence scores
- database
- data mining
- feature selection
- feature values
- empirical data
- domain knowledge
- classification algorithm
- classification accuracy
- classification method
- image reconstruction
- data points