A Deep Representation for Invariance And Music Classification
Representations in the auditory cortex might be based on mechanisms similar to the visual ventral stream; modules for building invariance to transformations and multiple layers for compositionality and selectivity. In this paper we propose the use of such computational modules for extracting invaria...
| Main Authors: | , , , , |
|---|---|
| Format: | Technical Report |
| Language: | en_US |
| Published: |
Center for Brains, Minds and Machines (CBMM), arXiv
2015
|
| Subjects: | |
| Online Access: | http://hdl.handle.net/1721.1/100163 |
| _version_ | 1826198698155048960 |
|---|---|
| author | Zhang, Chiyuan Evangelopoulos, Georgios Voinea, Stephen Rosasco, Lorenzo Poggio, Tomaso |
| author_facet | Zhang, Chiyuan Evangelopoulos, Georgios Voinea, Stephen Rosasco, Lorenzo Poggio, Tomaso |
| author_sort | Zhang, Chiyuan |
| collection | MIT |
| description | Representations in the auditory cortex might be based on mechanisms similar to the visual ventral stream; modules for building invariance to transformations and multiple layers for compositionality and selectivity. In this paper we propose the use of such computational modules for extracting invariant and discriminative audio representations. Building on a theory of invariance in hierarchical architectures, we propose a novel, mid-level representation for acoustical signals, using the empirical distributions of projections on a set of templates and their transformations. Under the assumption that, by construction, this dictionary of templates is composed from similar classes, and samples the orbit of variance-inducing signal transformations (such as shift and scale), the resulting signature is theoretically guaranteed to be unique, invariant to transformations and stable to deformations. Modules of projection and pooling can then constitute layers of deep networks, for learning composite representations. We present the main theoretical and computational aspects of a framework for unsupervised learning of invariant audio representations, empirically evaluated on music genre classification. |
| first_indexed | 2024-09-23T11:08:25Z |
| format | Technical Report |
| id | mit-1721.1/100163 |
| institution | Massachusetts Institute of Technology |
| language | en_US |
| last_indexed | 2024-09-23T11:08:25Z |
| publishDate | 2015 |
| publisher | Center for Brains, Minds and Machines (CBMM), arXiv |
| record_format | dspace |
| spelling | mit-1721.1/1001632019-04-10T11:36:22Z A Deep Representation for Invariance And Music Classification Zhang, Chiyuan Evangelopoulos, Georgios Voinea, Stephen Rosasco, Lorenzo Poggio, Tomaso Audio Representation Hierarchy Invariance Machine Learning Theories for Intelligence Representations in the auditory cortex might be based on mechanisms similar to the visual ventral stream; modules for building invariance to transformations and multiple layers for compositionality and selectivity. In this paper we propose the use of such computational modules for extracting invariant and discriminative audio representations. Building on a theory of invariance in hierarchical architectures, we propose a novel, mid-level representation for acoustical signals, using the empirical distributions of projections on a set of templates and their transformations. Under the assumption that, by construction, this dictionary of templates is composed from similar classes, and samples the orbit of variance-inducing signal transformations (such as shift and scale), the resulting signature is theoretically guaranteed to be unique, invariant to transformations and stable to deformations. Modules of projection and pooling can then constitute layers of deep networks, for learning composite representations. We present the main theoretical and computational aspects of a framework for unsupervised learning of invariant audio representations, empirically evaluated on music genre classification. This work was supported by the Center for Brains, Minds and Machines (CBMM), funded by NSF STC award CCF - 1231216. 2015-12-08T20:45:08Z 2015-12-08T20:45:08Z 2014-17-03 Technical Report Working Paper Other http://hdl.handle.net/1721.1/100163 arXiv:1404.0400v1 en_US CBMM Memo Series;002 Attribution-NonCommercial 3.0 United States http://creativecommons.org/licenses/by-nc/3.0/us/ application/pdf Center for Brains, Minds and Machines (CBMM), arXiv |
| spellingShingle | Audio Representation Hierarchy Invariance Machine Learning Theories for Intelligence Zhang, Chiyuan Evangelopoulos, Georgios Voinea, Stephen Rosasco, Lorenzo Poggio, Tomaso A Deep Representation for Invariance And Music Classification |
| title | A Deep Representation for Invariance And Music Classification |
| title_full | A Deep Representation for Invariance And Music Classification |
| title_fullStr | A Deep Representation for Invariance And Music Classification |
| title_full_unstemmed | A Deep Representation for Invariance And Music Classification |
| title_short | A Deep Representation for Invariance And Music Classification |
| title_sort | deep representation for invariance and music classification |
| topic | Audio Representation Hierarchy Invariance Machine Learning Theories for Intelligence |
| url | http://hdl.handle.net/1721.1/100163 |
| work_keys_str_mv | AT zhangchiyuan adeeprepresentationforinvarianceandmusicclassification AT evangelopoulosgeorgios adeeprepresentationforinvarianceandmusicclassification AT voineastephen adeeprepresentationforinvarianceandmusicclassification AT rosascolorenzo adeeprepresentationforinvarianceandmusicclassification AT poggiotomaso adeeprepresentationforinvarianceandmusicclassification AT zhangchiyuan deeprepresentationforinvarianceandmusicclassification AT evangelopoulosgeorgios deeprepresentationforinvarianceandmusicclassification AT voineastephen deeprepresentationforinvarianceandmusicclassification AT rosascolorenzo deeprepresentationforinvarianceandmusicclassification AT poggiotomaso deeprepresentationforinvarianceandmusicclassification |