https://github.com/arkaung/cs525-project

Final Project of CS525 (Deep Neural Networks - Jacob Whitehill)
https://github.com/arkaung/cs525-project

Last synced: 3 months ago
JSON representation

Final Project of CS525 (Deep Neural Networks - Jacob Whitehill)

• Host: GitHub
• URL: https://github.com/arkaung/cs525-project
• Owner: ArkAung
• License: mit
• Created: 2017-06-04T15:55:51.000Z (about 9 years ago)
• Default Branch: master
• Last Pushed: 2017-06-04T16:01:49.000Z (about 9 years ago)
• Last Synced: 2025-12-26T22:49:44.707Z (7 months ago)
• Language: Jupyter Notebook
• Size: 209 MB
• Stars: 1
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

          
# Classifying marine mammal vocalizations with Convolutional Neural Networks

#### Authors - Arkar Min Aung, Francisco Guerrero and Michael Sokolovsky

##### Repository of notebooks for final project of CS525-191N

An exciting aspect of modern machine learning techniques is their ability to uncover patterns in data that are both difficult for humans to notice or were previously unknown. In this paper, we attempt to use modern machine learning techniques to discover if dolphins species can be categorized by their vocalizations: a task that to our knowledge has not been attempted by humans or computer algorithms. We explore convolutional neural networks architectures as models for processing dolphin sound clips and categorizing them into four different species.

This repository contains notebooks for:

* 2D Convolutional Network on "Best of Cuts" data for four species of dolphins

* 2D Convolutional Network on "All Cuts" data for four species of dolphins

* 1D Convolutional Network on walrus and dolphins

* 2D Convolutional Network on walrus and dolphins

* Helper files for generating files

* Visualization notebooks for intput data and filters learnt by the Convolutional Networks