Competent Ultra Data Compression By Enhanced Features Excerption Using Deep Learning Techniques
| Autor: | V. Muneeswaran, A. Sudheer Kumar, P. Nagaraj, Surendra Rao. J, K. Muthamil Sudar |
|---|---|
| Rok vydání: | 2020 |
| Předmět: |
Computer science
business.industry Deep learning Computer Science::Neural and Evolutionary Computation Feature extraction ComputingMethodologies_IMAGEPROCESSINGANDCOMPUTERVISION 020206 networking & telecommunications Pattern recognition Data_CODINGANDINFORMATIONTHEORY 02 engineering and technology Autoencoder Convolutional neural network Redundancy (information theory) Computer Science::Computer Vision and Pattern Recognition Compression ratio 0202 electrical engineering electronic engineering information engineering 020201 artificial intelligence & image processing Artificial intelligence business MNIST database Data compression |
| Zdroj: | 2020 4th International Conference on Intelligent Computing and Control Systems (ICICCS). |
| Popis: | The objective of data compression is to extract the main features of the data and to restore the decompressed data from latent space i.e., compressed data without any quality or noise. In this paper, a Convolutional LSTM model is proposed to reduce the redundancy data and unnecessary information in the data set. The proposed methodology normalizes the picture to reduce blur and noises, with a compression ratio of 50%. The Convolutional LSTM model is compared with other models such as autoencoder, denoising autoencoder, convolutional neural network and our present work shows better RMSE compared to the other models. Datasets like MNIST and other datasets are used for testing and training the images. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|