Deep CNN (convolutional neural networks)

• 61 Speech recognition using CNN

Spec: Integrate signal processing (spectrogram) and convolutional neural network.

Language: Python (Keras-Tensorflow), Matlab (Deep Learning & Neural Networks Toolbox, Signal Processing Toolbox).

• 128 Speech denoising using fully-connected and convolutional NN

Spec: Integrate signal processing (spectrogram) and feedforward NN

Language: Python (Keras-Tensorflow), Matlab (Deep Learning & Neural Networks Toolbox, Audio System Toolbox).

• 62 Recognizing neurological disease on MRI image using CNN

Spec: Build, train & validate deep network to classify images.

Language: Python (Keras-Tensorflow), Matlab (Deep Learning & Neural Networks Toolbox, Computer Vision Toolbox).

• 136 Recognizing object in image using classification CNN

Spec: Build, train & validate deep network to classify images.

Language: Python (Keras-Tensorflow), Matlab (Deep Learning & Neural Networks Toolbox, Computer Vision Toolbox).

• 63 Evaluating rotation of object in image using regression CNN

Spec: Build, train & validate deep regression network to evaluate rotation angles of objects in images.

Language: Python (Keras-Tensorflow), Matlab (Deep Learning & Neural Networks Toolbox, Computer Vision Toolbox).

• 64 Recognizing neurological disease on MRI image using CNN trained in cloud

Spec: Build, train & validate deep network to classify images. It is done in cloud.

Language: Python (Keras-Tensorflow), Matlab (Deep Learning & Neural Networks Toolbox, Computer Vision Toolbox).

• 89 Image denosing using pretrained CNN

Spec: Remove normal noise from color image using deep learning.

Language: Matlab (Deep Learning & Neural Networks Toolbox, Image Processing Toolbox).

• 130 Approximating image processing filter using CNN with dilation

Spec: Use deep learning to approximate image processing filter that reduces noise while preserving edge sharpness.

Language: Matlab (Deep Learning & Neural Networks Toolbox, Image Processing Toolbox).

• 96 Estimating high-resolution image from single low-resolution image using VDSR (very-deep super-resolution) NN (neural network)

Spec: Improve quality of low-resolution image using deep learning.

Language: Matlab (Deep Learning & Neural Networks Toolbox, Image Processing Toolbox).

• 97 Reducing JPEG compression artifacts in image using CNN

Spec: Improve quality of JPEG decompression using deep learning.

Language: Matlab (Deep Learning & Neural Networks Toolbox, Image Processing Toolbox).

• 74 Evaluating area of vegetation on aerial image using ISS (image semantic segmentation)

Spec: Segment multispectral image using deep convolutional neural U-Net.

Language: Matlab (Deep Learning & Neural Networks Toolbox, Computer Vision Toolbox).

• 134 Image semantic segmentation using CNN with dilation

Spec: Segment image using deep learning.

Language: Matlab (Deep Learning & Neural Networks Toolbox, Computer Vision Toolbox).

• 10 Fingerprint identification using pretrained & retrained CNN

Spec: TL (transfer learning) from a network pretrained on images of one set of classes to a network which is supposed to recognize images of another set. The TL is done by retraining.

Language: Python (Keras-Tensorflow), Matlab (Deep Learning & Neural Networks Toolbox, Computer Vision Toolbox).

• 48 Birds detection from airplane using CNN trained from scratch

Spec: Build, train & validate network to classify images.

Language: Python (Keras-Tensorflow), Matlab (Deep Learning & Neural Networks Toolbox, Computer Vision Toolbox).

• 75 Stop road sign detection using R-CNN

Spec: Integrate computer vision (Viola-Jones or ACF) and neural network (CNN).

Language: Matlab (Computer Vision Toolbox).

• 76 Detecting cars on highway using Faster R-CNN

Spec: Build & train CNN to detect predefined object in images.

Language: Matlab (Computer Vision Toolbox).

Deep LSTM (long short-term memory) RNN (recurrent NN)

• 65 Detecting arrhythmia from ECG signal using LSTM

Spec: Integrate signal processing (instantaneous frequency & spectral entropy) and recurrent neural network.

Language: Python (Keras-Tensorflow), Matlab (Deep Learning & Neural Networks Toolbox, Signal Processing Toolbox).

• 129 Detecting P/QRS/T waves from ECG signal using LSTM

Spec: Integrate signal processing (spectrogram, filtering) and recurrent neural network.

Language: Python (Keras-Tensorflow), Matlab (Deep Learning & Neural Networks Toolbox, Signal Processing Toolbox).

• 66 Speech recognition using LSTM

Spec: Integrate signal processing (spectrogram) and recurrent neural network.

Language: Python (Keras-Tensorflow), Matlab (Deep Learning & Neural Networks Toolbox, Signal Processing Toolbox).

• 127 Speaker gender detection using LSTM

Spec: Integrate signal processing (extract voice features) and recurrent neural network.

Language: Python (Keras-Tensorflow), Matlab (Deep Learning & Neural Networks Toolbox, Audio System Toolbox).

• 67 Forecasting incidence of infection using LSTM

Spec: Build, train & validate deep recurrent network to predict continuous values.

Language: Python (Keras-Tensorflow), Matlab (Deep Learning & Neural Networks Toolbox).

• 68 Classifying accelerometer readings using LSTM

Spec: Build, train & validate deep recurrent network to classify waveform.

Language: Python (Keras-Tensorflow), Matlab (Deep Learning & Neural Networks Toolbox).

• 69 Predicting engine failure using LSTM

Spec: Build, train & validate deep recurrent network to predict continuous values.

Language: Python (Keras-Tensorflow), Matlab (Deep Learning & Neural Networks Toolbox).

• 121 Text classification using word embedding and LSTM

Spec: Classify text by mapping words into numerical vectors (signals).

Language: Python (Keras-Tensorflow), Matlab (Deep Learning & Neural Networks Toolbox, Text Analytics Toolbox).

• 125 Text classification using pretrained word embedding and LSTM

Spec: Classify text by mapping words into numerical vectors.

Language: Python (Keras-Tensorflow), Matlab (Deep Learning & Neural Networks Toolbox, Text Analytics Toolbox).

• 124 Word-by-word text generation using word embedding and LSTM

Spec: Predict next word in text by mapping words into numerical vectors.

Language: Python (Keras-Tensorflow), Matlab (Deep Learning & Neural Networks Toolbox, Text Analytics Toolbox).

• 123 Character-by-character text generation using character embedding and LSTM

Spec: Predict next character in sequence of characters by mapping characters into numerical vectors.

Language: Python (Keras-Tensorflow), Matlab (Deep Learning & Neural Networks Toolbox, Text Analytics Toolbox).

• 126 Character-by-character text generation using character vocabulary and LSTM

Spec: Predict next character in sequence of characters using their indices in vocabulary.

Language: Python (Keras-Tensorflow), Matlab (Deep Learning & Neural Networks Toolbox, Text Analytics Toolbox).

Deep transfer learning

• 135 Recognizing neurological disease on MRI image using TL with retrained GoogLeNet

Spec: TL (transfer learning) from a network pretrained on images of one set of classes to a network which is supposed to recognize images of another set. The TL is done by retraining.

Language: Python (Keras-Tensorflow), Matlab (Deep Learning & Neural Networks Toolbox).

• 131 Recognizing neurological disease on MRI image using TL with GoogLeNet retrained in deep network designer

Spec: TL (transfer learning) from a network pretrained on images of one set of classes to a network which is supposed to recognize images of another set. The TL is done by retraining.

Language: Matlab (Deep Learning & Neural Networks Toolbox).

• 70 Recognizing neurological disease on MRI image using TL with retrained AlexNet

Spec: TL (transfer learning) from a network pretrained on images of one set of classes to a network which is supposed to recognize images of another set. The TL is done by retraining.

Language: Python (Keras-Tensorflow), Matlab (Deep Learning & Neural Networks Toolbox).

• 132 Recognizing neurological disease on MRI image using TL with AlexNet retrained in deep network designer

Spec: TL (transfer learning) from a network pretrained on images of one set of classes to a network which is supposed to recognize images of another set. The TL is done by retraining.

Language: Matlab (Deep Learning & Neural Networks Toolbox).

• 71 Recognizing neurological disease on MRI image using TL with feature extraction from AlexNet

Spec: TL (transfer learning) from a network pretrained on images of one set of classes to a network which is supposed to recognize images of another set. The TL is done by feature extraction based on integration of neural network (CNN) with machine learning (SVM).

Language: Python (Keras-Tensorflow), Python (Scikit-learn), Matlab (Deep Learning & Neural Networks Toolbox).

• 133 Transfer learning by converting classification NN (project136) into regression NN (project63)

Spec: TL (transfer learning) from a network pretrained for classification to a network which is supposed to run for regression. The TL is done by retraining.

Language: Python (Keras-Tensorflow), Matlab (Deep Learning & Neural Networks Toolbox).

• 10 Fingerprint identification using pretrained & retrained CNN

Spec: TL (transfer learning) from a network pretrained on images of one set of classes to a network which is supposed to recognize images of another set. The TL is done by retraining.

Language: Python (Keras-Tensorflow), Matlab (Deep Learning & Neural Networks Toolbox, Computer Vision Toolbox).

• 49 Airplane detection from satellite using R-CNN on pretrained CNN

Spec: Integrate computer vision (Viola-Jones or ACF) and neural network (CNN).

Language: Matlab (Deep Learning & Neural Networks Toolbox, Computer Vision Toolbox).

• 46 Deer detection from car using feature extraction from pretrained CNN

Spec: TL (transfer learning) from a network pretrained on images of one set of classes to a network which is supposed to recognize images of another set. The TL is done by feature extraction based on integration of neural network (CNN) with machine learning (SVM).

Language: Python (Keras-Tensorflow), Matlab (Deep Learning & Neural Networks Toolbox, Computer Vision Toolbox, Machine Learning Toolbox).

• 57 Birds detection from airplane using feature extraction from pretrained CNN

Spec: TL (transfer learning) from a network pretrained on images of one set of classes to a network which is supposed to recognize images of another set. The TL is done by feature extraction based on integration of neural network (CNN) with machine learning (SVM).

Language: Python (Keras-Tensorflow), Matlab (Deep Learning & Neural Networks Toolbox, Computer Vision Toolbox, Machine Learning Toolbox).

Deep residual NN

• 72 Recognizing neurological disease on grayscale MRI image using residual NN

Spec: Build, train & validate deep residual network to classify images.

Language: Python (Keras-Tensorflow), Matlab (Deep Learning & Neural Networks Toolbox, Computer Vision Toolbox).

• 73 Recognizing neurological disease on colored MRI image using residual NN

Spec: Build, train & validate deep residual network to classify images.

Language: Python (Keras-Tensorflow), Matlab (Deep Learning & Neural Networks Toolbox, Computer Vision Toolbox).

Deep classification NN using autoencoders

• 115 Deep classification NN using autoencoders

Spec: Integrate unsupervised and supervised learning.

Language: Matlab (Deep Learning & Neural Networks Toolbox).

Cluster analysis (unsupervised learning)

• 11 Clustering vital sign features using competitive NN

Spec: Break data into clusters by computing their centroids.

Language: Matlab (Deep Learning & Neural Networks Toolbox).

Shallow fully-connected NN (including feedforward NN and time-series NN with input & output delays)

• 114NN Predicting whether US resident has health insurance or not based on survey of American households

Spec: Build, train & validate shallow network to classify multidimensional data.

Language: Python (Keras-Tensorflow, Scikit-learn), Matlab (Deep Learning & Neural Networks Toolbox, Machine Learning Toolbox).

• 128 Speech denoising using fully-connected and convolutional NN

Spec: Integrate signal processing (spectrogram) and feedforward NN

Language: Python (Keras-Tensorflow), Matlab (Deep Learning & Neural Networks Toolbox, Audio System Toolbox).

• 08 Extracting classification features from fetal heart rate waveform and classifying fetal clinical condition using shallow classification NN

Spec: Build, train & validate shallow network to classify multidimensional data.

Language: Python (Keras-Tensorflow), Matlab (Deep Learning & Neural Networks Toolbox).

• 07 Investigating feed-forward fully-connected NN for regression

Spec: Build, train & validate shallow network to predict continuous values.

Language: Python (Keras-Tensorflow), Matlab (Deep Learning & Neural Networks Toolbox).

• 09 Fetal heart signal recognition using shallow time-series NN

Spec: Build, train & validate shallow recurrent network to classify waveform.

Language: Python (Keras-Tensorflow), Matlab (Deep Learning & Neural Networks Toolbox).

Copyright © 2017-2019 Serguei Kabakov. All Rights Reserved.