Questions tagged [deep-learning]

The tag has no usage guidance.

Filter by
Sorted by
Tagged with
1 vote
1 answer
33 views

Decision level fusion

I have trained three deep learning models, and used majority voting for the final decision. Applying majority voting has led to improving the accuracy of the classifier with all test samples, except ...
user avatar
  • 367
0 votes
0 answers
58 views

How to remove noise from audio and convert the audio to text

This is a video link https://www.youtube.com/watch?v=78YHir50N4o I have used the audio from this video. I am using SpeechRecognition library to transform it to text but because of the noise it is not ...
user avatar
1 vote
0 answers
49 views

Labels in speaker verification

I'm a beginner. If I am using a Convolutional Neural Networks with Triplet Loss as a loss function (also combined with GAN and a Classifier) for building a model that performs Speaker Verification, ...
user avatar
  • 11
7 votes
2 answers
563 views

A Machine Learning Based Algorithm as an Alternative to the Matched Filter

Consider we have to detect a known signal added with Gaussian noise. In this scenario, the matched filter is known to be an optimal filter for SNR. The question: is there any machine learning ...
user avatar
  • 601
0 votes
0 answers
35 views

Negative SDR result for evaluating audio source separation

I'm trying to use eval_mus_track function of the museval package to evaluate my audio source separation model. The model I'm evaluating was trained to predict vocals and the results are similar to the ...
user avatar
  • 101
0 votes
3 answers
78 views

Is it possible to classify signal samples using STFT without (image-based) spectrograms?

I am aiming to conduct a classification-based study using signals collected from various devices. I've researched other approaches which make use of STFT for producing a spectrogram for speech and EEG/...
user avatar
  • 77
5 votes
1 answer
35 views

Image Standardization for Image Classification (Machine / Deep Learning)

I am trying to write a program to standardize an image which I need to better perform image classification. I found the definition of standardisation which is obtained simply by subtracting the mean ...
user avatar
0 votes
1 answer
64 views

How to convert depth to disparity

if $k = \text{baseline}\cdot\text{focal length}$ is known, then the disparity is the ratio of $k$ to depth $d_\text{image}$: $$D_\text{image} = \frac{k}{d_\text{image}}$$ I have a ...
user avatar
  • 11
1 vote
1 answer
44 views

When a face collapses into a swirling black hole on a video conference call like this is it due to some advanced AI-based error-correction gone wrong?

Within this newscast video1 a video conference call is shown. Between 02:23 and 02:37 the face of the interviewee appears to get ...
user avatar
  • 171
4 votes
3 answers
213 views

Role of window length and overlap in uncertainty principle?

I am trying to predict epilepsy using spectrograms and a convolutional neural network. So far I have achieved a validation accuracy of 86% which i feel like is pretty good. Lots of the papers doing ...
user avatar
0 votes
1 answer
91 views

How to train a FCNN with Spectrogram images?

I'm working on a audio dereverberation deep learning model, based on the U-net architecture. The idea of my project came from image denoising with autoencoders. I feed the reverbered spectrogram to ...
user avatar
5 votes
1 answer
85 views

Image Segmentation Using Deep Learning

I see in many reviews on Autonomous Car how they segment the images with person, cars, etc... How is it achieved in Deep Learning? Could anyone give an example of that? How it is done?
user avatar
  • 468
0 votes
0 answers
22 views

How can I speech enhancement with DNN?

I want to do speech enhancement with DNN on Matlab. I downloaded the TIMIT dataset from the internet as a .wav file. I also downloaded the noises from the Noisex-92 database as .m file. Now I need to ...
user avatar
  • 49
1 vote
2 answers
57 views

How to choose a deep learning model?

I have split the database available into 70% training, 15% validation, and 15% test. I have trained the model and got the following results: training accuracy 100%, validation accuracy 97.61%, test ...
user avatar
  • 367
4 votes
1 answer
86 views

Which Programming Language Should Be Used for Deep Learning (Deep Neural Network [DNN])?

I will do voice activity detection and speech enhancement based deep neural network. However, I don't know whether to do this via matlab or pyhton. In which programming language can I find more ready-...
user avatar
  • 49
4 votes
2 answers
81 views

When to Use Composite Filters and When to Use Separable Filters?

I’m a beginner in image processing, and was wondering since seperated (decomposed) filters help give faster and more efficient results, when do we even need to use composite filters? All I heard is ...
user avatar
  • 43
6 votes
2 answers
223 views

Explain the Process of Spectral Pooling and Spectral Activation in the Context of CNN in Frequency Domain

I am reading the paper Design of an energy efficient accelerator for training of convolutional neural networks using frequency Domain Computation: which uses Frequency Pooling, from Spectral ...
user avatar
0 votes
0 answers
43 views

I have two signals, recorded from the same device. How do I standardize/normalise them?

I essentially have two signals X and Y, recorded with PPG devices. These have been filtered already. I want to standardize(z-score) or min-max scale them but I don't know if I should do this on each ...
user avatar
0 votes
0 answers
21 views

Signal patterns between train and test sets are vastly inconsistent

I am trying to build a machine learning classification model of a given signal dataset of 3 classes (one file for train signal data is another one for test data). I tried different machine learning ...
user avatar
  • 109
1 vote
0 answers
72 views

Can we use AutoEncoder for Sparse Sensing?

Is there a way to introduce sparsity constraint on an autoencoder to achieve compressions in the Cosine/Fourier domain? I want to use the encoder part of the Auto encoder as the feature extractor from ...
user avatar
  • 21
0 votes
1 answer
38 views

Classification of very noisy EMG signals

I'm an absolutely newbie to signal processing. I'm trying to classify EMG signals which are very noisy (decibel values are more than -70 dB in some cases). After applying EMD technique these values ...
user avatar
  • 155
5 votes
1 answer
95 views

Perform Transposed Convolution in Spectral / Frequency Domain?

I'm doing some experimentation on performing end to end generative modeling in the frequency domain. I've got a working convolutional layer, but do not yet have a Conv2DTranspose equivalent. Please ...
user avatar
5 votes
1 answer
63 views

The Meaning of $ \mathbb{E} $ Operator in the Pix2Pix Loss Formula of a Neural Network / Convolutional Neural Network

I've been observing the Pix2Pix Paper - Image to Image Translation with Conditional Adversarial Networks and wondered on formulas. For example, the objective of the CGAN is: , where x - observed image,...
user avatar
  • 183
0 votes
1 answer
44 views

References of deep learning and ai for dsp researchers [closed]

Are there excellent references for machine learning, deep learning and ai for DSP researchers?
user avatar
  • 183
1 vote
0 answers
19 views

Backpropagation in a Network [closed]

Hi does anyone know how to solve this. Im really struggling to figuring this one out. I am unclear of how the backpropagation part would work in this network and how to calculate it for each weight. ...
user avatar
  • 23
0 votes
0 answers
46 views

Audio RMS normalization prior to a DNN classification

I want to normalize my data prior to a deep neural network model (DNN). It was recommended to me to use real mean square (RMS) normalization for audio, though I am not sure this is the best for a DNN. ...
user avatar
  • 621
4 votes
1 answer
187 views

how to handle different durations of audio data?

I am new to signal preprocessing, I read about mel_spectrograms, MFCC's. Now I want to apply it and use the CNN model, But the data which I have for practice is having audio of different durations, ...
user avatar
0 votes
0 answers
30 views

multi-frame image restoration

Suppose we have a sequence of still images each of which has been contaminated by some particles(ex, dust/sand/smoke) making the images very poor in certain areas. What approach would be best to teach ...
user avatar
6 votes
2 answers
549 views

Classic Signal Processing vs Deep Learning / Machine Learning (DNN / ML) Based Signal Processing

Are classic signal processing/statistics based approaches to optimum detection/estimation still relevant/important compared to ML based approaches using DL? There was a time when speech processing was ...
user avatar
1 vote
1 answer
38 views

Why not use synthetic datasets for training machine vision (deep learning) algorithms?

I am studying some examples of how to perform 3D instance segmentation in indoor scenes, and I have noticed many of the available datasets are from real environments. I was wondering with the ...
user avatar
  • 111
1 vote
0 answers
29 views

what should be a good bibliography for deep learning in 2020

What should be a good bibliography for having a good overview of deep learning (on image/signal processing) in 2020 ?
user avatar
0 votes
0 answers
54 views

Remove noise from VGA image

I have 2 color images generated from same X-ray machines, but one of the image is obtained using VGA cable. Therefore the RGB intensities differ in the 2 images. I am using Object detection to detect ...
user avatar
0 votes
0 answers
97 views

PCA with CNN Tensorflow

I need to improve my model of Convolutional Neural Network (CNN). The goal is to recognize facial expression. I've been using some strategies like dropout for regularization and Adam optimazer, but i ...
user avatar
0 votes
1 answer
65 views

How to prepare different input size in CNN

I'm doing a research about personality identification based on their signature using CNN method, however the learning feature for the personality traits have a different input size. I understand that ...
user avatar
0 votes
1 answer
67 views

Deep Learning: Classification vs. Convolution for Signal Restoration

Assume we have vector $X = [x_1, x_2, x_3, x_4 ,..... x_N], ∈ -1,1$. Therefore the value of $x$ is either 1 or -1. The vector $X$ is convoluted with random generated vector $Y$ whose length is the ...
user avatar
  • 593
0 votes
0 answers
31 views

Audio in Image,Graph or WaveForm Representation

Our task is to feed the audio data into the deep learning tensor flow model in the form of image representation(or graphical,waveform). Question is what is the best way of representing audio in image ...
user avatar
0 votes
0 answers
753 views

Understanding liftering as the final step in MFCCs features extraction

In the book here, they apply liftering, as a final step of MFCCs features extraction, to isolate the system component by multiplying the whole cepstrum by a rectangular window centred on lower ...
user avatar
1 vote
0 answers
34 views

Adversarial training: deep learning book

In the Deep Learning book of Ian Goodfellow, p. 261 it is shown how to build an "adversarial example" by adding to an image $x$ another image $x_{adversarial}$ build as epsilon times and image (same ...
user avatar
0 votes
0 answers
52 views

Deep Learning based NDA Channel Estimation

I was wondering if it is possible to use Deep Learning to estimate the Channel Impulse Response for NDA synchronization? I understand Deep Learning is not normally used for regression problems but I ...
user avatar
1 vote
2 answers
97 views

Could modulation/coding schemes be generated from an AI?

Warning: I'm a bit of a noob with respect to this entire field, but I took some classes on it (a few years ago) and found it fascinating. Anyhow, as far as I understand it, modulation is the process ...
user avatar
  • 119
0 votes
1 answer
37 views

Removing vocal but retaining all background ambient noises

I have one audio file containing human speeches, a lot of ambient noises like audience laughter, birds chirping, sounds of natures etc.. Now I want to separate it into two audio files where one ...
user avatar
1 vote
1 answer
73 views

Neural networks in system identification - What type of activation functions?

I made a free software for all operative systems, even Android. It's called Deeplearning2C. It can train a neural network and generate C code and MATLAB-code. C-code for embedded systems and MATLAB-...
user avatar
  • 389
0 votes
0 answers
42 views

convolutional deep Neural Networks for matrix

I have a basic question about using convolutional neural networks. It's not my field but I'd like to read and understand about it. What I know that convolutional neural networks is used for image ...
user avatar
  • 593
1 vote
1 answer
67 views

Which Machine or deep learning algorithm is appropriate of this issue?

Suppose I have $n$ features as $Y$ = ($y_1 , y_2 ...., y_n$), and a matrix of $J$ of dimension $M$x$N$, one feature of $Y$ is selected randomly to be convolved with one random column of $J$ resulting ...
user avatar
2 votes
1 answer
3k views

Frequency Domain Interpolation: Convolution with Sinc Function

I am reading the paper, Design of an energy-efficient accelerator for training of convolutional neural networks using frequency-domain computation, and I came across the following definition of sinc ...
user avatar
1 vote
1 answer
71 views

Are delta and double delta features needed when classifying with LSTM

I understand that the combination of MFCCs, deltas and double deltas is a good feature to be used with HMMs for keyword detection problems. HMMs are limited by Markov Property and this limitation is ...
user avatar
2 votes
1 answer
243 views

Augmentation for EEG signal classification using Deep Learning

Augmentation is a technique that we use in deep learning for expanding the training dataset. It includes different ways of modifying an image and adding it to the training dataset. My question is if ...
user avatar
  • 39
1 vote
1 answer
663 views

How to express STFT and ISTFT as a 1d convolution and 1d deconvolution in tensorflow/keras

I'm trying to implement this paper in tensorflow and keras. At the end of section 3 it says. ...
user avatar
  • 113
2 votes
2 answers
1k views

audio spectrogram normalization

I am performing language classification from audio signals using mel-spectrograms as my inputs for a ResNet. It works well as long as all of my audio data from different languages is from the same ...
user avatar
  • 21
5 votes
1 answer
486 views

How to Remove the Patch Artifacts of Neural Network Denoising Process?

I have written a python script which uses the Noise2Noise: Learning Image Restoration without Clean Data implementation of the Auto Encoder which is useful to remove noise from images. In the original ...
user avatar
  • 155