Questions tagged [deep-learning]

a new area of Machine Learning research concerned with the technologies used for learning hierarchical representations of data, mainly done with deep neural networks (i.e. networks with two or more hidden layers), but also with some sort of Probabilistic Graphical Models.

Filter by
Sorted by
Tagged with
3
votes
1answer
815 views

Convolutional neural networks for non-image applications?

If I remember correctly, Convolutional neural networks (CNN) have first been developed for image classification purposes (see work from LeCun et al.). The convolution process that "slides" over the ...
1
vote
2answers
94 views

Do anomalous input features to autoencoder result in high errors on the corresponding output features?

An autoencoder is trained by replicating each training instance to both input and output. However, when predicting for anomaly detection, will the output error be local to the same output feature(s) ...
2
votes
1answer
84 views

How to calculate $\phi_{i,j}$ in VGG19 network?

In the paper Photo-Realistic Single Image Super-Resolution Using a Generative Adversarial Network by Christian Ledig et al., the distance between images (used in the loss function) is calculated from ...
1
vote
1answer
651 views

Confused about false positive and false negative in confusion matrix?

I am working on binary classification for classifying cancer=1 and no-cancer=0, I use confusion matrix from sklearn, this is my confusion matrix on test set: ...
2
votes
1answer
67 views

Can I use a GAN to increase my Dataset used for Image detection?

I am currently working on a machine learning project where I use the YOLO Algorithm to detect an object inside of a picture or video. The problem I face is that the specific image set (side-scan sonar)...
7
votes
3answers
7k views

When using Data augmentation is it ok to validate only with the original images?

I'm working on a multi-classification deep learning algorithm and I was getting big over-fitting: My model is supposed to classify sunglasses on 17 different brands, but I only had around 400 images ...
1
vote
1answer
455 views

The mix of leaky Relu at the first layers of CNN along with conventional Relu for object detection

First of all, I know the usage of leaky RELUs and some other relevant leaky activation functions as well. However I have seen in a lot of papers on object detection tasks (e.g YOLO) to use this type ...
1
vote
0answers
154 views

100% classification accuracy

I am trying to perform a multi-class classification where the network is trained to classify objects into 3 categories: cars, pedestrians and miscellaneous. I am using the KITTI Dataset for car ...
6
votes
1answer
659 views

What is the neural network architecture behind Facebook's Starspace model?

Recently, Facebook released a paper concerning a general purpose neural embedding model called StarSpace. In their paper, they explain the loss function and the training procedure of the model, but ...
2
votes
2answers
371 views

Equivariance vs Invariance in Convolutional Neural Networks

Could someone please explain to me in details (possibly from mathematical point of view) what is the role of Equivariance and Invariance in Convolutional Neural Networks, and how are they actually ...
5
votes
2answers
1k views

issue with early-stopping on f1 score with imbalanced data

I have a highly imbalanced dataset with less than 0.5% of the minor class. Using Keras, I'm training DNN on the training set and evaluate performance on validation set. Loss function is ...
4
votes
2answers
1k views

How to use a NN architecture that is too big for GPU?

Initially posted in Stack Overflow. I would like to implement a model which is actually 2 neural networks stacked together. However the size of these 2 architecture is too big to fit in GPU at the ...
3
votes
1answer
870 views

How to extract characteristics from text using machine learning?

I would like to develop some kind of model/algorithm that allows me to extract the characteristics of a given product name. (let's say the brand, model and color). I am looking for a solution similar ...
0
votes
2answers
817 views

Does fine-tuning of transferred layers perform better than frozen transferred layers?

I recently learned concepts of transfer learning. Is it necessarily true that fine-tuning of transferred layers perform better than frozen transferred layer? why?
2
votes
0answers
407 views

Training detector without bounding box data

From what I can see most object detection NNs (Fast(er) R-CNN, YOLO etc) are trained on data including bounding boxes indicating where in the picture the objects are localized. Is there any model ...
0
votes
3answers
61 views

Using deep learning or random forest [closed]

I am very new to machine learning, and I am trying to build a model to classify this data set (UCI heart disease). I have built a simple model using random forest and got an 80% accuracy. The size of ...
1
vote
1answer
49 views

For text classification that has innumerable features, how do I choose the number of neurons and layers for MLPClassifier?

In my use case of text classification (identify the author from a subset of 10 authors), I find that post all processing with trigrams, there are a 100 thousand and odd features with nearly 50k ...
1
vote
0answers
95 views

Trainable parameters explode in CNN , how to control them and yet get desired accuracy

I am building a CNN mode using CSV of size (79999, 11) Here is my model below. When I do a model summary it gives me close to two billion trainable parameters , which my laptop or even AWS Ec2 is not ...
0
votes
1answer
466 views

How can I detect anomalies/outliers in my online streaming data on a real-time basis?

Say, I've a huge set of data(infinite in size) consisting of alternating sine wave and step pulses one after the other. What I want from my model is to parse the data sequence wise or point wise and ...
0
votes
1answer
263 views

train Neural Network with SGD and see that it overfits data.

Suppose you train Neural Network with SGD and see that it overfits data. Which of the following actions can help you to regularize model? Change optimization method to Adam. Insert (or increase rate ...
2
votes
0answers
685 views

How does the multi-input deep learning work?

I have a multi-input convolutional neural network model that inputs 2 images from 2 datasets to give one output which is the class of the two inputs. The two datasets have the same classes. I used 2 ...
1
vote
0answers
39 views

LSTM/RNN seems to be failing at testing

I'm relatively new to ML, keras and tensorflow and I working with a dataset (kerastest.csv) that is 400 lines of this ...
1
vote
0answers
209 views

How to train a multi inputs deep learning model using every combination of inputs? [closed]

I am beginner in deep learning. I want to create a multi inputs CNN model in Keras. The model takes two inputs of images to give the two images class. The two images from differnt datasets that have ...
2
votes
3answers
265 views

What is a suitable loss function and evaluation metric for a classification model with large number of unbalanced target classes?

I am building a multiclass classifier to predict the "Intent" of a question. There are some 100 classes in the target variable and each target class contains an unequal proportion of observations/...
0
votes
1answer
100 views

Multiple GPU in MXNet C++

I am trying to make MXNet (C++ API) learn, with a common sample in C++, on multiple GPU. According to this MXNet forum post, we need to aggregate manually the gradients that we fetch at the ...
3
votes
2answers
3k views

CNN not learning properly

Before marking my question as duplicate, I would like to say that I have tried all the possible solutions mentioned in similar questions, but that doesn't seem to work. I am currently working on ...
6
votes
3answers
25k views

How to change the names of the layers of deep learning in Keras?

I am using vgg16 to create a deep learning model. I want to know how to change the names of the layers of deep learning in Keras? I tried this ...
1
vote
2answers
1k views

Which method should be considered to evaluate the imbalanced multi-class classification?

I am working on multiclass-imbalanced data. My dependent variable is highly skewed. ...
2
votes
1answer
2k views

What is the difference between shuffle in fit_generator and shuffle in flow_from_directory?

I am using Keras to create a deep learning model and I would like to know that what is the difference between shuffle argument in ...
0
votes
1answer
36 views

not quite sure about the difference between RNN and feed forward neural net

I'm a bit confused after reading this paper: https://arxiv.org/abs/1705.09851 on page 22, the author writes response: \begin{equation} Y = softmax(Z^{L-1}) \end{equation} and hidden state \begin{...
2
votes
1answer
563 views

Train a deep reinforcement learning model using two computers

I would like to know if there is a way to train a deep rl model using two different computers. The first one would execute the game and send requisitions to the second computer which would store and ...
1
vote
1answer
2k views

Merge two vgg16 models in Keras

I am working in deep learning project using vgg16. I got the following error and I could not solve it. ...
0
votes
1answer
87 views

Why can distributed deep learning provide higher accuracy (lower error) than non-distributed one with the following cases?

Based on some papers which I read, distributed deep learning can provide faster training time. In addition, it also provides better accuracy or lower prediction error. What are the reasons? Question ...
1
vote
0answers
764 views

Examples for multi-input Convolutional Neural Network

I want to create a multi inputs Convolutional Neural Network (cnn) that takes two inputs and produces one output of the inputs class by using Keras. I searched for resources that explain multi inputs ...
11
votes
2answers
16k views

What is the advantage of using log softmax instead of softmax

i am wondering if there are any advantages of log softmax over softmax. And also, when i should use softmax or log-softmax. is there any specific reason for choosing one over another?
2
votes
1answer
784 views

Give Variable Length input to LSTM

My input data consist of list of list. Both list have dynamic length for every example like below. ...
0
votes
1answer
120 views

Best practice for short sentences in a deep learning network

In a deep learning network (CNN or RNN), we might use word embeddings such as FastText, Glove, etc. to represent the input text. My question is: If I'm working on a data from Twitter, and I have a ...
1
vote
0answers
206 views

Renaming deep learning layers causes bad results

I created two CNN models using Keras. The models give good accuracy results. To concatenate these models together to create a multi input model, I have to change the names of the layers to be unique. ...
1
vote
1answer
61 views

neural network training algorithms

When I first read about neural networks, I learned that Backpropagation is the algorithm used to train the neural network. I am interested if there are other alternatives (or better?) to BP. What ...
0
votes
1answer
3k views

input_dim for Dense Layer after LSTM layers Keras

Do I need to specify the input_dim (which means the number of features in one row/sample) after adding the first LSTM layer for the later Dense layers? I was trying to create an architecture with 2 ...
1
vote
1answer
925 views

Multi-inputs Convolutional Neural Network takes different number of images

I am using Keras to build a CNN model that takes two types of images as inputs (input1, input2) and produce one output. The model classifies the two inputs into a class, and the number of images in ...
2
votes
0answers
392 views

How to tune parameters for Time Series Analysis, when forecasting is only dominated by one feature and error is not getting reduced?

I am trying to predict time series based on 150 features. When I plot correlation of these features, I am getting 20 features with more or less importance but every model I use, it is completely ...
1
vote
1answer
596 views

CNN - Is this a Toeplitz Matrix?

I have been reading through Chapter 9 of www.deeplearningbbook.org, where convolutional networks are being described. The following image represents the output of a 2D convolution, without kernel ...
1
vote
0answers
33 views

Why is this convolution equation easier to apply than it's commutative counterpart?

The convolution is an operation on two functions of a real- valued argument. The convolution operation is typically denoted with an asterisk: s(t) = (x ∗ w)(t) It ...
7
votes
1answer
5k views

Convolution and Cross Correlation in CNN

What would be the intuition behind using the convolution and cross correlation operation inside Convolutional Neural Networks? I am interested in putting together the theory from Digital Image ...
2
votes
1answer
136 views

ROC curve for deep learning

I am working in a deep learning project for image classification. I am using Vgg16 to create the model and the dataset has 100 classes. The testing accuracy is 98.9% and loss is 0.1731 And I got the ...
9
votes
6answers
35k views

How to prepare the varied size input in CNN prediction

I want to make a CNN model in Keras which can be fed images of different sizes. According to other questions, I could understand how to set a model, like ...
6
votes
0answers
2k views

Why is my Keras model not learning image segmentation?

Edit: as is turns out, not even the model's initial creator could successfully fine-tune it. This is most likely a problem of implementation, or possibly related to the non-intuitive way in which the ...
0
votes
0answers
347 views

Why is my VGG16 not learning?

I've been following the series on Keras Python Deep Learning and, afaik, have produced the same code as used in the examples. However, I cannot produce the same result. ...
0
votes
1answer
345 views

LSTM - How to prepare train from a dataset which contains multiple observations for different events

I m using LSTM in a project related to MobiFall dataset which contains falls and daily activitives - such as walking, sitting etc - sensed by accelerometer, gyroscope and orientation sensors in x,y,z ...

1
50 51
52
53 54
80