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
6
votes
1answer
937 views

what actually word embedding dimensions values represent?

I am learning word2vec and word embedding , I have downloaded GloVe pre-trained word embedding (shape 40,000 x 50) and using this function to extract information from that: ...
1
vote
0answers
454 views

Generator loss not decreasing- text to image synthesis

I am implementing Scott Reed's paper on Generative Adversarial Text to Image Synthesis.(https://arxiv.org/pdf/1605.05396.pdf) The dataset I am using is a simple one, consisting of images of circles, ...
4
votes
1answer
972 views

“other” class in Image classification

In the MNIST dataset, you have 10 defined classes, one for each digit. But you don't have a "not a digit" class. It seems that most image classification datasets are the same. But in a business ...
1
vote
2answers
1k views

Image segmentation training set labeling

I am new to pytorch and Deep learning. I am trying to do image segmentation. But , I am stuck at how to label training set images. Can anyone please help me ? This is one of my training image I ...
5
votes
4answers
1k views

Alternatives to Logistic Regression

I have age, gender, height, weight and some other similar parameters of 15000 subjects. I also have one column showing if they had a medical condition (present in about 20% subjects). I now want to ...
1
vote
2answers
97 views

Weights in neural network

So I am newbie in deep learning, I came across activation functions which gives an output and compares it to label, if it's wrong, it adjusts its weight until it gives the same output as labelled data ...
0
votes
1answer
57 views

Technique to make NN (specifically autoencoders) more robust by artifcally making input data more sparse

I believe some month ago I have read somewhere that autoencoders can respond better to sparse input data when trained with such. For example when the training data is not sparse, but many features of ...
1
vote
1answer
727 views

trying to decrease overfitting with regularisation in CNN

I am doing transfer learning by retraining the publicly available inception layer, without regularisation here are my initial parameters and results: ...
3
votes
1answer
531 views

How to deal with skewed imbalanced image dataset to work with CNN?

I am working on multi-class classification problem on an image dataset. There is one class with 80% of the images and rest 20% is divided into rest 6 remaining classes. If I have to apply the image-...
1
vote
1answer
445 views

Resources to learn Tensorflow and Keras [closed]

What all should I learn in sequence to be expert in Tensorflow and Google Keras provided I know Python and Numpy, Pandas, Pyplot basics? Please provide the resources too as most online tutorials are ...
0
votes
1answer
54 views

Which cloud platform to maximize my impact as a data scientist? [closed]

I am looking to pick up the knowledge/software skills to move towards becoming an end to end deep learning engineer. By this I mean handling the following on my own: preprocess big data at low ...
1
vote
1answer
69 views

How does a Q algorithm consider future rewards?

I am trying to understand the underlying logic of Q learning (deep Q learning to be precise). At the moment I am stuck at the notion of future rewards. To understand the logic, I am reviewing some of ...
1
vote
1answer
51 views

Censored output data, which activation function for the output layer and which loss function to use?

I am building a neural network to solve a regression problem. The output is a single numerical value. Unfortunately, the output is censored: the values below 0 were recorded as 0, and postive values ...
1
vote
0answers
557 views

How to concatenate feature vectors of different dimensions?

I have been using different deep learning models and extracting features from different layers for the given images. My code goes like this: ...
2
votes
1answer
1k views

Estimating Predictive Uncertainty for unlabeled data

I am trying to estimate the predictive uncertainty for a deep neural network. While I do have a labeled training set, I´m trying to measure uncertainty for some unlabeled production data. This paper ...
10
votes
4answers
28k views

High model accuracy vs very low validation accuarcy

I'm building a sentiment analysis program in python using Keras Sequential model for deep learning my data is 20,000 tweets: positive tweets: 9152 tweets negative tweets: 10849 tweets I wrote a ...
1
vote
1answer
3k views

Are there any python libraries for sequences clustering?

I have a problem which I explained in other question. I've understood that my dataset is a sequence of states or something like that. Is there libraries to analyze sequence with python? And is it ...
1
vote
0answers
67 views

How to estimate Distance to Obstacles using Lidar's BEV(Bir's Eye view ) Representation?

I am using a implementation that use both camera and Lidar for 3D obstacle detection for self driving cars but I have no idea how to calculate the distance to the obstacles since the Lidar uses the ...
1
vote
2answers
31 views

What to do when we can't trust our human classifiers?

Suppose we want to design a neural network that can diagnose skin cancer. We want this neural network to consider the possibility that the doctor we hired misclassified some of our images while ...
1
vote
0answers
313 views

What should be the value of parallel iterations in tensorflow RNN implementations?

tf.nn.dynamic_rnn() and tf.nn.raw_rnn() take in an argument called parallel_iterations. The documentation says: ...
2
votes
3answers
513 views

What are the benefits of having ML in js? [closed]

What are the benefits of having ML in JavaScript I.e. the deeplearn.js (now tensorflow) stuff, as opposed to implementing the ML steps in a python backend?
2
votes
2answers
4k views

How to check and correct misspelling in the data of pairs of words?

I have user generated text containing names of ports often containing typos and the actual port names. I would like to correct the misspelling of user generated text containing the names of ports. Can ...
7
votes
1answer
199 views

Are deep learning models way over the required capacity for their datasets' estimated entropies?

this question might seem a bit odd. I was doing some self-studies into information theory and decided to do some more formal investigations into deep learning. Please bear with me as I try to explain. ...
2
votes
1answer
216 views

Any difference between adding epochs and duplicating data for neural nets?

Let's say I am training a neural net (e.g. convolutional network or LSTM). Generally, the longer the training (more epochs) leads to better accuracy, albeit at times at the expense of overfitting. ...
2
votes
1answer
300 views

Validation loss is less than training loss by 5 units. How this result is interpreted?

Iam training a Keras model for end-to-end speech recognition. I have my own dataset of speech containing about 400 wave files. Text transcriptions is also given as input. Model summary is: ...
1
vote
0answers
609 views

Why does my model accuracy rise and then drop, with the loss sharing similar characteristics?

I'm working on a multiclass classification problem using Keras. I have nearly 17k training data, and approximately 14k testing data (validation comes from the 17k training data). I have a number of ...
1
vote
0answers
25 views

Why do most of the research papers simply adopt mainstream CNN architectures designed for ImageNet dataset (AlexNet, VGG, ResNet, Inception etc.)?

Simply stacking multiple layers is not feasible. We need to optimize, and efficiently design the network, which has good learning capabilities (depending on the dataset domain). So what are the ...
1
vote
1answer
97 views

Scientific name for the problem given in the description. (In machine learning what kind of problem it is termed as)

Features marked as boolean values (around 50,000) , need to score a set of fixed output (around 25,000). I have wikipedia topics relevant to a particular url as features. And the output is the ...
2
votes
2answers
2k views

Order of normalization / augmentation for image classification

I'm currently working on a common image classification with CNN. I would like to use both normalization (substract mean / divide by std per channel) and data augmentation (rotation, color, blur, ...) ...
1
vote
1answer
162 views

Does classification of a balanced data-set lead to any problem?

So I came across a bioinformatics paper, where I found a line which says: One potential problem with using a training set with equal numbers of positive and negative examples in cross-validation ...
1
vote
0answers
29 views

How do Dynamic Memory Network scale to large inputs?

How do Dynamic Memory Network, for example from the paper Ask Me Anything: Dynamic Memory Networks for Natural Language Processing, scale to large inputs? The paper states the following: In these ...
4
votes
1answer
590 views

Theoretical and practical comparison of CTC and seq2seq loss in Tensorflow

Tensorflow has built-in implementations for both, the Connectionist Temporal Classification (CTC) loss and a special seq2seq loss (weighted cross-entropy). Since CTC loss is also intended to deal ...
3
votes
1answer
172 views

turn off parts of features in a neural network?

Suppose I have a neural network which accepts two sets of features as inputs and generates corresponding outputs, for instance, generate average final grade from: 1. working hours for N students in a ...
4
votes
6answers
452 views

Fast introduction to deep learning in Python, with advanced math and some machine learning backgrounds, but not much Python experience

I've the following somewhat unusual background and I've managed (probably by luck) to get an industry job of a computer vision researcher using deep learning. My background: I've a PhD in pure math, ...
6
votes
3answers
2k views

Are CNNs insensitive to rotations and shifts in images?

Can CNNs predict well if they are trained on canonical-like images but tested on a version of images that are little bit shifted? I tried it using ...
1
vote
1answer
937 views

Validation accuracy for neural network

When training a neural network, I usually plot the accuracy obtained on the validation data (validation accuracy) as an intermediate measure of the network's performance – the final measure being test ...
19
votes
6answers
79k views

Uploading images folder from my system into Google Colab

I want to train a deep learning model on a dataset containing around 3000 images. Since the dataset is huge, I want to use Google colab since it's GPU supported. How do I upload this full image folder ...
5
votes
1answer
631 views

Spatial Transformer Networks vs Deformable Convolutions

As I understand STN as described by the the deepmind paper https://arxiv.org/abs/1506.02025 allow a neural network to learn how to perform spatial transformations on the input image in order to ...
5
votes
1answer
801 views

The connection between optimization and generalization

Optimization algorithms such as gradient descent or particle swarm can find a minima in a function. On the other hand, learning methods such as back-prop define learning as an optimization problem ...
1
vote
0answers
98 views

What does the training time for a Neural Network include?

I recently developed a DNN model and I want to know what exactly is training time and what all steps are included in it? For ex I carried out the following steps 1) Determined best Network ...
5
votes
2answers
3k views

Unsupervised Anomaly Detection in Images

I would like to detect defects/anomalies in images. Due to the lack of images with anomalies, I try to solve the problem in an unsupervised manner. Until now, I trained a variational autoencoder ...
3
votes
1answer
46 views

How to find relation between N components and predict the value of any one component using the predicted relation?

I am new to Machine learning and trying to learn by practicing. I have a situation where I am reading a set of N data. Each of N data will have independent state at any moment of time. I want to use ...
2
votes
1answer
2k views

Drawing 1D CNN architecture

How can I draw CNN Architecture like this one here:
4
votes
2answers
2k views

How to sort numbers using Convolutional Neural Network?

Recently, in an interview I got this question: Design a convnet that sorts numbers. Operators are ReLU, Conv, and Pooling. E.g. input: 5, 3, 6, 2; output: 2, 3, 5, 6 I am not sure how can you sort a ...
1
vote
2answers
101 views

Homemade deep learning library: numerical issue with relu activation

For the sake of learning the finer details of a deep learning neural network, I have coded my own library with everything (optimizer, layers, activations, cost function) homemade. It seems to work ...
3
votes
1answer
77 views

Setting “missing” distance values to zero when training a neural network

Not sure if missing values is the right name to use here. I want to train a DNN on data given by a sensor. The sensor gives the (x,y) coordinates of the founded objects. The sensor can keep track of ...
1
vote
1answer
139 views

Forward and backward process in pyTorch

When I write a network, do I have to write the whole forward property in nn.Module.forward()? I mean if I do some operations outside the net, does grad correctly ...
5
votes
1answer
7k views

What is the input size of Alex net

In the paper ImageNet Classification with Deep Convolutional Neural Networks, the size of input image is 224x224. The following figure shows the input size. From caffe, deploy.prototxt file from the ...
2
votes
1answer
124 views

Is color information only extracted in the first input layer of a convolutional neural network?

In a convolutional neural network (CNN), since the RGB values get multiplied in the first convolutional layer, does this mean that color is essentially only extracted in the very first layer? ...
6
votes
2answers
911 views

In a convolutional neural network (CNN), when convolving the image, is the operation used the dot product or the sum of element-wise multiplication?

The example below is taken from the lectures in deeplearning.ai shows that the result is the sum of the element-by-element product (or "element-wise multiplication". The red numbers ...

1
51 52
53
54 55
70