Questions tagged [data-augmentation]

For questions related to the concept of data augmentation, where a dataset can be augmented in terms of number and diversity of the samples, which can be useful to avoid over-fitting, especially, when the available dataset(s) is relatively small.

Filter by
Sorted by
Tagged with
0 votes
0 answers
19 views

Fine-tuned GPT-2 producing padding tokens and nonsense

I'm trying to fine-tune GPT-2 so that it can produce artificial data similar to this: ...
0 votes
1 answer
23 views

why by adding additional information as number of sequence on dataset can avoid overfitting

I am developing a regression model to analyze walking styles. The dataset I am using to build the model is from 2 different sources, let's call them dataset A and dataset B. Dataset A has a shape of <...
0 votes
0 answers
19 views

NLP - F1 score for positive class drops to 0 after data augmentation

I'm working on a 3-class text classification problem where my initial class distribution looked like this: positive: 50% negative: 25% and neutral: 25% And training on a model on this slightly ...
0 votes
1 answer
56 views

Why does data augmentation using synthetic data generated by one model improve the performance of another model?

I understand from articles like this one that synthetic data generated by one model based on real data can improve the performance of a second model. Can anyone help me understand the intuition behind ...
-1 votes
1 answer
39 views

Generating synthetic time series data with limited data

I would like some opinions on my current situation. I have a set of time series data that I want to forecast. The data however is not very long (around 500 rows) so I was looking into generating many ...
0 votes
0 answers
14 views

If you have a small amount of labeled data and a limitless amount of pseudo-labeled data, does the ratio of labeled to pseudo-labeled data matter?

Suppose I have a labeled dataset $L$ and unlabeled dataset $U$, where $U \gg L$. Suppose I focus on a subset of $U$ called $u$ and generate a subset of $u$ I'll call $u_L$ that consists of ...
0 votes
1 answer
29 views

Is data augmentation beneficial even if the dataset is large/diverse enough?

I'm training a deep learning model to map binary images to grayscale values of the same shape. For the dataset, I can genearate one as large and diverse as I want it to be. My question is: let's say ...
0 votes
2 answers
32 views

Can I shuffle data for delivery duration forecast problem?

I'm new to ML and trying to write a solution to a food delivery duration time problem (so called lead time). I used algorithms such as random forest and gradient boosting which gave OK results but not ...
1 vote
1 answer
73 views

Is data augmentation inducing bias?

I am using Keras to build a CNN model to classify spectograms and using the following layers: ...
1 vote
1 answer
41 views

How to preserve Markov Property in Deep Reinforcement Learning when using "mixup" or "mixreg"?

I've read through these two papers: (original about "mixup") https://arxiv.org/pdf/1710.09412.pdf (variant for RL, "mixreg") https://arxiv.org/pdf/2010.10814.pdf They are about a ...
5 votes
1 answer
148 views

Does the term "data augmentation" imply increasing the training dataset?

I have a manuscript that has been reviewed and one of the reviewers commented on my use of the term " data augmentation", saying that it might not be the appropriate term in my case (...
0 votes
0 answers
77 views

How to annotate images for instance segmentation with lots of objects per image?

I have (thermal) images which contain multiple objects which I want to extract with instance segmentation. The objects do not overlap and all belong to the same single class. Do I need to annotate all ...
  • 133
2 votes
1 answer
107 views

Is using separate channels of a RBG image a valid data augmentation technique?

Suppose there is a ML network that takes grayscale images as the input. The images that I have are RGB images. So, instead of converting these RGB images to grayscale, I treat each individual colour ...
0 votes
2 answers
45 views

Performance of augmented dataset with or without original images

I am training on yolo and I had a small dataset. I decided to increase it by augmenting it with rotation, shearing, etc to increase the size and increase accuracy. Now I have seen augmented datasets ...
0 votes
0 answers
91 views

Test accuracy go down after decreasing learning rate

My project include classification of images into several classes. I'm having a strange issue related to adding mixup augmentation. The accuracy of the training set and the validation set keep rising ...
  • 183
1 vote
1 answer
133 views

What exactly is data augmentation?

Data augmentation is useful in training. But, I am not sure when can a modification applied to data can be called data augmentation. Suppose a technique is applied to the instances of a dataset and ...
  • 3,481
2 votes
0 answers
42 views

What is the sensible amount of augmentation?

I am playing with the transforms from Torchvision. There are plenty of different kinds of these like: Resize RandomCrop ...
1 vote
0 answers
72 views

Batch normalization for multiple datasets?

I am working on a task of generating synthetic data to help the training of my model. This means that the training is performed on synthetic + real data, and tested on real data. I was told that batch ...
  • 221
1 vote
2 answers
975 views

Train Validation Test Splitting After or Before Data Augmentation?

I have seen tutorials online saying that you should do data augmentation AFTER doing the train/val/test split. However, when I go online to read some research papers, I see numerous instances of ...
  • 21
2 votes
1 answer
106 views

Data augmentation for very small image datasets

I am looking for techniques for augmenting very small image datasets. I have a classification problem with 3 classes. Each class consists of 20 different shapes. The shapes are similar between the ...
1 vote
0 answers
21 views

Does distribution of data augmentation parameters matter?

Idea Let's say we have simple pictures dataset containing 40x40 images of digits. We have only one image of each digit. We want to use that as training set, but we need more data, so we use data ...
1 vote
2 answers
126 views

Is creating dataset only by augmentation a bad practice?

I wonder if creating data set only by augmentation base images is a bad practice. I mean the situation when you have to train net to predict really simple patterns, for example printed-like digits. ...
2 votes
1 answer
388 views

Should one rescale (normalize) image before or after data augmentation?

During image preprocessing pipeline, should one rescale each pixel value to [0, 1] by dividing 255 first, and then perform data transformation such as color distortion, gaussian blur? or vice versa? I ...
2 votes
6 answers
408 views

How do I increase the size of an (almost) balanced dataset?

I am trying to add more data points in my (almost) balanced dataset for training my neural network. I have come across techniques such as SMOTE or Random Over Sampling, but they work best for ...
1 vote
1 answer
812 views

What is the difference between feature extraction with or without data augmentation?

Here's an extract from Chollet's book "Deep Learning with Python" about using pre-trained CNN to predict class from a photo set (p. 146): At this point, there are two ways you could proceed:...
  • 113
1 vote
2 answers
62 views

Late Onset Augmentation

If I train a U-Net model for image segmentation (e.g. medical images) and start training until it converges and then add augmentation - can i expect similar results as if i train with augmentation ...
0 votes
0 answers
54 views

Data Augmentation of store images using handwritten labels

I am new to AI and NN. I've started learning using Geron's book on Tensorflow. My first project ("Smart Shelf") is to determine which items in a store have been purchased and need refilled. ...
0 votes
0 answers
66 views

Is there any rule of thumb to determine the amount of data needed to train a CNN

I am training an AlexNet Convolutional Neural Network to classify images in a dataset. I want to know if there is any general rule for using data augmentation in training a neural network. How can I ...
  • 140
3 votes
0 answers
33 views

If random rotations are included in the data augmentation process, how are the new bounding boxes calculated?

When studying bounding box-based detectors, it's not clear to me if data augmentation includes adding random rotations. If random rotations are added, how is the new bounding box calculated?
2 votes
1 answer
152 views

Should I remove the text overlaying some images in the dataset before training the CNN?

If I am attempting to train a CNN on some image data to perform image classification, but some of the images have pieces of text overlaying them (for the purpose of description to humans), then is it ...
2 votes
0 answers
895 views

How much should we augment our training data?

I am wondering how much I should extend my training set with data augmentation. Is there somewhere a pre-defined number I can go with? Suppose I have 10000 images, can I go as far as 10x or 20x times, ...
  • 21
1 vote
0 answers
130 views

What is the amount of test data needed to evaluate a CNN?

I have an image dataset of about 400 images. 70% of these data points were used for training, 15% for validation, and 15% for testing. I am using the 70% to train a CNN-based binary classifier. I ...
1 vote
1 answer
263 views

Do I need to rotate the masks, if I also rotate the images and the masks are generated from the input?

I am training a neural network that takes an input (H, W, 3) and has the output of size (H', W', C). Now, to augment my dataset, ...
  • 165
2 votes
1 answer
81 views

What could cause a big fluctuation of the loss in the last epochs of training an AlexNet?

I am training an AlexNet neural network, with about 12000 images which 80% is for training, 10% is for validation and another 10% is for testing. I have a problem in my plots. There is a big ...
  • 140
2 votes
2 answers
193 views

Validation Loss Fluctuates then Decrease alongside Validation Accuracy Increases

I was working on CNN. I modified the training procedure on runtime. As we can see from the validation loss and validation accuracy, the yellow curve does not fluctuate much. The green curve and red ...
7 votes
2 answers
443 views

How does rotating an image and adding new 'rotated classes' prevent overfitting?

From Meta-Learning with Memory-Augmented Neural Networks in section 4.1: To reduce the risk of overfitting, we performed data augmentation by randomly translating and rotating character images. We ...
  • 171
2 votes
2 answers
1k views

What is the effect of training a neural network with randomly generated fake data that satisfies certain constraints?

I have a neural network with 2 inputs and one output, like so: ...
  • 123
2 votes
1 answer
2k views

How to label edited images after data augmentation?

I am new to neural networks, I've only started studying and learning about the subject a year ago, and I just started building my first neural network. The project is a little bit ambitious: A browser ...
  • 176
4 votes
3 answers
235 views

Would this relatively small dataset be enough to train a CNN?

Scenario: I am trying to create a dataset with images of choice for different animal classes. I am going to train those images for classification using CNN. Problem: Let's assume I somehow don't have ...