Questions tagged [training]

The tag has no usage guidance.

109 questions with no upvoted or accepted answers
Filter by
Sorted by
Tagged with
3
votes
0answers
27 views

A Deep CNN model delivering better results with standardization, when compared with normalization

I developed a deep CNN model, based on the architecture discussed in this paper, to generate predictions for time series data. My training data is shown in the figure below: In order to train the ...
3
votes
1answer
33 views

Validation data shall be in broken down into batches or not?

I am using fit_generator to train the model. The training dataset is being read from a generator function which gives data in a constant batch size. Now I want to ...
3
votes
0answers
122 views

Training deep CNN with noisy dataset

I am training a Mask RCNN model with a train dataset that has been generated from some simple computer vision operations (color thresholding) and some morphological filtering. The train set captures ...
3
votes
1answer
209 views

Why do most GAN (Generative Adversarial Network) implementations have symmetric discriminator and generator architectures?

For example, if the discriminator is a vanilla network of n layers, each with n(i) units, then, typically, the generator will also be a vanilla network of n layers, each with n(n-i) units (except the ...
3
votes
1answer
279 views

What are the effects of clipping the reward in stability?

I am looking for stabilizing my results of DQN, I found clipping is one technique to do it but I did not understand it completely! 1- what are the effects of clipping the reward, clipping the ...
3
votes
0answers
194 views

How much text is enough to train a good embedding model?

I need to train a word2vec embedding model on Wikipedia articles using Gensim. Eventually, I will use the entire Wikipedia for that but for the moment, I'm doing some experimentation/optimization to ...
2
votes
1answer
47 views

Given a machine learning algorithm, what is the minimum size of the training set for it?

I understand that the more data we have, the more reliable is our model trained on that data. I also understand that the more parameters a machine learning model has, the more training data it ...
2
votes
1answer
60 views

how is correct usage of the validation split in neural networks?

I have a dataset separated in train, test and validation splits. After each epoch, I evaluate the loss and accuracy in the validation split. When the loss in validation split is not better, I stop ...
2
votes
2answers
25 views

How to gather training data for simple voice commands?

I'm trying to build a machine learning model for recognizing simple voice commands like up, down, left, etc. On similar problems based on images, I'd just take the picture and assign a label to it. ...
2
votes
1answer
136 views

Smart data split (train/eval) for Object Detection

I am looking for a smart way of splitting object detection data (images with labelled objects inside them) while taking into account the distribution of the objects themselves and not just the images. ...
2
votes
1answer
51 views

Training a model where each response in the observation data has a different known varience

I have a dataset where each response variable is the number of successes of N Bernoulli trials with N and p (the probability of success) being different for each observation. The goal is to train a ...
2
votes
1answer
22 views

Running multiple times of a model is for model randomness or data randomness?

When a paper report the average and std of a model on a dataset, it means that they have changed the split of training and test sets and run the model multiple times or they just run the model on ...
2
votes
0answers
38 views

Replacing mean by median over batch-size to lessen the impact of outliers

In the case of training a Neural Network on a regression task. Assuming the data has a significant amount of outliers. Provided that the error needs to be RMS and not MAE. Can it be better (as in less ...
2
votes
1answer
326 views

Why is performance worse when my time-series data is not shuffled prior to a train/test split vs. when it is shuffled prior to the split?

We are running RandomForest model on a time-series data. The model is run in real time and is refit every time a new row is added. Since it is a timeseries data, we set shuffle to false while ...
2
votes
0answers
28 views

Correct approach to usage of class labels in cell imaging data

As part of a group project at university, we are given a series of videos of cell cultures over a 24 hour period. A number of these cells (the "knockout" cells) have had a particular gene removed, ...
2
votes
0answers
409 views

How to train the generator in a recurrent GAN (Keras)

I am trying to train a Recurrent GAN that is meant to generate geospatial movement data (sequences of 3-tuples of latitude, longitude and time). You may simply consider it a sequences of vectors with ...
2
votes
0answers
416 views

Time series regression using SVR

I have time series data stored in a data frame as follows: Time, c1, c2, c3 0, 0.55, 0.4 , 0.3 1, 0.8 , 0.1 , 0.6 2, 0.9 , 0.5 , 0.7 .... And I want to ...
2
votes
0answers
120 views

How to resolve the instability of average reward per episode in training of DQN (Deep Q-Network)?

what is shown when average reward per episode in training is unstable? If there is big difference between average reward per episode and final reward by test section, what we can say? For ...
2
votes
0answers
21 views

Force Neural Net to attempt to predict every class

I am training a (deep) neural net to classify approximately 60 different classes. The range of occurrences of each class in the dataset is wide, 3 orders of magnitude from the most represented to the ...
2
votes
0answers
332 views

Matlab: setting static iterations per epoch in a CNN

I'm building a convolutional neural network using Matlab's neural network toolbox. I have code designed to cross-train the network with different data sets, using the previous network's layers in ...
2
votes
0answers
604 views

Backpropagation with step or threshold activation function

I understand that gradient descent is local and it deals only with the inputs to the neuron, what it outputs and what it should output. In all I've seen, gradient descent needs the activation function ...
2
votes
0answers
35 views

Term for Methods of Representing Repeated Text in Classifier

A colleague told me that there are terms for two different methods of representing repeated text in the training set for a classifier, but he could not recall them. What are the terms for the options ...
2
votes
0answers
248 views

How to add a new label to a multi-label dataset (like Open Images)

Given N classes in a multi-label dataset and a trained classifier C, how would we add a new class N+1 to the dataset, and fine-tune the trained classifier C such that it now predicts N+1 labels? (lets ...
2
votes
0answers
118 views

Multiple models vs. Single model for prediction

I am using the Darknet Convolutional Neural Networks to detect people (as in, humans) and furniture in a single image. If I train the model twice, one for people, one for furniture. I seem to get ...
1
vote
0answers
18 views

How to teach algorithm to mimic paths in a certain enviroment

I have a set of scenarios which represent the movement of a car in a certain environment containing some obstacles. So for each scenario I have the position of the car (x,y,t) and a description of the ...
1
vote
1answer
105 views

Training an ensemble of small neural networks efficiently in TensorFlow 2

I have a bunch of small neural networks (say, 5 to 50 feed-forward neural networks with only two hidden layers with 10-100 neurons each), which differ only in the weight initialization. I want to ...
1
vote
1answer
56 views

How to split a dataset into train and test sets for time series (multiple step-multiple output forecasting)?

I am trying to use a LSTM neural net to do multiple step / multiple output forecasting (I predict multiple values in one time knowing some values in the past). But, I have realized that I must be ...
1
vote
1answer
42 views

Neural network is getting partially trained

So I am writing my own neural network library using back-propagation as my training algorithm. Everything seems fine the error is getting decreased more and more at each iteration however when I am ...
1
vote
1answer
27 views

How interpret keras training loss without compare with validation loss?

I have several implementation of the same neural network, but each one with different starting parameter. This is one of my plot comparing the training loss of the base experiment with the training ...
1
vote
1answer
71 views

Training data : forecasted or actual?

I am working on a time series prediction problem. I am using keras models for machine learning. For this prediction, weather variables are used as input. They can be of two types: forecasted and ...
1
vote
1answer
83 views

Accuracy and Loss in MLP

I am trying to explore models for predicting whether the a team will win or lose based on features about the team and their opponent. My training data is 15k samples with 760 numerical features. Each ...
1
vote
0answers
30 views

Performance diagnostics in mxnet gluon (e.g. plotting training vs validation loss over time)?

Tensorflow has tensorboard, is there any recommended way to plot classification error/loss over time in mxnet?
1
vote
0answers
25 views

Unexpected shape of “training curves” in NN

I'm trying to find the best configuration for my NN (in terms of batch size, learning rate etc) and noticed the following unexpected behavior. The AUC scores, computed on validation data, as ...
1
vote
0answers
22 views

Strangeness in validation loss between CPU vs GPU when training CNN

I've been training an implementation of Mask R-CNN and it was training very successfully on my CPU but I've just set up my GPU and it is giving some strange results when looking at my validation loss. ...
1
vote
0answers
25 views

Dynamic window regression model

I have a signal and want to predict y which present Number of requests, using regression models. Currently, I am using OLS regression model to predict y. But the prediction error is very high, as my ...
1
vote
1answer
49 views

Effect of adding extra unrelated features to linear perceptron

Suppose that we are training a linear regressor (perceptron). Adding extra features that are not related to the target (e.g. randomly generated values) before training will typically ____ our training ...
1
vote
0answers
64 views

Add training data to YOLO post-training

I've been playing around with YOLOv3 and obtaining some good results on the ~20 custom classes I trained. However, one or two classes look like they can use some additional training data (not a lot, ...
1
vote
0answers
21 views

How to prepare future data for training

Assume I have a large data of ecommerce website sessions with user id key (a user can have multiple sessions with random time between them). The data is on S3 in json gzipped format. On some sessions ...
1
vote
0answers
25 views

Training Gaussian Restricted Boltzmann Machines with Noisy Rectified (nrelu or ssu) linear hidden units

I'm not sure how to implement this architecture. I'm following this thesis (pages 17-19) or this paper but I'm not sure how to train it. I want to use this to extract features from raw audio. I know ...
1
vote
0answers
47 views

How can I train a model to modify a vector by rewarding the model based on the modified vectors nearest neighbors?

I am experimenting with a document retrieval system in which I have documents represented as vectors. When queries come in, they are turned to vectors by the same method as used for the documents. The ...
1
vote
0answers
25 views

keras evaulate method results vary with equal testset

I trained segnet on a dataset of remote sensing imagery. When I run model.evaluate a set of metrics is returned. When I compile the model again with the same ...
1
vote
0answers
70 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 ...
1
vote
1answer
99 views

Optimization methods used in machine learning

I don't have too much knowledge in the field of ML, but from my naive point of view it always seems that some variant of gradient descent is used when training neutral networks. As such, I was ...
1
vote
0answers
173 views

How many epochs to run during hyperparameter search?

If I'm doing a hyperparameter search and comparing two different hyperparameters (but not number of epochs), is there some established rule of thumb for how many epochs to run? If I just compare ...
1
vote
0answers
63 views

How to select samples for a trainings set

My dataset contains half a million unlabeled entries with over 100 binary features. A third of these features are present in less than 1000 samples. I want to classify a few samples by hand (into ...
1
vote
0answers
139 views

Ideal difference in the training accuracy and testing accuracy

In a data classification problem (with supervised learning), what should be the ideal difference in the training set accuracy and testing set accuracy? What should be the ideal range? Is a difference ...
1
vote
0answers
91 views

Inconsistent validation accuracy? is that expected?

I am currently training a pattern classification network, and seem to get very inconsistent result. The network seem to overfit, but the validation accuracy is very inconsistent. I am currently ...
1
vote
0answers
759 views

Tensorflow Training Time Estimates

When I build models and before I train them, I have the habit of estimating the training time to make sure it is reasonable. This usually comes in the form of my naive estimate, training hours = ...
1
vote
0answers
53 views

Predict arguments of indicator function or the value the indicator function itself

I have an evenly-spaced timeseries $a_1, a_2, a_3,..., a_n$ and a function $f_k=\mathbf 1(a_k-a_{k-5})$, where $\mathbf 1$ is the unit step function. I want to predict $f_k$ using $a_1, a_2, a_3,..., ...
1
vote
0answers
340 views

Tips for retraining convolutional neural networks given a drastically different loss surface

For the image dataset I am working with, I need to use B&W version of images (otherwise, I would need to build a network to give false colors to a set of my images, since they have an overpowering ...