Questions tagged [loss-functions]

For questions related to the concept of loss (or cost) function in the context of machine learning.

Filter by
Sorted by
Tagged with
19
votes
3answers
24k views

Understanding GAN loss function

I'm struggling to understand the GAN loss function as provided in Understanding Generative Adversarial Networks (a blog post written by Daniel Seita). In the standard cross-entropy loss, we have an ...
11
votes
1answer
3k views

Loss jumps abruptly when I decay the learning rate with Adam optimizer in PyTorch

I'm training an auto-encoder network with Adam optimizer (with amsgrad=True) and ...
6
votes
1answer
404 views

What loss function to use when labels are probabilities?

What loss function is most appropriate when training a model with target values that are probabilities? For example, I have a 3-output model. I want to train it with a feature vector $x=[x_1, x_2, \...
6
votes
1answer
432 views

What's the advantage of log_softmax over softmax?

Previously I have learned that the softmax as the output layer coupled with the log-likelihood cost function (the same as the ...
5
votes
3answers
3k views

Can the mean squared error be negative?

I'm new to machine learning. I was watching a Prof. Andrew Ng's video about gradient descent from the machine learning online course. It said that we want our cost function (in this case, the mean ...
5
votes
1answer
107 views

What is the formula used to calculate the loss in the FaceNet model?

The FaceNet model returns the loss of the predictions and ground-truth classes. How is this loss calculated?
4
votes
3answers
194 views

Which function $(\hat{y} - y)^2$ or $(y - \hat{y})^2$ should I use to compute the gradient?

The MSE can be defined as $(\hat{y} - y)^2$, which should be equal to $(y - \hat{y})^2$, but I think their derivative is different, so I am confused of what derivative will I use for computing my ...
4
votes
1answer
61 views

Is there a reason to choose regular momentum over Nesterov momentum for neural networks?

I've been reading about Nesterov momentum from here and it seems like a nice improvement over regular momentum with no extra cost whatsoever. However, is this really the case? Are there instances ...
4
votes
1answer
76 views

Loss function for choosing a subset of objects

I'm trying to train a neural net to choose a subset from some list of objects. The input is a list of objects $(a,b,c,d,e,f)$ and for each list of objects the label is a list composed of 0/1 - 1 for ...
4
votes
1answer
65 views

How to define a loss function for a classifier where the confusion between some classes is more important than the confusion between others?

I have a dataset of images belonging to $N$ classes, $A_1, A_2...A_n,B_1,B_2...B_m$ and I want to train a CNN to classify them. The classes can be considered as subclasses of two broader classes $A$ ...
4
votes
1answer
80 views

Why is Jensen-Shannon divergence preferred over Kullback-Leibler divergence in measuring the performance of a generative network?

I have read articles on how Jensen-Shannon divergence is preferred over Kullback-Leibler in measuring how good a distribution mapping is learned in a generative network because of the fact that JS-...
4
votes
1answer
159 views

Why does the binary cross-entropy work better than categorical cross-entropy in a multi-class single label problem?

I was just doing a simple NN example with the fashion MNIST dataset, where I was getting 97% accuracy, when I noticed that I was using Binary cross-entropy instead of categorical cross-entropy by ...
4
votes
0answers
14 views

How do weights changes handles during back-propagation when there are unknown labels

I have a question about how weights are updated during back-propagation for some of my samples that have unknown labels (please note, unknown, not missing). The reason they are unknown is because this ...
3
votes
3answers
269 views

What's the function that SGD takes to calculate the gradient?

I'm struggling to fully understand the stochastic gradient descent algorithm. I know that gradient descent allows you to find the local minimum of a function. What I don't know is what exactly that ...
3
votes
2answers
63 views

Could error surface shape be useful to detect which local minima is better for generalization?

The following plot shows error function output based on system weights. Two equal local minima are shown in green pointers. Note that the red dots are not related to the question. Does the right one ...
3
votes
1answer
472 views

What's the difference between RMSE and Euclidean distance, and when to use a custom loss? [closed]

I'm searching for a loss function that fits my project. Actually, I have two questions, but they are in the same direction. I take a look at the definition of the root mean squared error (RMSE) and ...
3
votes
2answers
85 views

How to calculate the advantage in policy gradient functions?

From my understanding of the REINFORCE policy gradient method, we gently nudge the probabilities of actions based on the advantages. More specifically, the positive advantages increase the ...
3
votes
1answer
231 views

Alphazero policy head loss not decreasing

I am now working on training an alphazero player for a board game. The implementation of board game is mine, MCTS for alphazero was taken elsewhere. Due to complexity of the game, it takes a much ...
3
votes
1answer
127 views

How is equation 8 derived in the paper “Self-critical sequence training for image captioning”?

In the paper "Self-critical sequence training for image captioning", on page 3, they define the loss function (of the parameters $\theta$) of an image captioning system as the negative expected reward ...
3
votes
1answer
365 views

Advantages of Kullback-Leibler over L1/L2?

I've recently encounter different articles that are recommending to use KL instead of L1/L2 norm when trying to minimize a probability distribution. But none of the articles are giving a clear ...
3
votes
2answers
58 views

If loss reduction means model improvement, why doesn't accuracy increase?

Problem Statement I've built a classifier to classify a dataset consisting of n samples and four classes of data. To this end, I've used pre-trained VGG-19, pre-trained Alexnet and even LeNet (with ...
3
votes
1answer
562 views

Dice loss gives binary output whereas binary crossentropy produces probability output map

On recommendation of Kanak on stackoverflow I am posting this question here: Currently I am experimenting with various loss functions and optimizers for my binary image segmentation problem. The loss ...
3
votes
1answer
1k views

How do I calculate the gradient of the hinge loss function?

With reference to the research paper entitled Sentiment Embeddings with Applications to Sentiment Analysis, I am trying to implement its sentiment ranking model in Python, for which I am required to ...
3
votes
1answer
115 views

Should the input to the negative log likelihood loss function be probabilities?

I am trying to train a supervised model where the output from the model is output of a linear function $WX + b$. Kindly note that I'm not using any softmax or $\log$ softmax on the result of the ...
3
votes
1answer
38 views

How to add weights to one specific input feature to ensure fair training in the network?

I am trying to create a multiclass product-rating network based on product reviews and other input features. Two of the other input features are "product category" and "gender". However, I want to ...
3
votes
1answer
62 views

when should I create a custom loss function?

Hi I'm using neural network to solve a multi regression problem. I'm trying to predict continuous values, to be more specific I'm making a tracking algorithm to track the position of an Object, I'm ...
3
votes
0answers
47 views

When to use RMSE as opposed to MSE and vice versa?

I understand that RMSE is just the square root of MSE. Generally, as far as I have seen, people seem to use MSE as a loss function and RMSE for evaluation purposes, since it exactly gives you the ...
2
votes
2answers
390 views

In which cases is the categorical cross-entropy better than the mean squared error?

In my code, I usually use the mean squared error (MSE), but the TensorFlow tutorials always use the categorical cross-entropy (CCE). Is the CCE loss function better than MSE? Or is it better only in ...
2
votes
3answers
113 views

How do you interpret this learning curve?

Loss is MSE; orange is validation loss, blue training loss. The task is NN regression (18 inputs, 2 outputs), one layer 300 hidden units. Tuning the lr, mom, l2 regularization parameters this is the ...
2
votes
1answer
29 views

Should illegal moves be excluded from loss calculation in DQN algorithm?

I'm implementing DQN algorithm to train my agent to play a turn-based game. The action space for the game is small, but not all moves are available at all the states. Therefore, when deciding on which ...
2
votes
2answers
290 views

What are the major differences between cost, loss, error, fitness, utility, objective, criterion functions?

I find the terms cost, loss, error, fitness, utility, objective, criterion functions to be interchangeable, but any kind of minor difference explained is appreciated.
2
votes
4answers
201 views

How to stop DQN Q function from increasing during learning?

Following the DQN algorithm with experience replay: We calculate the $loss=(Q(s,a)-(r+Q(s+1,a)))^2$. Assume I have positive but changing rewards. Meaning, $r>0$. Thus, since the rewards are ...
2
votes
1answer
173 views

Heavy loss and inaccurate answer in pytorch

As my first AI model I have decided to make an AI model to predict multiplication of two numbers EX - [2,4] = [8]. I wrote the following code, but the loss is very high, around thousands, and it's ...
2
votes
1answer
1k views

Why is the hyperbolic tangent with MSE better than the sigmoid with cross-entropy?

Usually, in binary classification problems, we use sigmoid as the activation function of the last layer plus the binary cross-entropy as cost function. However, I have already experienced (more than ...
2
votes
1answer
37 views

Is there a way of deriving a loss function given the neural network and training data?

There is some sort of art to using the right loss function. However, I was wondering if there is a way to derive the loss function if I gave you a neural network model (the weights) as well as the ...
2
votes
1answer
94 views

Why does TensorFlow docs discourage using softmax as activation for the last layer?

The beginner colab example for tensorflow states: Note: It is possible to bake this tf.nn.softmax in as the activation function for the last layer of the network....
2
votes
1answer
32 views

How should I penalize the model proportionally to the error?

I am making an MNIST classifier. I am using categorical cross-entropy as my loss function. I want to make it so that if the correct label is 3, then it will penalize the model less heavily if it ...
2
votes
1answer
49 views

When and how to use a mix of loss functions for back-propagation?

I am trying to understand the best loss function to be used with a convolutional neural network. I came to know that we can mix two loss functions. Can any body share in what case was it done and how?
2
votes
1answer
35 views

What loss function is appropriate for finding “points of interest” in a array of x,y inputs

I am looking into whether a neural network is appropriate to detect "points of interest" (POI) in a set of tuples (say length, and some sensor value). A POI is essentially a quick change in the value ...
2
votes
1answer
289 views

Are the training loss and validation loss plotted per sample or per batch?

I am using a CNN to train on some data, where training size = 21700 samples, and test size is 653 samples, and say I am using a batch_size of 500 (I am accounting for samples out of batch size as well)...
2
votes
1answer
273 views

Which loss function should I use for binary classification?

I plan to create a neural network using Python, Keras and TensorFlow. All the tutorials I have seen so far are concerned with image recognition. However, the goal of my program would be to take in 10+ ...
2
votes
2answers
73 views

How do we get the true value in the prediction objective in reinforcement learning?

In the book "Reinforcement Learning: An Introduction" (2018) Sutton and Barto define the prediction objective ($\overline{VE}$) as follows (page 199): $$\overline{VE}\doteq\sum_{s\epsilon S} \mu(s)[v_{...
2
votes
1answer
300 views

Chess policy network

I am interested in making a simple chess engine using neural networks. I already have a fairly good value network but I can't figure out how to train a policy network. I know that Leela chess zero ...
2
votes
0answers
23 views

Single label classification into hierarchical categories using a neural network

I am working on a classification problem into progressive classes. In other words, there is some hierarchy of categories in such a way, that A < B < C, e.g. low, medium, high, very high. What ...
2
votes
0answers
35 views

Is Mean Squared Error Loss function a good loss function for continuous variables $0 < x < 1$

Suppose I am utilising a neural network to predict the next state, $s'$ based on the current $(s, a)$ pairs. all my neural network inputs are between 0 and 1 and the loss function for this network ...
2
votes
0answers
20 views

Is there any wrong in my focal loss derivation?

Assume $\mathbf{X} \in R^{N, C}$ is the input of the softmax $\mathbf{P} \in R^{N, C}$, where $N$ is number of examples and $C$ is number of classes: $$\mathbf{p}_i = \left[ \frac{e^{x_{ik}}}{\sum_{j=...
2
votes
0answers
33 views

Why does GAN loss converge to log(2) and not -log(2)?

In Goodfellow's paper, he says: Hence, by inspecting Eq. 4 at $D^*_G (\mathbf{x}) = \frac{1}{2}$, we find $C(G) = \log \frac{1}{2}+ \log \frac{1}{2} = − \log 4$. To see that this is the best ...
2
votes
0answers
53 views

Why is the loss associated with my neural network increasing?

I am currently learning neural networks. Using data from http://www.mariofrank.net/touchalytics/index.html, I am trying to predict "User ID" by training the neural network model shown below. However, ...
2
votes
0answers
23 views

Tversky Loss paper implementation: Recall/Precision do not improve as stated

I have been trying to implement this paper and I am very much intrigued. I am working on a medical image problem where I have to segment very small specimens on Whole Slide Images (gigapixel ...
2
votes
0answers
34 views

How to understand my CNN's training results?

I created a multi-label classification CNN to classify chest X-ray images into zero or more possible lung diseases. I've been doing some configuration tests on it and analyzing its results and I'm ...