Questions tagged [deep-learning]

For questions related to deep learning, which refers to a subset of machine learning methods based on artificial neural networks (ANNs) with multiple hidden layers. The adjective deep thus refers to the number of layers of the ANNs. The expression deep learning was apparently introduced (although not in the context of machine learning or ANNs) in 1986 by Rina Dechter in the paper "Learning while searching in constraint-satisfaction-problems".

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

PyTorch: LSTM error while trying to update the hidden state

I am trying to train an LSTM while keeping its hidden state (LSTM stateful) until the moment when I am going to start a new epoch(episode). But here it's come an interesting situation because I am ...
0
votes
2answers
18 views

Do larger numbers of hidden layers have a bigger effect on a classification model's accuracy?

I trained different classification models using Keras with different numbers of hidden layers and the same number of neurons in each layer. What I found was the accuracy of the models decreased as the ...
1
vote
0answers
21 views

The MLP output of a neural network can be written as $\|x\|\|w_l\|\cos(\theta_l)$: why is the norm easier to maximize?

The MLP output of a neural network is a dot product between the weights and the input and therefore can be written as $\|x\|\|w_l\|\cos(\theta_l)$ (see this for more details), where $x$ is the input, $...
3
votes
1answer
490 views

Alternative to sliding window neural network (was: Object detect (or) image classification at specific locations in the frame)

Recent advances in Deeplearning and dedicated hardware has made it possible to detect images with a much better accuracy than ever. Neural networks are the gold standard for computer vision ...
1
vote
0answers
7 views

How does the loss landscape look like or change when a model is overfitting?

My understanding is that when a model starts overfitting, it no longer learns useful features and starts remembering the training data set. Given enough epochs and sufficient parameters, a model can ...
1
vote
1answer
30 views

How to define a “don't care” class in time series classification in Pytorch?

This is a theoretical question. Setup I have a time series classification task in which I should output a classification of 3 classes for every time stamp t. All ...
1
vote
0answers
24 views

Why do you calculate the mean and standard deviation over the complete dataset before training rather than for every batch?

In most implementations of neural networks the features are scaled to make the optimization of the loss function as stable as possible. Mostly a min-max scaler is used. Alternatively, there is also a ...
1
vote
1answer
20 views

What would be the state of the art image captioning deep learning model?

I saw a couple of architectures, like CNN-LSTM, with and without attention model, use of Glove vector, self-critical models, etc. I am overwhelmed looking at different notebooks and architectures, ...
4
votes
2answers
3k views

What is the purpose of “reshaping it into the shape the network expects and scaling it so that all values are in the [0, 1] interval.”?

I am a deep learning beginner recently reading this book "Deep learning with Python", the example explains the process of implementing a greyscale image classification using MNIST in keras, in the ...
0
votes
1answer
19 views

How does high entropy targets relate to less variance of the gradient between training cases?

I've been trying to understand the Distilling the Knowledge in a Neural Network paper by Hinton et al. But I cannot fully understand this: When the soft targets have high entropy, they provide much ...
2
votes
1answer
31 views

How to improve a trained model over time (i.e. with more predictions)?

I built a model using the tutorial on the TensorFlow site. It was a simple image classification neural network. I trained it and saved the model and weights together on a ...
0
votes
0answers
10 views

How to make CNN to recognize whole picture, not just the details?

In my current project I use CNNs to analize plots (CNN autoencoders for feature extraction and KMeans for clusterization) and I get the feeling that these networks, can extract only features that are ...
1
vote
1answer
231 views

Is Mask R-CNN suited to solve a multi-class classification problem where the classes are related?

I want to create a model to solve a multi-class classification problem. Here are more details about my problem. Every picture contains only one object The background is very simple All objects ...
2
votes
1answer
56 views

How to represent integer values in sequence to sequence prediction task in encoder-decoder LSTM?

I have a large 2D grid having 30k rows and 35k columns, so a total of 30x35k grid cells. Each grid cell is represented by a unique integer number (identity of grid cell). I have several trajectories ...
-2
votes
0answers
18 views

extract desire keyword/text pair

I am looking for extract keyword pair from text files. They might not be next to each other and do not have same pattern for each occurrence. And I would not think regex will works because there is no ...
1
vote
1answer
65 views

Can the law of iterated expectation be used on the inner expectation of the DQN cost function described in the DQN paper

Is the expression for the DQN cost function, Equation (2) of the DQN paper $$\begin{align}L_1 &= E_{\mu,\pi}\left[\left(y_i - q(s,a;\theta)\right)^2\right]\\ &=E_{\mu,\pi}\left[\left(E_{\...
2
votes
1answer
78 views

Spikes in of Train and Test error

I learn a DNN for image recognition. During each epoch, I calculate mean loss in the training set. After each epoch, I calculate loss and number of errors over both training and test set. The problem ...
1
vote
1answer
31 views

Do RNNs/LSTMs really need to be sequential?

There are many articles comparing RNNs/LSTMs and the Attention mechanism. One of the disadvantages of RNNs that is often mentioned is that while Attention can be computed in parallel, RNNs are highly ...
2
votes
1answer
80 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 ...
1
vote
1answer
40 views

Can RNNs get inputs and produce outputs similar to the inputs and outputs of FFNNs?

RNN and LSTM models have many interesting architectures that can be modified in various ways. We can also compose their input and output data in quite interesting ways. However, in the examples that I ...
1
vote
1answer
27 views

Can residual connections be beneficial when we have a small training dataset?

I have a classification problem, for which an inadequate amount of training data is available. Also, there is no known practical data augmentation approach for this problem (as no unlabelled data is ...
4
votes
1answer
158 views

Will there be some promising techniques that can make AI greener and affordable in the future?

The recent advances in machine learning were mostly achieved by the hardware, and the hardware is said to continue driving the development of AI, but I was still shocked by this thread which reads ...
13
votes
2answers
644 views

Is there any scientific/mathematical argument that prevents deep learning from ever producing strong AI?

I read Judea Pearl's The Book of Why, in which he mentions that deep learning is just a glorified curve fitting technology, and will not be able to produce human-like intelligence. From his book ...
2
votes
2answers
738 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.
0
votes
0answers
20 views

How can AI algorithms be used in regards to cryptocurrency and token mining?

I am new to AI and lack the knowledge of it's capabilities. A question popped in my head, in regards to Blockchains and the mining of cryptocurrencies and tokens, how can machine learning algorithms ...
5
votes
1answer
116 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 ...
1
vote
0answers
26 views

What are some use cases of discrete optimization in Deep Learning?

When we talk of optimization, it usually boils down to gradient descent and its variants in the context of deep learning. However, I wonder if there are some works that use discrete optimization in ...
45
votes
19answers
15k views

Can digital computers understand infinity?

As a human being, we can think infinity. In principle, if we have enough resources (time etc.), we can count infinitely many things (including abstract, like numbers, or real). For example, at least, ...
1
vote
2answers
34 views

How can “any process you can imagine” be thought of as function computation?

I stumbled upon this passage when reading this guide. Universality theorems are a commonplace in computer science, so much so that we sometimes forget how astonishing they are. But it's worth ...
30
votes
5answers
16k views

What is the purpose of an activation function in neural networks?

It is said that activation functions in neural networks help introduce non-linearity. What does this mean? What does non-linearity mean in this context? How does the introduction of this non-...
0
votes
0answers
15 views

How do the trainable projection layer used in PRADO and pQRNN work?

Trainable projection layers are said to be a very powerful thing but after reading: https://www.aclweb.org/anthology/D19-1506.pdf https://arxiv.org/pdf/2101.08890.pdf I don't understand how it works....
0
votes
2answers
74 views

Could I just choose the other (non-predicted) class when the accuracy is low?

I have a binary classification problem. My neural network is getting between 10% and 45% accuracy on the validation set and 80% on the training set. Now, if I have a 10% accuracy and I just take the ...
13
votes
1answer
162 views

Will parameter sweeping on one split of data followed by cross validation discover the right hyperparameters?

Let's call our dataset splits train/test/evaluate. We're in a situation where we require months of data. So we prefer to use the evaluation dataset as infrequently as possible to avoid polluting our ...
0
votes
1answer
19 views

Handling imbalanced data with multiple targets

I have the model which has 3 outputs (it is a regression task, I have the angle of the steering wheel, brake and acceleration). I can divide my values to some smaller bins and in this way I can change ...
0
votes
1answer
23 views

Validation Accuracy remains constant while training VGG?

I posted this question on stackoverflow and got downvoted for unmentioned reason, so I'll repost it here, hoping to get some insights This is the plot This is the code: ...
1
vote
1answer
56 views

Single-Shot Learning for Object Re-Identification

I am looking for a way to re-identify/classify/recognize x real life objects (x < 50) with a camera. Each object should be presented to the AI only once for learning and there's always only one of ...
0
votes
0answers
23 views

Do Gradient Descent and Natural Gradient solve the same problem?

I am troubled by natural gradient methods. If we have a function f(x) we wish to minimize, gradient descent minimizes f(x) of course, but what does the natural gradient do? I found on https://...
0
votes
1answer
34 views

How to transfer declarative knowledge into neural networks

Humans learn facts about the world like "most A are B" by own experience and by being told so (by other people or texts). The systems and mechanisms of storage and usage of such facts (by an ...
2
votes
2answers
180 views

Is the target assumed to be a noisy version of the output of the model in machine learning?

I wonder if the following equation (you can find it in almost every ML book) refers to a general assumption that we make when using machine learning: $$y = f(x)+\epsilon,$$ where $y$ is our output, $f$...
0
votes
0answers
15 views

Dealing with huge peak in data distribution

I am trying to predict a continuous value using a deep neural network. I have about 100,000 samples, where input is a sequence of RNA, and output is a continuous metric determining the quality of the ...
2
votes
1answer
55 views

Why is the sample size of stochastic gradient descent a power of 2?

I watched in the video lecture of cs224: Stanford CS224N: NLP with Deep Learning | Winter 2019 | Lecture 2 – Word Vectors and Word Senses. They take the sample size of the window to be $2^5 = 32$ or $...
1
vote
0answers
11 views

Use case for dilated pooling operator used in a Machine Learning model?

I have seen that most of the deep-learning frameworks have the ability to do dilated pooling. Many frameworks have recently been updated to add the dilated property to the pooling. However, I have not ...
2
votes
1answer
119 views

Why is this ResNet50 misclassifying objects?

I'm new to Deep Learning, and I have some conceptual problems. I followed a simple tutorial here, and trained a model in Keras to do image classification on 10 classes of logos. I prepared 10 classes ...
5
votes
2answers
103 views

Is it possible to control asymptotic behaviour of neural network models?

Is it possible to specify what the asymptotic behaviour of a Neural Networks (NN) model should be? I am thinking of a NN which tries to learn a mapping $\vec y=f(\vec x)$ with $\vec x$ a vector of ...
0
votes
1answer
42 views

Is the final model scaling done on the full training set?

We have our training set and our test set. When we scale our data we "fit" the scaler transform to the training set and then we scale both the training set and test set using this scaler ...
0
votes
2answers
94 views

Why is non-linearity desirable in a neural network?

Why is non-linearity desirable in a neural network? I couldn't find satisfactory answers to this question on the web. I typically get answers like "real-world problems require non-linear ...
3
votes
1answer
653 views

What's the definition of “singularity” in the context of neural networks?

The paper Skip connections eliminate singularities explains the use of skip connections to break the singularity in deep networks, but I have not fully understood what a singularity is. Any easy-...
0
votes
2answers
29 views

Does anybody know what would happen if I changed input shape of pytorch models?

In this https://pytorch.org/vision/stable/models.html tutorial it clearly states: All pre-trained models expect input images normalized in the same way, i.e. mini-batches of 3-channel RGB images of ...
3
votes
1answer
102 views

Is batch normalization not suitable for non-gaussian input?

I generate some non-Gaussian data, and use two kinds of DNN models, one with BN and the other without BN. I find that the model DNN with BN can't predict well. The codes is shown as follow: <...
0
votes
0answers
16 views

Multi dimensional LSTM modeling in KERAS

I have a database of time series signals with multiple features and Im trying to build a model to predict whether or not two samples are related to each other. For example : a database of 1000 sample ...

1
2 3 4 5
30