Questions tagged [weight-initialization]
Use this tag when asking about the weight initialization of neural networks which are used in machine and deep learning.
50
questions
0
votes
1
answer
36
views
Train CNN weights by using FFT - Reinforcement Learning?
Assume that you are doing convolution inside a CNN network, by using FFT because FFT is much way faster than using 4-5 for-loops etc.
But how should I train the weights if I know the output of my CNN ...
- 115
0
votes
0
answers
19
views
How to put WEIGHT to the binary data?
I have a dataset with ship and vessel AIS messages as well as other pertinent information. Names of dataset features are similar to those in the table below. Data from the Hamelin and Minden locks ...
1
vote
0
answers
12
views
confusion with Xavier Initiliazation definition
When researching online, I keep finding that Xavier/Glorot initialization is:
however, the original paper by Glorot said that this was a common initialization strategy that they soon found did not ...
- 11
0
votes
0
answers
36
views
Initialization of Nodes in Pytorch mps
I’ve been running some GAN tests both on my local machine (Apple M1 with mps) and on a remote server (with cuda), and I recently realized that the very same network can generate sufficient MNIST ...
0
votes
0
answers
23
views
Rules/Guidelines for Custom Weightage and Hyper-parameter tuning
I have a movie and user-ratings dataset. After implementing the content-based filtering technique, I figured, I can improvise the results even further by assigning weightage to the parameters based on ...
- 111
0
votes
0
answers
38
views
CNN sharing weights in feature map
what do they mean when they say all
neurons in a channel share weights with one another? Do they mean that in a chanel or a featue map the weights are the same ?
- 13
0
votes
0
answers
11
views
Assigning Final Scores to Identified Technologies: Considering Users' Reputation, badge counts, post scores, no.of posts, & post date
I am trying to determine the importance of various factors in assigning a score for identified technologies using the user's StackOverflow post tags and content.
The considering factors are users' ...
- 103
2
votes
0
answers
81
views
TF: What is the difference between the 'kernel weights' and the 'recurrent kernel weights' in LSTMs/GRUs?
Context:
I am trying to understand the differences between the GRU/LSTM cells from tensorflow and pytorch (for research reproducibility) and noticed that TensorFlow differentiates between the ...
- 164
0
votes
0
answers
87
views
Effect of torch weight_norm when dim = None
torch.nn.utils.weight_norm(module, name='weight', dim=0)
When dim = None, g parameter becomes equal to $\|v\|$.
Therefore, $w=g \frac{v}{\|v\|} = v$.
So, I think ...
- 101
1
vote
1
answer
123
views
How to add significance weighting in user based collaborative filtering
I have been learning about recommender systems these past days. More specifically about the collaborative filtering. While exploring I found that it can be useful to use "significance weighting&...
- 31
0
votes
1
answer
315
views
pytorchs LSTMs use of 'bias' and 'weight' strings
Hi I am new to RNN and have come across this the following implementation of Pytorchs LSTM, but I cant understand how (or why) the 'bias' and ...
- 125
0
votes
2
answers
83
views
NNs for fitting highly oscillatory functions
in a scientific computing application of neural networks, I have to maximize several neural networks with scalar output with respect to a target/loss function (coming from a weak form of a PDE).
It is ...
- 1
4
votes
4
answers
1k
views
Why are deep learning models unstable compare to machine learning models?
I would like to understand why deep learning models are so unstable. Suppose I use the same dataset to train a machine learning model multiple times (for example logistic regression) and a deep ...
- 151
1
vote
0
answers
69
views
Is saddle point a cause for the vanishing gradient problem
I am a beginner to neural networks and I am writing a report summarising on the causes and solutions to the vanishing gradient problem. From what I have read, the 2 main causes are the repeated ...
- 113
6
votes
1
answer
2k
views
Where Does the Normal Glorot Initialization Come from?
The famous Glorot initialization is described first in the paper Understanding the difficulty of training deep feedforward neural networks. In this paper, they derive the following uniform ...
- 167
0
votes
0
answers
46
views
Create weights network with randomly initialized weights for Keras Models
I work with a tool for audio feature extraction which has several layers (DenseNet, etc) for the extraction. The default is to use pre-trained imagenet weights, however I want to evaluate the ...
- 21
1
vote
0
answers
55
views
Question regarding weight initialization of an artificial neural network
This is what i'm trying to implement in Python.
w0,...,w8 = vector w1 of shape (9,1)
w9,...,w11 = vector w2 of shape (3,1)
b0 (first bias) is of shape (3,1)
b1 is of shape (1,1)
vector X is of shape (...
- 11
1
vote
1
answer
68
views
Shared classifier for 3 neural networks (is this weights sharing?)
I would like to create 3 different VGGs with a shared classifier. Basically, each of these architectures has only the convolutions, and then I combine all the nets, with a classifier.
For a better ...
- 229
1
vote
0
answers
983
views
Bad marshal data YoLo model
I tried to run a project from repo and got the following log which, I believe, tells a problem with weights load.
...
- 131
0
votes
1
answer
44
views
CNN Design for Counting on Simple Images
This is the first CNN I'm designing following college examples and assignments. I'm working on a CNN that I'd like to use to classify images by the number of shapes on them. My basic problem is that I ...
- 3
1
vote
1
answer
102
views
How to force a NN to ouput the same output given a reverse input?
I want to choose an architecture that can deal with an input symmetry.
As input, I have a sequence of zeros and ones, like [1, 1, 1, 0, 1, 0] and at the output layer I have N neurons that outputs a ...
1
vote
1
answer
3k
views
class weights formula for imbalanced dataset
I am trying to make some semantic segmentation. I have 7 imbalanced classes in my case.
I found several methods for handling Class Imbalance in a dataset is to perform Undersampling for the Majority ...
- 11
1
vote
1
answer
226
views
Update of mean and variance of weights
I'm trying to understand the Bayes by Backprop algorithm from the paper Weight Uncertainty in Neural Networks, the idea is to make a NN in which each weight has it's own probability distribution. I ...
- 21
1
vote
1
answer
405
views
Mathematical bias and weight vs machine learning bias and weight
I am a little confused about the term Bias and Weight with respect to machine learning.
Say we want to predict the heights of people whose weights are given. So plot weights to x-axis and height to ...
- 359
1
vote
0
answers
16
views
Can we talk about vanishing activations?
When updating the weights of a deep neural network using backpropagation, to update the weights of a given hidden layer, we use both the partial derivatives of the objective function with respect to ...
- 201
1
vote
1
answer
224
views
Matrix factorization how to initialize weights and biases?
I have a matrix factorization and I'm wondering how I should initialize its weights and biases.
When getting prediction (recommendation), after computing a dot product and adding bias I want to use ...
- 111
3
votes
2
answers
69
views
Result of uniform weight initialization in all neurons
Background
cs231n has the question regarding how to initialize weights.
Question
Please confirm or correct my understandings. I think the weight value will be the same in all the neurons with ReLU ...
- 575
2
votes
1
answer
514
views
Why is it okay to set the bias vector up with zeros, and not the weight matrices?
We do not initialize weight matrices with zeros because the symmetry isn’t broken during the backward pass, and subsequently in the parameter updating process.
But it is safe to set the bias vector up ...
- 280
1
vote
1
answer
73
views
Initializing weights that are a pointwise product of multiple variables
In two-layer perceptrons that slide across words of text, such as word2vec and fastText, hidden layer heights may be a product of two random variables such as positional embeddings and word embeddings ...
- 111
0
votes
1
answer
108
views
Does training of neural networks follow the same order in each epoch?
Each epoch uses the weight from the end of the previous epoch(correct me if I am wrong). Is the updating of parameters after each batch always in the same order? To rephrase, are the batches always in ...
- 189
0
votes
1
answer
29
views
Why don't different output weights break symmetry?
My deep learning lecturer told us that if a hidden node has identical input weights to another, then the weights will remain the same over the training/there will be no separation. This is confusing ...
- 113
1
vote
1
answer
37
views
Weight-initialisation and model instability
Trying to calibrate a relatively vanilla NN, setting the hyper-parameter tuning aside*, it appears that weight initialisation has a lot of impact on the model output. Ie. Models calibrated with ...
- 2,567
1
vote
2
answers
198
views
What is wrong with a neural network model which is so dependent on the seed of initialization?
I have a fully-connected neural network with one hidden layer with 2 units which its goal is to classify a dataset with 324 samples and 300 features. 50% of the dataset is used for train and 50% of it ...
- 379
1
vote
0
answers
101
views
Is the hyperbolic tangent function a solution to the weight clipping problem of WGAN?
Instead of clipping to the range $[-c,c]$ in WGAN (Wasserstein generative adversarial network), why not smoothly map into the range $[-c,c]$ by using $c\times \mathrm{tanh}(w)$? This would guarantee ...
- 1
1
vote
1
answer
231
views
Should weight distribution change more when fine-tuning transformers-based classifier?
I'm using pre-trained DistilBERT model from Huggingface with custom classification head, which is almost the same as in the reference implementation:
...
- 121
4
votes
2
answers
953
views
Compare Coefficients of Different Regression Models
in my project, I am using asuite of shallow and deep learning models in order to see which has the best performance on my data. However, in the pool of shallow machine learning models, I want to be ...
- 187
6
votes
1
answer
2k
views
Is it wrong to use Glorot Initialization with ReLu Activation?
I'm reading that keras' default initialization is glorot_uniform.
However, all of the tutorials I see are using relu ...
- 519
1
vote
0
answers
42
views
How could we use a neuron instead of 100 neurons in 0.01 standard Weight initialization
Many books explain that we don't need to use 100 neurons and just can use a neuron if all result values(sigmoid(Wx)) are same at each of hidden layers.
but, I don't know that we can just use a ...
- 115
1
vote
0
answers
79
views
Cold-start problem in Real Time Bidding
I'm currently on the reading stage of the deployment of an RTB system. I've seen the problem of a cold start (having no initial guess of how to bid) in several papers, but I haven't really seen it be ...
- 111
0
votes
1
answer
30
views
Why is the variance going down so much in this weight initialization problem(using pytorch)?
first look at this example
>>> x = t.randn(512)
>>> w = t.randn(512, 500000)
>>> (x @ w).var()
tensor(513.9548)
it makes sense that ...
- 199
1
vote
1
answer
30
views
Same probability for all classes
I implemented a fully connected MLP of shape [783 (input), 128 (hidden layer) and 10 (output)] the hidden layer had a sigmoid activation function and the output a sofmax.
I tested with the dataset of ...
1
vote
1
answer
217
views
How can I have the same initialization for all my networks?
I want to have the same weights for layer initializations in all my networks, so that when I'm comparing their first epoch loss they all start from the same value. Is there a way in keras to do this?
...
- 221
2
votes
1
answer
468
views
A model that only works by setting all initial weights to zero
In this model from MusicNet, they set the initial weights of their neural network to all zeros.
...
- 123
2
votes
3
answers
646
views
weight training speed too slow in CNNs
I'm writing my own CNN code from scratch. Though I got fast, converged and satisfactory results, the trained weights change very little in value (while cost/loss function drops in time rapidly in a ...
- 227
3
votes
1
answer
365
views
How large of a value should a weight have in a neural network?
If you're assigning random values to the weights in a neural network before back-propagation, is there a certain maximum or minimum value for each weight ( for example, 0 < w < 1000 ) or can ...
- 33
1
vote
1
answer
148
views
Weights initialization in Neural Network
I was viewing code for custom neural network for sentiment analysis. It had 3 layers (1 hidden layer). I am more concerned with weight initialization for the layers
...
- 135
0
votes
2
answers
217
views
C++ return array from function
I would like to implement machine learning algorithm in C++ without using any C++ machine learning library. So I'm writing this initializer function for generating zero matrices but can't figure out ...
6
votes
2
answers
296
views
What are the reasons for drawing initial neural network weights from the Gaussian distribution?
Are there theoretical or empirical reasons for drawing initial weights of a multilayer perceptron from a Gaussian rather than from, say, a Cauchy distribution?
- 161
10
votes
2
answers
18k
views
what is difference between the DDQN and DQN?
I think I did not understand what is the difference between DQN and DDQN in implementation.
I understand that we change the traget network during the running of DDQN but I do not understand how it is ...
- 1,824
7
votes
2
answers
8k
views
What are the cases where it is fine to initialize all weights to zero
I've taken a few online courses in machine learning, and in general, the advice has been to choose random weights for a neural network to ensure that your neurons don't all learn the same thing, ...
- 272