Linked Questions

16 votes
3 answers

MNIST digit recognition: what is the best we can get with a fully connected NN only? (no CNN)

To fully understand how it works internally, I'm re-writing a neural network from scratch in Python + numpy only. (As it's for learning purposes, performance is not an issue). Before moving to ...
Basj's user avatar
  • 385
8 votes
2 answers

What values should initial weights for a ReLU network be?

For a standard feed-forward Neural Network, what range should my initial weights fall under if I'm planning to use Rectified Linear Unit as an activation function? A mathematical justification for the ...
Bn.F76's user avatar
  • 201
0 votes
0 answers

Choice of initial value for weights in Tensorflow MNIST Tutorial [duplicate]

I was tweaking the Tensorflow MNIST Tutorial, and I am not clear why the weights are initialized using tf.zeros. If I switch to using tf.random_normal, or tf.truncated_normal, I see a very significant ...
reddragon's user avatar
  • 121
0 votes
1 answer

Neural net: Cost goes down but performance on train isnt! [duplicate]

So, While building a simpel neural net (MLP) for recognizing digits, I ordered my function to print me both the mean cost overall off the train dataset and %currect answers, also over the train ...
Moran Reznik's user avatar
0 votes
1 answer

Why my weight Initialization saturates the activation function?

I am trying to initialize the Hidden Layers of my Convolutional Neural Network with 7 hidden layers, all with ELU Activation with Sigmoid Activation at the end for binary classification. For ...
mamafoku's user avatar
  • 201
1 vote
1 answer

Initialising the weights of a neural network in regards to the Var{y}

So for a neural network, I'm trying to find the best weight $a$ to form the weight interval $[-a, a]$, given the function for the output of a single layer neural network $y = \sigma(Wx)$ where $x \...
roughosing's user avatar
5 votes
2 answers

Why does pre-training help avoid the vanishing gradient problem?

I read that a problem with the Classic approach to deep NN is the vanishing gradient, which is caused by the derivative of the logistic activation function - broadly speaking, the update flowing down ...
volperossa's user avatar
22 votes
0 answers

When should I use the Normal distribution or the Uniform distribution when using Xavier initialization?

Xavier initialization seems to be used quite widely now to initialize connection weights in neural networks, especially deep ones (see What are good initial weights in a neural network?). The ...
MiniQuark's user avatar
  • 2,150
5 votes
4 answers

Deep Neural Network weight initialization [duplicate]

Given difficult learning task (e.g high dimensionality, inherent data complexity) Deep Neural Networks become hard to train. To ease many of the problems one might: Normalize && handpick ...
Joonatan Samuel's user avatar
2 votes
1 answer

Feed forward Neural Network and MSE issues

I've been implementing a Feed-forward Neural Network in C++ and CUDA. It is a basic Multi-layered Feed Forward ANN, using various activation functions (sigmoid bipolar, tanh, tanh scaled, and soft-...
Alex's user avatar
  • 123
120 votes
3 answers

tanh activation function vs sigmoid activation function

The tanh activation function is: $$tanh \left( x \right) = 2 \cdot \sigma \left( 2 x \right) - 1$$ Where $\sigma(x)$, the sigmoid function, is defined as: $$\sigma(x) = \frac{e^x}{1 + e^x}$$. ...
satya's user avatar
  • 1,353