How to deal with memory in java?

Question

Java

So, I decided to play around with neural networks, for this I chose the FANN library, made a small network, trained it, everything seems to be fine, well, I decided what little things I will do right away with a network that will recognize Linux Torvald)) (in the pictures 240X320) I did everything , what you need and moved on to creating the network itself and training it, I create:

layerList.add(Layer.create(76800, ActivationFunction.FANN_SIGMOID_SYMMETRIC, 0.01f));
        layerList.add(Layer.create(153600, ActivationFunction.FANN_SIGMOID_SYMMETRIC, 0.01f));
        layerList.add(Layer.create(153600, ActivationFunction.FANN_SIGMOID_SYMMETRIC, 0.01f));
        layerList.add(Layer.create(153600, ActivationFunction.FANN_SIGMOID_SYMMETRIC, 0.01f));
        layerList.add(Layer.create(1, ActivationFunction.FANN_SIGMOID_SYMMETRIC, 0.01f));

And what do you think happened? Correctly ! jvm ran out of memory.
So I have two questions:
1. How to allocate more RAM to jvm (I have 16gb on my computer)
2. Are there ways to work with memory more economically?

Answer 1

1. How to allocate jvm more RAM (I have 16gb on my computer)

With the xms and xmx command line options when you call java.exe.
Radically: abandon Java, I'm not a data scientist, but I have a person who does, and he does not use Java, and I do not see much advantages in it for neural networks.

Answer 2

2. Are there ways to work with memory more economically?

- As advised to abandon Java.
- Throw more RAM
- Optimize the code (do not shit code)
- Clusters?
- Transfer part of the logic to native
- Transfer data to the database?
- Rewrite a more optimized lib?