I have the following problem. We have $4$ separate discrete input values in the range $[-63,63]$. The output is also supposed to be a value in the range $[-63,63]$. Another constraint is that the solution should allow for online learning with singular values or mini-batches, as the dataset is too big to load all the training data into memory.

I have tried the following method, but the predictions are not good.

MLP based feedforward network: $4$ inputs and $127$ outputs. The inputs are being fed without normalization. The number of hidden layers is $4$ with $[8,16,32,64]$ units in each (respectively). So, essentially, this treats the problem like a sequence classification problem. For training, we feed the non-normalized input along with a one-hot encoded vector for that specific value as output. The inference is done the same way. Finding the hottest output and returning that as the next number in the sequence.

I have the following problem. We have $4$ separate discrete inputs, which can take any integer value between $-63$ and $63$. The output is also supposed to be a discrete value between $-63$ and $63$. Another constraint is that the solution should allow for online learning with singular values or mini-batches, as the dataset is too big to load all the training data into memory.

I have tried the following method, but the predictions are not good.

I created an MLP or feedforward network with $4$ inputs and $127$ outputs. The inputs are being fed without normalization. The number of hidden layers is $4$ with $[8,16,32,64]$ units in each (respectively). So, essentially, this treats the problem like a sequence classification problem. For training, we feed the non-normalized input along with a one-hot encoded vector for that specific value as output. The inference is done the same way. Finding the hottest output and returning that as the next number in the sequence.

The problem:

We have four separate discrete values in the range [-63,63]. The output once again is supposed to be a value in the range [-63,63].

  Another constraint is that the solution should allow for online learning with singular values or minibatches as the dataset is too big too load all the training data into memory.

I have tried a few things but they do not work very well. The predictions are not good

What I have tried:

1. MLP based feedforward network : 4 inputs and 127 outputs. The inputs are being fed without normalization. No of hidden layers 4 with [8,16,32,64] perceptrons in each. So essentially this treats the problem like a sequence classification problem. For training we feed the non normalized input along with a one hot encoded vector for that specific value as output. Inference is done the same way. Finding the hottest output and returning that as the next number in the sequence.

I have the following problem. We have $4$ separate discrete input values in the range $[-63,63]$. The output is also supposed to be a value in the range $[-63,63]$. Another constraint is that the solution should allow for online learning with singular values or mini-batches, as the dataset is too big to load all the training data into memory.

I have tried the following method, but the predictions are not good.

MLP based feedforward network: $4$ inputs and $127$ outputs. The inputs are being fed without normalization. The number of hidden layers is $4$ with $[8,16,32,64]$ units in each (respectively). So, essentially, this treats the problem like a sequence classification problem. For training, we feed the non-normalized input along with a one-hot encoded vector for that specific value as output. The inference is done the same way. Finding the hottest output and returning that as the next number in the sequence.

The problem:

We have four separate discrete values in the range [-127,127]. The output once again is supposed to be a value in the range [-127,127].

Another constraint is that the solution should allow for online learning with singular values or minibatches as the dataset is too big too load all the training data into memory.

I have tried a few things but they do not work very well. The predictions are not good

What I have tried:

1. MLP based feedforward network : 4 inputs and 127 outputs. The inputs are being fed without normalization. No of hidden layers 4 with [8,16,32,64] perceptrons in each. So essentially this treats the problem like a sequence classification problem.

The problem:

We have four separate discrete values in the range [-63,63]. The output once again is supposed to be a value in the range [-63,63].

Another constraint is that the solution should allow for online learning with singular values or minibatches as the dataset is too big too load all the training data into memory.

I have tried a few things but they do not work very well. The predictions are not good

What I have tried:

1. MLP based feedforward network : 4 inputs and 127 outputs. The inputs are being fed without normalization. No of hidden layers 4 with [8,16,32,64] perceptrons in each. So essentially this treats the problem like a sequence classification problem. For training we feed the non normalized input along with a one hot encoded vector for that specific value as output. Inference is done the same way. Finding the hottest output and returning that as the next number in the sequence.