Neural networks *Must see details
Budget: $2 – $8 USD
Neural network solution
This will require the ability to create an AI program (possibly from scratch but not necessarily), write functions and be able to think conceptually about the problem.
Here is the problem
1. Create a neural network
2. Create a function that can store and recall the unique state of the network
3. Generate a unique combination of nodes
4. Objective is 1 trillion unique combinations
Description
Premise x-> f(x)->y
Propose a dataset of x (10 parameters)
Create a function f(x) to reliably output 1 trillion (y)
Given a combination of x(100 parameters) the Network can generate Y (1 trillion unique combinations)
Given x the network can calculate y
Given y the network can calculate x
and do this reliably
thus the network can store and recall 1 trillion unique states
Delivery
python or Golang Python
Notes
Here is an idea of what the end product should look like
Let me see
Let’s say
X could be any alphanumeric string 1234467890
Y is the state of the network - however you want to notate it , Let’s say I notate that as g1g5g8n12b32f25
When I input{ x=1234567890} I get {y = g1g5g8n12b32f25} and of course if I input {y = g1g5g8n12b32f25} I get { x=1234567890}
This gives me a way to store 1 trillion unique states using an algorithm instead of a table
I collect one trillion x’s and y’s and check for collisions
If no collisions then success
1t is 10^11 - so we may need 11 numbers
This is where I got the idea so maybe it can help
https://www.microsoft.com/en-us/research/blog/metalearned-neural-memory-teaching-neural-networks-how-to-remember/
This is a common problem in cryptography they use distributive properties to solve it. a*(b+c)= (a*b) + (a*c) so given any two you can calculate the third. We just have to think about how to do it in another way. I have been thinking about your idea of a tensor I think your on to something that’s why you won the job
If you have four dimensions of coordinates and you add time you get a near infinite number set and Unicode characters you get 147,000+ character input which is more than enough combinations . I think your on to something
This will require the ability to create an AI program (possibly from scratch but not necessarily), write functions and be able to think conceptually about the problem.
Here is the problem
1. Create a neural network
2. Create a function that can store and recall the unique state of the network
3. Generate a unique combination of nodes
4. Objective is 1 trillion unique combinations
Description
Premise x-> f(x)->y
Propose a dataset of x (10 parameters)
Create a function f(x) to reliably output 1 trillion (y)
Given a combination of x(100 parameters) the Network can generate Y (1 trillion unique combinations)
Given x the network can calculate y
Given y the network can calculate x
and do this reliably
thus the network can store and recall 1 trillion unique states
Delivery
python or Golang Python
Notes
Here is an idea of what the end product should look like
Let me see
Let’s say
X could be any alphanumeric string 1234467890
Y is the state of the network - however you want to notate it , Let’s say I notate that as g1g5g8n12b32f25
When I input{ x=1234567890} I get {y = g1g5g8n12b32f25} and of course if I input {y = g1g5g8n12b32f25} I get { x=1234567890}
This gives me a way to store 1 trillion unique states using an algorithm instead of a table
I collect one trillion x’s and y’s and check for collisions
If no collisions then success
1t is 10^11 - so we may need 11 numbers
This is where I got the idea so maybe it can help
https://www.microsoft.com/en-us/research/blog/metalearned-neural-memory-teaching-neural-networks-how-to-remember/
This is a common problem in cryptography they use distributive properties to solve it. a*(b+c)= (a*b) + (a*c) so given any two you can calculate the third. We just have to think about how to do it in another way. I have been thinking about your idea of a tensor I think your on to something that’s why you won the job
If you have four dimensions of coordinates and you add time you get a near infinite number set and Unicode characters you get 147,000+ character input which is more than enough combinations . I think your on to something