Electronic device for encoding event indicated by spatial-temporal input signals and operating method thereof

1. An electronic device comprising:

first to n-th cells (‘n’ is an integer of 2 or more) configured to receive spatial-temporal input signals that indicate an event unit in a time window;

a summation circuit configured to sum first to n-th cell signals recorded in the first to n-th cells for each of first to m-th unit times (‘m’ is an integer of 2 or more) dividing the time window to generate first to m-th summation signals;

a sensor configured to receive spike signals;

a conversion circuit configured to convert the spike signals into the spatial-temporal input signals;

an encoding circuit configured to compare each of the first to m-th summation signals with a threshold value to encode the spatial-temporal input signals into a code of the event unit; and

a memory cell array configured to store the code and output a final code to a neural network or other electronic device to model a neuron of the neural network.

2. The electronic device of claim 1 , wherein the code of the event unit is a one-dimensional array of 1×m.

3. The electronic device of claim 1 , wherein addresses of the first to n-th cells indicate spatial information of the spatial-temporal input signals, and

wherein the first to m-th unit times indicate temporal information of the spatial-temporal input signals.

4. The electronic device of claim 1 , wherein the first to n-th cells are further configured to receive spatial-temporal input signals in second to k-th time windows (‘k’ is an integer of 2 or more) after a first time window that is the time window,

wherein the spatial-temporal input signals in the first time window indicate the event unit that is a first event unit,

wherein the spatial-temporal input signals in the second to k-th time windows indicate second to k-th event units, respectively, and

wherein the first to k-th event units indicate one event object.

5. The electronic device of claim 4 , wherein the summation circuit is further configured to:

sum the first to n-th cell signals recorded in the first to n-th cells for each of the first to m-th unit times dividing the second time window to generate first to m-th summation signals of the second event unit; and

sum the first to n-th cell signals recorded in the first to n-th cells for each of the first to m-th unit times dividing the k-th time window to generate first to m-th summation signals of the k-th event unit.

6. The electronic device of claim 5 , wherein the code is a first code of the first event unit, and

wherein the encoding circuit is further configured to:

compare each of the first to m-th summation signals of the second event unit with the threshold value to encode the spatial-temporal input signals of the second event unit into a second code of the second event unit; and

compare each of the first to m-th summation signals of the k-th event unit with the threshold value to encode the spatial-temporal input signals of the k-th event unit into k-th code of the k-th event unit.

7. The electronic device of claim 6 , wherein the encoding circuit is further configured to store the first to k-th codes in a memory cell array of k×m.

8. The electronic device of claim 7 , wherein the encoding circuit is further configured to store the first to k-th codes in the memory cell array of k×m,

wherein the first to n-th cells are further configured to receive spatial-temporal input signals that indicate a k+1-th event unit in a k+1-th time window after the k-th time window, and

wherein the summation circuit is further configured to sum the first to n-th cell signals recorded in the first to n-th cells for each of the first to m-th unit times dividing the k+1-th time window to generate first to m-th summation signals of the k+1-th event unit.

9. The electronic device of claim 8 , wherein the encoding circuit is further configured to:

compare each of the first to m-th summation signals of the k+1-th event unit with the threshold value to encode the spatial-temporal input signals of the k+1-th event unit into a k+1-th code of the k+1-th event unit;

select codes of the first to k-th codes, based on differences between the k+1-th code and the first to k-th codes; and

sum the selected codes to restore the k+1-th code.

10. The electronic device of claim 6 , wherein the summation circuit is a first summation circuit and the encoding circuit is a first encoding circuit, and

the electronic device further comprises:

a second summation circuit configured to sum the first to k-th codes for each of the first to m-th unit times dividing each of the first to k-th time windows to generate first to m-th summation signals of the event object; and

a second encoding circuit configured to compare each of the first to m-th summation signals of the event object with the threshold value to encode the event object into a final code.

11. The electronic device of claim 10 , wherein the final code of the event object is a one-dimensional array of 1×m.

12. A method of operating an electronic device that encodes an event object indicated by spatial-temporal input signals, the method comprising:

dividing the event object into first to k-th event units based on first to k-th time windows (‘k’ is an integer of 2 or more);

generating first to m-th summation signals of each of the first to k-th event units by summing first to n-th cell signals recorded in first to n-th cells (‘n’ is an integer of 2 or more) that receive the spatial-temporal input signals for each of first to m-th unit times (‘m’ is an integer of 2 or more) dividing each of the first to k-th time windows;

encoding the spatial-temporal input signals into first to k-th codes of the first to k-th event units by comparing each of the first to m-th summation signals with a threshold value;

storing the first to k-th codes in a memory cell array; and

outputting a final code to a neural network or other electronic device to model a neuron of the neural network;

wherein the dividing of the event object into the first to k-th event units includes converting spike signals into the spatial-temporal input signals.

13. The method of claim 12 , wherein a size of the memory cell array in which the first to k-th codes are stored is k×m.

14. The method of claim 12 , wherein the event object is stored in a memory cell array of n×m×k that is distinct from the memory cell array.

15. The method of claim 12 , wherein each of the first to k-th event units of the event object is stored in an n×m memory cell array that is distinct from the memory cell array.

16. The method of claim 12 , wherein the spatial-temporal input signals are digital signals.

17. The method of claim 12 , further comprising:

generating first to m-th summation signals of the event object by summing the first to k-th codes for each of the first to m-th unit times;

encoding the event object into the final code by comparing each of the first to m-th summation signals of the event object with a threshold value; and

storing the final code in the memory cell array.

18. The method of claim 17 , wherein the final code is a one-dimensional array of 1×m.