Homomorphic encryption

1. A method implemented in hardware, in software, or in a combination thereof, the method comprising:

receiving data characterizing tailoring of a characteristic of an input of secure data into a recurrent artificial neural network, wherein the data characterizes either:

synapses and nodes into which bits of the secure data are to be injected, or

an order in which bits of the secure data are to be injected;

tailoring the input of the secure data into the recurrent artificial neural network in accordance with the data;

homomorphically encrypting secure data, comprising

determining whether signal transmission activity between three or more nodes in the recurrent artificial neural network comports with patterns of signal transmission activity, wherein the signal transmission activity is responsive to the tailored input of the secure data into the recurrent artificial neural network, and

storing binary data, wherein each digit in the binary data represents whether the signal transmission activity in the recurrent artificial neural network comports with a respective pattern of the patterns of signal transmission activity; and

making the stored binary data available for statistical analysis that draws conclusions about the secure data.

2. The method of claim 1 , wherein the patterns of signal transmission activity in the recurrent artificial neural network comprise simplex patterns of signal transmission activity in the network.

3. The method of claim 2 , wherein the simplex patterns enclose cavities.

4. The method of claim 1 , wherein determining whether the signal transmission activity comports with the patterns comprises:

determining a timing of signal transmission activity having a complexity that is distinguishable from other signal transmission activity that is responsive to the input; and

determining whether the signal transmission activity comports with the patterns based on the timing of the signal transmission activity that has the distinguishable complexity.

5. The method of claim 1 , wherein the method further comprises:

tailoring the response of the recurrent artificial neural network to the input of the secure data by changing one or more properties of a node or a link within the network.

6. A method implemented in hardware, in software, or in a combination thereof, the method comprising:

receiving data characterizing tailoring a characteristic of an input of first data into a recurrent artificial neural network, wherein the data characterizes either:

synapses and nodes into which bits of the first data are to be injected, or

an order in which bits of the first data are to be injected;

tailoring the input of the first data into the recurrent artificial neural network in accordance with the data characterizing the tailoring of the characteristic;

homomorphically encrypting first data, including:

inputting the first data into a recurrent artificial neural network,

identifying patterns of signal transmission activity between three or more nodes within the recurrent artificial neural network, wherein the patterns of signal transmission activity are responsive to the tailored input of the first data, and

storing second data representing whether the identified patterns of signal transmission activity comports with topological patterns; and

making the stored second data available for statistical analysis that draws conclusions about the first data.

7. The method of claim 6 , wherein the identified patterns of signal transmission activity in the recurrent artificial neural network comprise simplex patterns of signal transmission activity in the network.

8. The method of claim 7 , wherein the simplex patterns enclose cavities.

9. The method of claim 6 , wherein identifying the patterns of signal transmission activity comprises:

determining a timing of signal transmission activity having a complexity that is distinguishable from other signal transmission activity that is responsive to the input; and

identifying the patterns based on the timing of the signal transmission activity that has the distinguishable complexity.

10. The method of claim 6 , wherein the method further comprises:

tailoring the response of the network to the input of the first data by changing one or more properties of a node or a link within the network.

11. A homomorphic encryption device comprising:

a recurrent artificial neural network comprising an input configured to receive secure data, wherein the input includes an input layer or one or more injection sites into nodes or links in the recurrent artificial neural network;

data processing apparatus configured to:

receive data characterizing tailoring of a characteristic of an input of secure data into the recurrent artificial neural network, wherein the data characterizes either:

synapses and nodes into which bits of the secure data are to be injected, or

an order in which bits of the secure data are to be injected;

tailor the input of the secure data into the recurrent artificial neural network in accordance with the data; and

identify patterns of signal transmission activity between three or more nodes within the recurrent artificial neural network, wherein the patterns of signal transmission activity are responsive to the tailored input of the secure data; and

a data storage device configured to store second data representing whether the identified patterns of activity comports with topological patterns.

12. The homomorphic encryption device of claim 11 , wherein the patterns of activity in the recurrent artificial neural network comprise simplex patterns of activity in the network, wherein the simplex patterns enclose cavities.

13. The homomorphic encryption device of claim 11 , wherein the data processing apparatus of the homomorphic encryption device is configured to:

determine a timing of activity having a complexity that is distinguishable from other activity that is responsive to the input; and

identify the patterns based on the timing of the activity that has the distinguishable complexity.

14. The homomorphic encryption device of claim 11 , wherein the data processing apparatus of the homomorphic encryption device is configured to:

tailor the response of the network to the input of the secure data by changing one or more properties of a node or a link within the network.