1. A computer-based method of computing a similarity function from a data set of objects, the method comprising the steps of:
obtaining a training set of objects;
presenting a user with one or more subsets of objects based on the training set of objects, each subset comprising at least two objects of the data set;
receiving user feedback regarding similarity between the one or more subsets of objects;
defining one or more sets of feature variables based on features in the one or more subsets of objects;
defining one or more class variables in accordance with the user feedback; and
constructing a similarity function which relates the one or more sets of feature variables to the one or more class variables;
wherein each of the sets of feature variables is defined by a pairwise subtraction of the features in the corresponding subset of objects.
2. The method of claim 1, wherein the one or more class variables are based on one or more similarity values returned by the user.
3. The method of claim 1, wherein each of the sets of feature variables is defined as a parametric function of the features in the corresponding subset of objects.
4. The method of claim 3, wherein one or more parameters of the function are determined in accordance with the user feedback.
5. A computer-based method of computing a similarity function from a data set of objects, the method comprising the steps of:
obtaining a training set of objects;
presenting a user with one or more subsets of objects based on the training set of objects, each subset comprising at least two objects of the data set;
receiving user feedback regarding similarity between the one or more subsets of objects;
defining one or more sets of feature variables based on features in the one or more subsets of objects;
defining one or more class variables in accordance with the user feedback; and
constructing a similarity function which relates the one or more sets of feature variables to the one or more class variables;
wherein the similarity function is constructed for subsequent use in automatically determining the similarity between two or more objects of substantially unknown similarity.
6. Apparatus for computing a similarity function from a data set of objects, the apparatus comprising:
at least one processor operative to: (i) obtain a training set of objects; (ii) present a user with one or more subsets of objects based on the training set of objects, each subset comprising at least two objects of the data set; (iii) receive user feedback regarding similarity between the one or more subsets of objects; (iv) define one or more sets of feature variables based on features in the one or more subsets of objects; (v) define one or more class variables in accordance with the user feedback; and (vi) construct a similarity function which relates the one or more sets of feature variables to the one or more class variables; and
memory, operatively coupled to the at least one processor, for storing at least one of the data set of objects and the constructed similarity function;
wherein each of the sets of feature variables is defined by a pairwise subtraction of the features in the corresponding subset of objects.
7. The apparatus of claim 6, wherein the one or more class variables are based on one or more similarity values returned by the user.
8. The apparatus of claim 6, wherein each of the sets of feature variables is defined as a parametric function of the features in the corresponding subset of objects.
9. The apparatus of claim 8, wherein one or more parameters of the function are determined in accordance with the user feedback.
10. Apparatus for computing a similarity function from a data set of objects, the apparatus comprising:
at least one processor operative to: (i) obtain a training set of objects: (ii) present a user with one or more subsets of objects based on the training set of objects, each subset comprising at least two objects of the data set; (iii) receive user feedback regarding similarity between the one or more subsets of objects; (iv) define one or more sets of feature variables based on features in the one or more subsets of objects; (v) define one or more class variables in accordance with the user feedback; and (vi) construct a similarity function which relates the one or more sets of feature variables to the one or more class variables; and
memory, operatively coupled to the at least one processor, for storing at least one of the data set of objects and the constructed similarity function;
wherein the similarity function is constructed for subsequent use in automatically determining the similarity between two or more objects of substantially unknown similarity.
11. An article of manufacture for computing a similarity function from a data set of objects, the article comprising a machine readable medium containing one or more programs which when executed implement the steps of:
obtaining a training set of objects;
presenting a user with one or more subsets of objects based on the training set of objects, each subset comprising at least two objects of the data set;
receiving user feedback regarding similarity between the one or more subsets of objects;
defining one or more sets of feature variables based on features in the one or more subsets of objects;
defining one or more class variables in accordance with the user feedback; and
constructing a similarity function which relates the one or more sets of feature variables to the one or more class variables;
wherein each of the sets of feature variables is defined by a pairwise subtraction of the features in the corresponding subset of objects.
12. The article of claim 11, wherein the one or more class variables are based on one or more similarity values returned by the user.
13. The article of claim 11, wherein each of the sets of feature variables is defined as a parametric function of the features in the corresponding subset of objects.
14. The article of claim 13, wherein one or more parameters of the function are determined in accordance with the user feedback.
15. An article of manufacture for computing a similarity function from a data set of objects, the article comprising a machine readable medium containing one or more programs which when executed implement the steps of:
obtaining a training set of objects;
presenting a user with one or more subsets of objects based on the training set of objects, each subset comprising at least two objects of the data set;
receiving user feedback regarding similarity between the one or more subsets of objects;
defining one or more sets of feature variables based on features in the one or more subsets of objects;
defining one or more class variables in accordance with the user feedback; and
constructing a similarity function which relates the one or more sets of feature variables to the one or more class variables;
wherein the similarity function is constructed for subsequent use in automatically determining the similarity between two or more objects of substantially unknown similarity.
The claims below are in addition to those above.
All refrences to claim(s) which appear below refer to the numbering after this setence.
1. A reconfigurable, multi-user optical communications network, with low latency time, comprising:
a first transmitter configured to generate a first brightness-modulated light signal from a first digital data signal,
a second transmitter configured to generate a second brightness-modulated light signal from a second digital data signal
optical means configured to combine and transport the first and second light signals to at least one receiver, wherein
each of the first and second transmitters comprise,
a pseudo-random code generator configured to transmit a code generated from a predetermined software key,
means to encode the data signal with said code and configured to output an encoded digital signal,
light emission means, comprising a multimode light source whose optical power is directly modulated by a signal consisting of the encoded digital signal to generate a corresponding encoded light signal, and wherein
the at least one receiver comprises,
direct optical detection means configured to generate an analog electrical signal characteristic of the sum of the combined first and second light signals transmitted,
a pseudo-random code generator identical to the pseudo-random code generator of either the first and second transmitter, and
means to decode said analog signal using a code transmitted by said code generator with a key identical to the predetermined software key associated with at least one of the first and second transmitters and configured to recreate the data signal intended for the user.
2. The optical communications network according to claim 1, wherein said multimode light source is a laser diode.
3. The optical communications network according to claim 1, wherein said direct optical detection means comprise a photodiode associated with a transimpedance amplification stage and an amplification stage.
4. The optical communications network according to one of claims 1 to 3, wherein the optical transport of the encoded light signal is carried out in free propagation.
5. The optical communications network according to claim 1, wherein the encoding means of each of the first and second transmitters comprise digital multiplication means between the data signal and the code generated using said key.
6. The optical communications network according to claim 1, wherein the decoding means of the at least one receiver comprises:
first analog signal formatting means,
analog multiplication means between said formatted analog signal and said code configured to obtain a decoded signal, and
low-pass filtering means applied to the decoded signal configured to recreate at least one of the first and second data signals.
7. The optical communications network according to claim 6, wherein the decoding means further comprises:
means for thresholding, after the filtering means, configured to recreate at least one of the first and second data signals in a digital format.
8. The optical communications network according to claim 1, wherein the codes generated by the pseudo-random code generators are Gold type codes.
9. The optical communications network according to claim 1, wherein the codes generated by the pseudo-random code generators are M-sequence type codes.
10. The optical communications network according to claim 1, applied to a data distribution system with several transmitters and several receivers, wherein the light emission means of each transmitter are virtually identical and comprise a multimode light source whose optical power is modulated according to the encoded digital signal, and wherein each transmitter is associated with a specific software key, network reconfiguration being carried out by the decoding means of the receivers by selecting the key identical to the key specific to each transmitter with which the user wants to communicate.
11. The optical communications network according to claim 1, applied to a system for distribution of data between at least one of the first transmitter and the second transmitters and several receivers, wherein each receiver is associated with a specific software key, network reconfiguration being carried out by the encoding means of the at least one of the first transmitter and the second transmitter by selecting the key identical to the key specific to each receiver whose associated user wants to communicate with the at least one of the first transmitter and the second transmitter.
12. The optical communications network according to claim 11, wherein the propagation of the encoded light signal transmitted by each of the first and second transmitters is carried out in free propagation to the receivers.
13. The optical communications network according to claim 12, wherein the light transmission means of each of the first and second transmitters comprise a flash lamp whose optical power is modulated according to the encoded light signal.
14. A reconfigurable, multi-user optical communications network, comprising:
a first transmitter configured to generate first a brightness-modulated light signal from a first digital data signal,
a second transmitter configured to generate a second brightness-modulated light signal from a second digital data signal,
an optical coupler configured to combine and transport the first and second light signals to at least one receiver, wherein
each of the first and second transmitters comprise,
a first pseudo-random code generator configured to transmit a code generated from a predetermined software key,
an encoder configured to encode the data signal with said code and configured to output an encoded digital signal,
a light emitter, comprising a multimode light source whose optical power is directly modulated by a signal consisting of the encoded digital signal, to generate a corresponding encoded light signal, and wherein
the at least one receiver comprises,
an optical detector configured to generate an analog electrical signal characteristic of the sum of the first and second light signals transmitted,
a second pseudo-random code generator, and
a decoder configured to decode said analog signal by using a code transmitted by said first pseudo-random code generator with the predetermined software key of either the first or second transmitter and configured to recreate at least one of the first and second data signals.