System implemented by a processor controlled machine for inductive determination of pattern probabilities of logical connectors
View Patent ↗The invention relates to a system for the inductive determination of pattern probabilities of 16 dyadic logical connectors. Said system is especially useful for technically predicting human behavior, e.g. inclusion fallacies, and is therefore useful for controlling of technical systems and for data mining, search engines or human-machine interfaces. The system can be implemented in software with a processor controlled machine.
1. A system for the inductive determination of pattern probabilities of all 16 dyadic logical connectors comprising:
a processor controlled machine configured to implement:
at least one input unit for frequency data, prior probabilities of connectors and noise levels, and model parameters;
a central representation unit for representing a field of specific ideal probability tables based on specific model assumptions;
a first determination unit for calculation the data probability given this field of probability tables via multinomial distributions for each table;
a second determination unit for a Bayesian update process for each probability table; and
an output unit for providing pattern-sensitive hypothesis probabilities for connectors at a particular noise level and aggregated probabilities for noise levels or connectors.
2. The system according to claim 1 , wherein the central representation unit is provided for representing a field comprising logical connectors and related noise levels of specific ideal probability tables.
3. The system according to claim 1 , wherein the first determination unit is provided to determine separately the likelihood of the frequency data entered in to the input unit for each connector together with one given noise levels.
4. The system according to claim 1 , further comprising a weighting unit provided to weight the probabilities of the data given the noise-level-connector combinations in a way which monotonically decreases with the inverse of their noise value.
5. The system according to claim 4 , wherein the weighting unit is provided to normalize the weighted noise level probabilities.
6. The system according to claim 1 , wherein the second determination unit is provided to update the probability for noise levels and connector patterns separately for each combination of connector and noise level.
7. The system according to claim 1 , wherein the output unit is provided to update the probability for noise levels and connector patterns by making use of the aggregated probabilities for each noise level and the probabilities for each connector, by recombining both.
8. The system according to claim 1 , wherein the output unit is provided to determine separately the most probable connector pattern or a ranking of connector patterns for a given table of frequency data or the most probable noise level or a ranking of noise levels.
9. The system according to claim 1 , wherein the system is configured for technically predicting human behavior.
10. The system according to claim 1 , further comprising a data mining unit, a search engine unit or a human-machine interface.
11. A method for inductive determination of pattern probabilities of all 16 dyadic logical connectors comprising the steps of:
a) inputting frequency data, prior probabilities of connectors and noise levels, and model parameters;
b) representing a field of specific ideal probability tables based on specific model assumptions;
c) simultaneously calculating the data probability given said ideal probability tables via multinomial distributions;
d) updating of each of the probability tables by use of a Bayesian logic;
e) outputting pattern-sensitive hypothesis probabilities for connectors at a particular noise level and aggregated probabilities for noise levels of connectors.