Abstract: A method involves using a predictor rule base and a corrector rule base to map an acquired context to a rule when the context may not be completely covered by a rule. All involved rules may be non-deterministic. Non-deterministic predictor consequents provide the context for the corrector rule base, while the associated consequent is the correct action to be taken. A minimal of predicates is deleted from every rule in the predictor or corrector base to obtain a contextual covering of at least one rule in the predictor or corrector base. Each produced action may be associated with a possibility metric, which is inversely proportional to the maximum percentage of deleted antecedent predicates produced in matching any single rule in the predictor and corrector bases. The generalized matching capability allows systems to be arranged in a networked system of systems where each system is a domain-specific knowledgebase segment.

Abstract: A method involves providing a parent schema, generating a plurality of distinct and domain-specific instantiations of the parent schema, user-validating at least two of the instantiations of the parent schema, and creating a symmetric schema by combining the user-validated instantiations of the parent schema. The parent schema and the instantiations of the parent schema may be user-provided or iteratively generated. The parent schema contains more than one tractable search spaces. Prior to user-validating the instantiations of the parent schema, they may be optimized using domain-specific knowledge. Additionally, one or more existing Boolean features in one or more cases of a case-based reasoning system may be replaced with the instantiations of the parent schema and non-zero weights for the cases may evolve that include the instantiations of the parent schema. The method may also include generating instantiations of the created symmetric schema.

Abstract: The invention is a method and system updating the automated responses of an autonomous system using sensor data from heterogeneous sources. An array of cases representing known situations are stored as data structures in a non-transitory memory. Each case in the array of cases is associated with an action to create a database of identifiable situation-action pairs. The system determines an acceptable range of correctness of partial matches of sensed data for new cases to the data properties of known cases and creates and overwrites now situation-action pairs in a process of autonomous learning of new responses.

Abstract: A system and method include comparing a context to cases stored in a case base, where the cases include Boolean and non-Boolean independent weight variables and a domain-specific dependency variable. The case and context independent weight variables are normalized and a normalized weight vector is determined for the case base. A match between the received context and each case of the case base is determined using the normalized context and case variables and the normalized weight vector. A skew value is determined for each category of domain specific dependency variables and the category of domain specific dependency variables having the minimal skew value is selected. The dependency variable associated with the selected category is then displayed to a user.

Abstract: A method includes comparing a user-specified context having natural language contextual antecedents to cases stored in a case base. Each stored case includes case antecedents and case consequents. A matching case exists and is selected if the case antecedents exactly match the contextual antecedents. If no match exists, a best-matching case is determined and selected. The best-matching case may be determined by generalizing the situational part of a rule and comparing the user-specified context to the stored (generalized) cases. The best-matching case is the case having the highest ratio of matching generalized case antecedents to contextual antecedents and having a matching error ratio that does not exceed an error ratio threshold. The case consequents of the selected matching case or best matching case are then displayed to a user, with case base adjustment performed based upon feedback provided by the user in response to the displayed case consequents.

Abstract: Systems and methods are described for increasing the burn efficiency of fuel in a jet engine by charging the intake air molecules and inversely charging the fuel molecules, while reducing the fuel droplet size using a fuel-atomizing transducer in the air/fuel path. Generally speaking, an apparatus (and method) for increasing fuel burning efficiency in a jet engine is described, comprising: an intake air high voltage charger; a fuel injector downstream of the intake air charger; a fuel atomizer formed from an acutely angled transducer in a flow path of the fuel, wherein the atomized fuel is charged to an opposite voltage to that of the intake air; and an ignitor for igniting the atomized charged fuel and charged intake air for combustion.

Abstract: A relational database is used to determine a possibility of events, such as terrorist threats. A database is populated or updated in an automated fashion by using appropriate sensor sources. Whenever a field is augmented or updated, an event is defined. Events trigger intelligent data collection agents using a push technology. A list of events is defined over a relative time interval. A selection of lists of events is made in response to events. The defined database is updated according to an iterative architecture for the defined database.

Abstract: Systems and methods are described for remotely controlled vehicles that hierarchically transform sensor-detected information into levels of increasing fidelity for highly efficient machine recognition of attributes and features of detected objects/events. Fusion of the feature space representations of the various levels allows correlation with the operator's attention to the respective objects for automated processing/controlling of the unmanned vehicle. High efficiencies in object/event recognition are attained with reduced memory and processor requirements, enabling near autonomous or fully autonomous operation of the vehicle.

Abstract: Anticipatory logistics is used to predict observable events and respond to the predictions of the observable events in the control of automated equipment that perform highly repetitive functions such as elevator cars. A set of table entries is obtained, and the table entries are metricized and stored as cell entries. All cell entries are normalized. Ten weighted values to the cell entries are initialized. An algorithmically defined subset of weighted values is normalized and an instruction is selected based on the computed dependency using an algorithm incorporating uniform chance selection for exploratory optimization, such as the Mersenne Twister algorithm. Here, the search space is delimited by careful selection of the salient variables as well as by the algorithm itself, which only relies on chance to find truly novel solutions as time (and space) permit.

Abstract: A method includes receiving a user-specified context comprising one or more natural language contextual antecedents. Then, for each contextual antecedent, a modified contextual antecedent is created by converting each contextual antecedent to a sequence of integers using a word base. Each modified contextual antecedent is compared to each of a plurality of cases stored in a case base, where each case includes one or more case antecedents and one or more case consequents. The case antecedents and case consequents are stored in the case base as sequences of integers representing the respective case antecedents and case consequents. The case having the case antecedents that best match the contextual antecedents is then selected and the case consequents of the selected case are displayed to a user. The user then provides feedback regarding the displayed case consequents. The feedback may be integrated into the contextual antecedent for a new search of the case base.

Abstract: Sea characteristics are determined by using a buoy with multiple tilt switches configured to sense at plural tilt angles of the buoy. The tilt switches provide binary outputs of tilt data corresponding to the tilt angles. The binary outputs are received by a processor and are compared with sample patterns of tilt data. The comparison is used to provide an output corresponding to sea conditions.

Abstract: A method includes selecting, from a data table containing a plurality of weighted independent variable columns and rows, where each of the columns represents an independent input variable feature and each row represents a set of stored independent input variables and a corresponding stored dependent variable, a row that best matches received sensor independent input variable data from at least one vehicle sensor, causing a vehicle accelerator to gradually accelerate a vehicle to a vehicle acceleration level corresponding to the stored dependent variable of the selected best match row, and modifying a weight of the weighted independent variable columns based upon received user feedback to the vehicle acceleration level. The weights may then be normalized and the received sensor independent input variable data may be stored within a new row of the data table if the weight modification produces a net overall improvement in the data table dependency mapping.

Abstract: A system and method for inductive and analogical reasoning provide capabilities for generalization, creative reasoning, possibilistic analysis, massive parallelism, and distributed processing. The system and method involve using matched cases/generalizations/analog inferences from a case base or generalization/analog base to provide output inferences based upon a user-supplied context. A user then provides feedback to indicate that the output inference is either correct or incorrect. If the user indicates that the output is incorrect, new rules may be computer-generated during runtime by performing, for example, an analogical contextual transformation. If the user indicates that the output is correct, the matched case/generalization/analog may be moved to the head of its respective base.

Abstract: Systems and methods for extracting information from large databases include the initial step of searching a database using a query. The search results are then normalized by formatting the search results in a standard text format. The normalized search results are then crawled according to a predetermined algorithm to yield candidate information, which includes establishing sequences of keywords and generating hash tables that correspond to the keyword sequences. For each paragraph of candidate information, a defined possibility for each hash table entry is calculated. If the possibility is above a predetermined value, the candidate information paragraph is extracted from candidate information and presented to the user. Only the paragraphs that have a possibility that is above the predetermined threshold are displayed and presented to the user. Displayed search results rated as useful by the user are added to the query and to the keyword sequences to accomplish iterative, context-directed searches.

Abstract: A knowledge-based decision support system that allows for communication and learning to occur using natural language is presented. The system has a capability to automatically extract features from the natural language using symmetric reductions and random search. The iterative generalization of the rule base and checking of the resultant base against a case base from which the generalizations are induced is also provided. The decision support system can be used to search semi-structured databases and automatically learns new knowledge and search control knowledge where it is most needed based on the pattern of previous rule firings.

Abstract: In various embodiments, evolutionary expert systems and methods are disclosed. For example, a method for evolving a rule base of an expert system includes creating a set of meta-rules from a set of first rules associated with the expert system, creating a set of one or more generalized virtual rule candidates based on the set of first rules and the set of meta-rules, filtering the set of generalized virtual rule candidates to remove generalized virtual rule candidates that conflict with at least one rule of the set of first rules to form a set of virtual rules, and incorporating at least one virtual rule of the set of virtual rules into the set of first rules to evolve the first set of rules.

Abstract: A laser interferometer, such as a dual path Michelson interferometer, is used to generate fringe patterns resulting from one of a plurality of optical paths passing through a sample gas. An artificial neural network, such as, for example, a KASER neural network, is used to recognize patterns in the fringe interference patterns corresponding to known target gases.

Abstract: A distributed biohazard surveillance system including a plurality of robust miniaturized remote monitoring stations for the detection, localized analysis and reporting of a broad range of biohazards. The remote monitoring station may be adapted to identify many different biological particles and is not limited to particular predetermined biohazard profiles. It is centrally and dynamically reconfigurable and can be adapted to operate unattended in a remote location. The distributed system may be used to locate and report unsuspected sources of biohazards and to monitor the localized effects in real-time cooperation with a centralized data processing facility.

Abstract: A Knowledge Amplifier with Structured Expert Randomization (KASER) that exploits a structured expert randomization principle. One KASER embodiment allows the user to supply declarative knowledge in the form of a semantic tree using single inheritance. Another KASER embodiment includes means for automatically inducing this semantic tree, such as, for example, means for performing randomization and set operations on the property trees that are acquired by way of, for example, database query and user-interaction.

Abstract: A distributed biohazard surveillance system including a plurality of robust miniaturized remote monitoring stations for the detection, localized analysis and reporting of a broad range of biohazards. The remote monitoring station may be adapted to identify many different biological particles and is not limited to particular predetermined biohazard profiles. It is centrally and dynamically reconfigurable and can be adapted to operate unattended in a remote location. The distributed system may be used to locate and report unsuspected sources of biohazards and to monitor the localized effects in real-time cooperation with a centralized data processing facility.