Abstract: An object in a digital image is automatically recognized by obtaining the digital image, generating a first vector representation of a plurality of pixels of the obtained digital image, randomizing the first vector representation to generate a first generalized vector representation of the digital image that is substantially an order of magnitude smaller than the first vector representation, and searching a repository of hierarchically generalized vector representations of previously recognized digital images, to determine whether the first generalized vector representation of the digital image matches one or more of the previously recognized digital images, within a predetermined threshold value, based on determining a focus area within the obtained digital image, using associative memory.

Abstract: An object in a digital image is automatically recognized by obtaining the digital image, generating a first vector representation of a plurality of pixels of the obtained digital image, randomizing the first vector representation to generate a first generalized vector representation of the digital image that is substantially an order of magnitude smaller than the first vector representation, and searching a repository of hierarchically generalized vector representations of previously recognized digital images, to determine whether the first generalized vector representation of the digital image matches one or more of the previously recognized digital images, within a predetermined threshold value, based on determining a focus area within the obtained digital image, using associative memory.

Abstract: A computing device for expert case-based natural language learning includes a blackboard database, a top level mapper, and a bottom level case-based inference engine, and a bottom level translator. The blackboard database is configured to store context information corresponding to case semantics associated with natural language sentential forms. The case semantics include situation semantics and action semantics. The top level mapper is configured to query the blackboard database for the context information, map the situation semantics to the action semantics using the context information to form new case semantics, and store the new case semantics in a bottom level case database. The bottom level case-based inference engine is configured to match an input natural language sentential form to a matching case semantic stored in the bottom level case database. The bottom level translator is configured to translate the matching case semantic into natural language sentential form.

Abstract: A method for protecting a software system against cyber attacks comprising the following steps: subdividing the software system into components, wherein each component maps a set of input vectors to a non-deterministic set of stochastic output vectors; saving the components in a database; testing the software system by associating each component in the database with at least two mutually random input/output (I/O) test vector pairs that serve to map a random input vector to correct a non deterministic output vector; storing I/O test vectors with their respective component, wherein components in satisfaction of their I/O test vectors are valid by definition; and expunging components having invalid output vectors.

Abstract: A method for protecting a software system against cyber attacks comprising the following steps: subdividing the software system into components, wherein each component maps a set of input vectors to a non-deterministic set of stochastic output vectors; saving the components in a database; testing the software system by associating each component in the database with at least two mutually random input/output (I/O) test vector pairs that serve to map a random input vector to correct a non deterministic output vector; storing I/O test vectors with their respective component, wherein components in satisfaction of their I/O test vectors are valid by definition; and expunging components having invalid output vectors.

Abstract: A method for directed mining of a heterogeneous dataset with a computer comprising: populating a rule base with known rules, wherein each rule has a context and a situation; populating a case base with known cases, wherein each case has a context and a situation, and wherein the case base is partitioned from the rule base; ascribing a natural language semantics to predicates of the known cases and rules; randomly transforming the known rules and the known cases to form new rules by extracting a maximum number of common predicates; segmenting the rules and the cases on the basis of shared predicates without making distinction between context and situation predicates; abducing new knowledge from the dataset by fuzzily matching the context of a new rule to a situation the new rule does not cover; and issuing a query to a user to supply missing predicates of the fuzzy match.

Abstract: A method for directed mining of a heterogeneous dataset with a computer comprising: populating a rule base with known rules, wherein each rule has a context and a situation; populating a case base with known cases, wherein each case has a context and a situation, and wherein the case base is partitioned from the rule base; ascribing a natural language semantics to predicates of the known cases and rules; randomly transforming the known rules and the known cases to form new rules by extracting a maximum number of common predicates; segmenting the rules and the cases on the basis of shared predicates without making distinction between context and situation predicates; abducing new knowledge from the dataset by fuzzily matching the context of a new rule to a situation the new rule does not cover; and issuing a query to a user to supply missing predicates of the fuzzy match.

Abstract: A method for machine-learning by analogy, comprising: providing a network of component computer systems, each component having sequential layers, each layer having parallel segments, each segment comprises a processor and a memory store, each memory store is configured to store a domain-specific case base, which is defined by a situation-action pair; stochastically transforming cases in each case base through automatic deterministic generalization and analogy when the corresponding processor is in a dream mode to create transformed cases; providing a user-entered situation; modifying the user-entered situation by expanding its contextual mnemonics; searching each case base for cases and transformed cases that include contextual subsets of the modified, user-entered situation; for a given case base, mapping the modified, user-entered situation to a matched action within the given case base; creating a new case comprising the user-entered situation and the matched action.

Abstract: A new production generation method comprising: a.) storing a list of productions on a memory store on a computer, wherein each production comprises a context and a consequent, each context and consequent comprising at least one feature; b.) searching the production list for productions with contexts that match a user-provided context; and c.) if no context-matching production is found, i.) randomly selecting a sub-set of features from the user-provided context, ii.) selecting from the production list by uniform chance a first production with a context that matches the selected feature sub-set, iii.) substituting the feature or feature sub-set with the consequent of the first production to create a first feature set, iv.) replacing features in the first feature set as specified by predefined rules to create a new feature set, and v.) displaying a new production consisting of the user-provided context and the new feature set.

Abstract: A method for reaching conclusions from stratified knowledge statements (SKSs) comprising: storing a list of SKSs in at least one memory store on a computer, wherein the list of SKSs is composed of cases, generalizations, and analogs, wherein cases are the most valid, and analogs are the least valid of the SKSs; creating a new generalization based on one of the SKSs; redefining a user-validated generalization as a new case; moving the validated new generalization to the logical head of the case list; expunging the new generalization from the list if the new generalization contradicts an existing case creating with the computer a new analog based on three of the SKSs; redefining a user-validated analog as a new case; moving the new case to the logical head of the case list; and expunging the new analog from the list if the new analog contradicts an existing case, or generalization.

Abstract: A graphical user interface (GUI) is used for rendering semantic concepts and provides menu selections that are rendered in accordance with the semantic concepts. The display includes fields for rendering logical associations in multiple display fields. A first field is provided for a semantic concept, and a second field is provided for receiving a consequent of the semantic concept. First and second menus are used to display the semantic concept in a hierarchical display of a set of semantic concepts and to display consequents corresponding to the semantic concepts. A third field for display a semantic concept and a fourth field displays a consequent of the semantic concept. If a user makes a selection from one of the first and second menus, the GUI renders displays in the third and fourth fields in accordance with the selection.

Abstract: A KASER (Knowledge Amplification by Structured Expert Randomization) engine reaches conclusions in a semantic format or which take the form of a series of rules. The conclusions are parsed into an array structure having a hierarchical order of validity. A set of inserted rules is received as an initial rules array and are configured so that an antecedent comprises a non-empty, sorted set and a consequent comprises a non-empty sequence. A hierarchy of validity for the rules is determined and the rules are sorted according to the hierarchy. At least one rule set which optimizes the selection of rules is determined, with the rule set meeting predetermined validity requirements, and the rules are ordered in an order of validity, such as an order based on a maximal length antecedent set as a highest rank.

Abstract: A method includes the steps of comparing a first body of text with a user-created summary of the first body of text, creating rules based on the comparison of the first body of text with the user-created summary of the first body of text, selecting one or more summary rules for generating a computer-created summary of a second body of text, and applying the selected summary rules to the second body of text to generate a computer-created summary of the second body of text. The first body of text may be a user-corrected summary of a computer-created summary of the first body of text. The rules may be selected based on previous use, frequency of use, context of the body of text, or most-specific applicability. The rules may be iteratively applied to generate a summary. A method is also provided for generating a heading for a summary of text.

Abstract: A graphical user interface for a semantic normalizer of natural language comprising: a link to a preference menu, in which a user may set the semantic normalizer to operate in a predictive or learning mode; an input textbox disposed to display user-typed text in a first font color and user-spoken text in a second font color; a semantic echo textbox disposed to display semantically normalized text in a third font color, wherein the first, second, and third font colors are different from each other; graphical buttons that are only enabled when the semantic normalizer is in learning mode, wherein the graphical buttons may be selected by the user after the user has been prompted by the semantic normalizer to verify the accuracy of the semantically normalized text, the graphical buttons comprising a “Yes” button, a “No” button, and an “Enter Correction” button.

Abstract: In various embodiments, geodesic mining systems and methods are disclosed. For example, a method for forming and adapting a geodesic adaptive network may include embedding a set of rules into a two-dimensional adaptive network having N-rows and M-columns with rule independent variables embedded in a first column of the two-dimensional adaptive network and rule dependent variables embedded in the last column of the two-dimensional adaptive network, where N and M are positive integers greater than two, and repetitively selecting a pair of rows of the two-dimensional adaptive network having common dependent attributes using a random process, then adapting the two-dimensional adaptive network based on the selected pair of rows.

Abstract: A method for object recognition includes generating a set of rules, using multiple systems to recognize a target object, applying the set of rules to a set of responses to determine an output, and displaying the output to a user. Each rule contains predicates and a consequent, each predicate comprising a rule token identifier and a rule probability of recognition. The rule token identifiers are generated from multiple systems. Each rule token identifier represents a system recognized object. Each rule is derived by associating a range of rule probabilities of recognition for one or more rule token identifiers to a known object. The range of rule probabilities of recognition is determined by at least one system and by combining multiple rule probabilities of recognition. Each system produces a response having a response token identifier and a response probability of recognition. Responses are combined to form the sets of responses.

Abstract: A method for hard real-time adaptive wave modulation, comprising: populating a case base with waveform pairs, each waveform pair comprising a primary waveform, and a secondary waveform, wherein the primary waveform induces the secondary waveform; synthesizing a known target waveform s(t) by selecting waveform pairs from the case base such that the expression ?s(t)?g?2 is minimized, where g is the secondary waveform; and mutating the primary waveforms for each waveform pair based on a normal distribution until ?s(t)?g??2<?s(t)?g?2 where g? is a new secondary waveform induced by the mutated primary waveform.