WO2016098366A1 - Dispositif pour construire un système de connaissances autonome - Google Patents

Dispositif pour construire un système de connaissances autonome Download PDF

Info

Publication number
WO2016098366A1
WO2016098366A1 PCT/JP2015/062160 JP2015062160W WO2016098366A1 WO 2016098366 A1 WO2016098366 A1 WO 2016098366A1 JP 2015062160 W JP2015062160 W JP 2015062160W WO 2016098366 A1 WO2016098366 A1 WO 2016098366A1
Authority
WO
WIPO (PCT)
Prior art keywords
pattern
information
patterns
knowledge
connection
Prior art date
Application number
PCT/JP2015/062160
Other languages
English (en)
Japanese (ja)
Inventor
洋彰 宮崎
Original Assignee
洋彰 宮崎
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 洋彰 宮崎 filed Critical 洋彰 宮崎
Publication of WO2016098366A1 publication Critical patent/WO2016098366A1/fr

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N5/00—Computing arrangements using knowledge-based models
    • G06N5/02—Knowledge representation; Symbolic representation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N99/00—Subject matter not provided for in other groups of this subclass

Definitions

  • the present invention relates to artificial intelligence that builds logic and knowledge automatically from input information sequentially.
  • the machine processes input information, etc.
  • it is realized by incorporating and executing a program created in advance in a program language in a computer mounted on the machine.
  • a program is created so that the corresponding operation is executed when a preset condition is detected. If the condition detection and the corresponding operation are not appropriate, modify the program installed on the computer.
  • the knowledge built inside There is no conventional machine that adds new knowledge to a system or updates knowledge, and autonomously carries out knowledge expansion and knowledge organization.
  • the machine processes input information, etc.
  • it is realized by incorporating and executing a program created in advance in a program language in a computer mounted on the machine.
  • a program is created so that the corresponding operation is executed when a preset condition is detected. If the condition detection and the corresponding operation are not appropriate, modify the program installed on the computer.
  • the machine processes input information, etc.
  • it is realized by incorporating and executing a program created in advance in a program language in a computer mounted on the machine.
  • a program is created so that the corresponding operation is executed when a preset condition is detected. If the condition detection and the corresponding operation are not appropriate, modify the program installed on the computer. Convert the input information into a pattern and record it one after another, and analyze the common points and differences of the thoughts by inputting the books, sentences, conversations and utterances of multiple persons, etc. and compare the thinking systems There is no conventional machine. [Prior Art] (Corresponding to Claim 4)
  • the machine processes input information, etc.
  • it is realized by incorporating and executing a program created in advance in a program language in a computer mounted on the machine.
  • a program is created so that the corresponding operation is executed when a preset condition is detected. If the condition detection and the corresponding operation are not appropriate, modify the program installed on the computer. Convert the input information into a pattern and record it one after another, analyze the superiority of the logical thinking by inputting the books, sentences, conversations and remarks of multiple persons, etc., and based on the better logical thinking There is no machine that builds knowledge in the past. [Prior art] (corresponding to claim 5)
  • the machine processes input information, etc.
  • it is realized by incorporating and executing a program created in advance in a program language in a computer mounted on the machine.
  • a program is created so that the corresponding operation is executed when a preset condition is detected. If the condition detection and the corresponding operation are not appropriate, modify the program installed on the computer.
  • the knowledge built inside There is no conventional machine that performs addition of new knowledge, update of knowledge and generalization of knowledge to a system, and autonomously performs knowledge expansion and systematization of knowledge. [Prior art] (corresponding to claim 6)
  • the machine processes input information, etc.
  • it is realized by incorporating and executing a program created in advance in a program language in a computer mounted on the machine.
  • a program is created so that the corresponding operation is executed when a preset condition is detected. If the condition detection and the corresponding operation are not appropriate, modify the program installed on the computer.
  • the knowledge built inside There is no conventional machine that carries out the expansion of knowledge and the systematization of knowledge by adding new knowledge, updating knowledge and generalizing knowledge to a system, and autonomously performing an appropriate action on input information.
  • Human thinking is expressed by language, and we will convert the information expressed by this language into what is called a pattern.
  • individual thoughts of human beings can be expressed as individual patterns, and changes in human thinking can be regarded as changes from patterns to patterns.
  • the pattern is not limited to merely expressing a language, and for example, image information can be expressed, and information processing can also be performed.
  • image information can also be expressed as a pattern.
  • a process of generating an answer to a question can also be expressed as a pattern.
  • a control signal for driving a driving device can also be expressed as a pattern.
  • human thinking can be expressed as a change from pattern to pattern, but the change from pattern to pattern is not chaotic.
  • patterns and patterns Expressing information such as speech, speech, books and documents, which are human thinking activities, as individual patterns, and carefully observing the appearance of each pattern, there is a relationship between patterns and patterns. I understand.
  • the document is a set of individual sentences, but a certain sentence appears based on the sentence and context before the sentence appears, so words or meanings In terms of levels, the words and meanings contained in the sentence have a strong correlation with the words and meanings of the sentences that appeared earlier. Identify individual information as a pattern, convert the pattern to an equivalent pattern at the level of meaning, record the pattern and the appearance of the pattern, and extract the relationship between the strongly related pattern and pattern when accumulated Can.
  • a logical relationship between this pattern and the pattern (cause and result, action and its reason, action and its means, problem and solution, question and answer, certain item and its detailed explanation, certain item and its brief explanation , Concreteness and abstraction, an item and an example, an item and its features, an item and its state, an explanation and another explanation, a meaning and equivalent meaning, a story and a continuation of a story, etc.)
  • the identification of logical relationships is first carried out by humans.
  • this knowledge system construction machine is made to learn the identification of the logical relation which a person carries out.
  • the present knowledge system builder can autonomously output patterns and pattern identification results. If there is an error in the output result, correction is made such as strengthening the connection with the logical relationship to be originally identified and weakening or prohibiting the erroneously output logical relationship.
  • the knowledge construction machine can autonomously generate information and information connection relation and logical relation. The information input to this knowledge system builder can be collated with the knowledge accumulated in the past, and since it is possible to analyze and accumulate the relationship, it is constructed as a knowledge system rather than accumulating information individually.
  • the knowledge system builder converts information into patterns.
  • the converted patterns are sequentially recorded in the recording module of the pattern recorder.
  • the pattern registration device collates the input information with the recording module of the pattern recorder, confirms whether the same pattern as the input information is recorded, and if it is not recorded, records and excites the recording module as a new pattern.
  • the excitation refers to a state in which the recording module is activated by collating information or the like. If the same pattern as the input information is already recorded, the corresponding recording module is excited.
  • the history of the activated recording module is sequentially recorded on the pattern recorder.
  • connection information is transferred to the connection information recording unit recording the information.
  • the method of increasing the coupling coefficient which indicates the strength of the transfer and connection of the history of the associated recording module, detects the correlation with the already recorded history information with regard to the individual excitation history as well as a simple superposition. Include methods to increase the appropriate coupling factor. Although this method is somewhat complicated in processing, an increase in the coupling coefficient is caused by the history of the recording module which is not directly related to the excitation of the recording module included in the history information, and it is possible to avoid making an erroneous judgment.
  • the sentence S1 is a cause
  • the sentence S2 is a result of the sentence S1
  • the sentence S3 is a task
  • the sentence S4 is a solution of the sentence S3.
  • connection relationships between patterns described above have been generated.
  • connection information recording unit of the pattern S2 has a connection relationship with the pattern S1
  • the connection information recording unit of the pattern S4 has a connection relationship with the pattern S3.
  • the pattern S2 is excited, it is detected that there is a connection relation with the pattern S1, but a human being excites the pattern R2 indicating the result in the logical relation (the result is learned And the pattern S1 and the pattern S2 are recorded as a combination excitation pattern in the connection information recording portion of the pattern R2.
  • the pattern S4 when the pattern S4 is excited, it is also detected that there is a connection relation with the pattern S3, but a human being excites the pattern R4 showing the solution in the logical relation (learning that it is a solution And the pattern S4 and the pattern S4 are recorded as a combination excitation pattern in the connection information recording portion of the pattern R4.
  • the connection information recording portion of the pattern R4 when combinations of corresponding logical relationships are recorded for each connection relationship between patterns, when a specific combination of patterns appears in the connection information recording unit of the recording module indicating each logical relationship, the relevant logical relationship The combination of connection information is generated so that the recording module corresponding to ⁇ ⁇ is excited.
  • the present knowledge system builder can autonomously output patterns and pattern identification results.
  • the knowledge construction machine can autonomously generate information and information connection relation and logical relation.
  • each input information is compared with a pattern (corresponding to knowledge) already recorded, and a logical connection relation (corresponding to a knowledge system) constructed by the collated patterns.
  • the logical validity, novelty, and validity of the input information can also be evaluated by comparing with the logical relationship of the input information.
  • Allowing a knowledge system to grow as a knowledge system rather than recording it in a single state by autonomously incorporating new and valid information into the knowledge system of the knowledge construction machine while evaluating the validity of input information Become an artificial intelligence device that can Next, a method of evaluating the value of input information in the present knowledge construction machine will be described. It is possible to identify what kind of field information is related to input information from words included in the input information. This can be implemented in the same way as the learning of logical relationships described above. Perform semantic analysis on input information and excite patterns that are semantically equivalent to input information. Next, the human judges which field the input information is, and excites a pattern indicating the corresponding field.
  • connection information recording part of the pattern indicating the field indicates the input information and the equivalent meaning Is recorded.
  • a history of excitation of a pattern related to the field is accumulated in the connection information recording unit of the pattern indicating the field of the present knowledge constructing machine. That is, the connection corresponding to the word etc. which shows the said field is strengthened.
  • this knowledge construction machine autonomously performs the field identification, and if there is an error in the output result, the connection with the field to be originally identified is strengthened, and the field is erroneously output Make corrections such as weakening or prohibiting bonding with
  • this knowledge construction machine can identify the field, theme, etc. of the input information.
  • the field, theme, etc. of the input information is identified, it is compared with the already recorded knowledge of the present knowledge construction machine to check whether there is related information. This is because, as described above, when a related pattern is detected, a pattern having a logical relationship with that pattern can also be searched, so that it may be judged whether the input information includes new information. Can.
  • the input information is valid. This can be determined by checking whether the pattern of the semantic level indicated by the input information and the pattern of the semantic level indicated by the related information already recorded excite the opposite pattern.
  • the value of the input information can be evaluated by evaluating the field of the input information, the consistency with the related knowledge, the presence or absence of novelty, the reliability, and the like. If it is judged that the input information has a value, incorporation into the knowledge system is promoted by recording it in the relevant part of the knowledge system of this knowledge construction machine. [Means for Solving the Problems] (Corresponding to Claim 2)
  • a logical relationship between this pattern and the pattern (cause and result, action and its reason, action and its means, problem and solution, question and answer, certain item and its detailed explanation, certain item and its brief explanation , Concreteness and abstraction, an item and an example, an item and its features, an item and its state, an explanation and another explanation, a meaning and equivalent meaning, a story and a continuation of a story, etc.)
  • the identification of logical relationships is first carried out by humans.
  • this knowledge system construction machine is made to learn the identification of the logical relation which a person carries out.
  • the present knowledge system builder can autonomously output patterns and pattern identification results. If there is an error in the output result, correction is made such as strengthening the connection with the logical relationship to be originally identified and weakening or prohibiting the erroneously output logical relationship.
  • the knowledge construction machine can autonomously generate information and information connection relation and logical relation. The information input to this knowledge system builder can be collated with the knowledge accumulated in the past, and since it is possible to analyze and accumulate the relationship, it is constructed as a knowledge system rather than accumulating information individually.
  • connection patterns of the converted patterns are generated.
  • time-series input time-series connection relations are generated between the preceding and succeeding patterns.
  • a logical connection relationship is also generated.
  • Books, sentences, conversations and remarks etc. of an individual are recorded in the recording area classified according to the field and the theme.
  • the pattern to be recorded is collated with the already recorded pattern, and if collated, the connection relation is generated in consideration of the logical relation with the already recorded pattern. If not collated, it is recorded as a new pattern in a recording area corresponding to a predetermined field and theme. In this way, when recording in consideration of the connection relationship between patterns, it becomes possible to record as a thought system of a certain individual instead of being recorded in a disjointed state. [Means for Solving the Problems] (Corresponding to Claim 3)
  • a logical relationship between this pattern and the pattern (cause and result, action and its reason, action and its means, problem and solution, question and answer, certain item and its detailed explanation, certain item and its brief explanation , Concreteness and abstraction, an item and an example, an item and its features, an item and its state, an explanation and another explanation, a meaning and equivalent meaning, a story and a continuation of a story, etc.)
  • the identification of logical relationships is first carried out by humans.
  • this knowledge system construction machine is made to learn the identification of the logical relation which a person carries out.
  • the present knowledge system builder can autonomously output patterns and pattern identification results. If there is an error in the output result, correction is made such as strengthening the connection with the logical relationship to be originally identified and weakening or prohibiting the erroneously output logical relationship.
  • the knowledge construction machine can autonomously generate information and information connection relation and logical relation. The information input to this knowledge system builder can be collated with the knowledge accumulated in the past, and since it is possible to analyze and accumulate the relationship, it is constructed as a knowledge system rather than accumulating information individually.
  • connection patterns of the converted patterns are generated.
  • time-series input time-series connection relations are generated between the preceding and succeeding patterns.
  • the logical connection relationship is also generated.
  • Books, sentences, conversations and remarks etc. of an individual are recorded in the recording area classified according to the field and the theme.
  • the pattern to be recorded is collated with the already recorded pattern, and if collated, the connection relation is generated in consideration of the logical relation with the already recorded pattern. If not collated, it is recorded as a new pattern in a recording area corresponding to a predetermined field and theme.
  • the patterns of person A and person B are matched against the recording area defined above in the same field, theme, and logical connection.
  • This collation can identify the collation of person A, the collation of person B, and other persons by using a plurality of independent collation lines. If the pattern matched by the person A matches the pattern matched by the person B, it is detected that they are common, and conversely the pattern matched by the person A and the pattern matched by the person B show opposite patterns Excitation makes it possible to detect differences. (Note that if there is a pattern on one side in the same field, theme, or logical connection, but there is no pattern on the other side, it indicates that there is more information in the one where the pattern exists. ) When the same field, theme, logical connection, etc. are analyzed one after another in common, difference, no relation etc. between patterns, what is common in thinking of person A and person B and what is different It is possible to analyze. [Means for Solving the Problems] (Corresponding to Claim 4)
  • connection patterns of the converted patterns are generated.
  • time-series input time-series connection relations are generated between the preceding and succeeding patterns.
  • the logical connection relationship is also generated. Books, sentences, conversations and remarks etc. of an individual are recorded in the recording area classified according to the field and the theme.
  • the pattern to be recorded is collated with the already recorded pattern, and if collated, the connection relation is generated in consideration of the logical relation with the already recorded pattern. If not collated, it is recorded as a new pattern in a recording area corresponding to a predetermined field and theme.
  • the patterns of person A and person B are matched against the recording area defined above in the same field, theme, and logical connection.
  • This collation can identify the collation of person A, the collation of person B, and other persons by using a plurality of independent collation lines. If the pattern matched by the person A matches the pattern matched by the person B, it is detected that they are common, and conversely the pattern matched by the person A and the pattern matched by the person B show opposite patterns Excitation makes it possible to detect differences. (Note that if there is a pattern on one side in the same field, theme, or logical connection, but there is no pattern on the other side, it indicates that there is more information in the one where the pattern exists. ) When the same field, theme, logical connection, etc.
  • identify the information source (whether from a reliable source or not), the type of information (truth, fact, rumor, present state, future talk, imagination, assumption, etc.), etc. deep.
  • the new information is analyzed and if there is something to match with the already recorded information, the new information is evaluated as how it was combined with the already recorded information.
  • A is B. Because it is C. It is assumed that the new information is collated with the already recorded information and the "C" is collated with the already recorded information. At this time, the above sentence mentions that "It is C" as a ground "A is B".
  • a logical relationship between this pattern and the pattern (cause and result, action and its reason, action and its means, problem and solution, question and answer, certain item and its detailed explanation, certain item and its brief explanation , Concreteness and abstraction, an item and an example, an item and its features, an item and its state, an explanation and another explanation, a meaning and equivalent meaning, a story and a continuation of a story, etc.)
  • the identification of logical relationships is first carried out by humans.
  • this knowledge system construction machine is made to learn the identification of the logical relation which a person carries out.
  • the present knowledge system builder can autonomously output patterns and pattern identification results. If there is an error in the output result, correction is made such as strengthening the connection with the logical relationship to be originally identified and weakening or prohibiting the erroneously output logical relationship.
  • the knowledge construction machine can autonomously generate information and information connection relation and logical relation. The information input to this knowledge system builder can be collated with the knowledge accumulated in the past, and since it is possible to analyze and accumulate the relationship, it is constructed as a knowledge system rather than accumulating information individually.
  • (1) Treatment of input information When information is input, collation with already recorded information, position analysis, recording, updating of already recorded information, etc. are performed.
  • the input information is analyzed in terms of fields and themes, and collated with information already recorded. If there is any collation with the information already recorded, the position of the input information is analyzed. The position is evaluated from the logical connection relation with the recorded information, reliability, etc., and if the input information is useful (newness, logical validity, reliability), the corresponding field, theme, logical connection Add the relevant knowledge to Also, if necessary, update or change the priority of already recorded information.
  • These actions can be defined by patterns. Analysis of information and themes and analysis of position is the method described in the previous section.
  • the evaluation of the usefulness of input information can also be performed by the method described in the previous section.
  • the addition of the corresponding field, theme, and logical connection can be implemented by additionally recording the information in the corresponding recording area.
  • (2) Interpretation and response of information Generate instructions, response to requests, responses to questions, and questions to confirm unknown points. Identify whether the input sentence corresponds to an instruction or request.
  • the identification can be carried out by making the present knowledge system builder learn the human identification described in the previous section. Since the sentence includes a term that can identify the command sentence and the request sentence, the command sentence and the request sentence can be identified by detecting this word and phrase. Have you responded to the orders and requests in the past? Is there any problem with the response when I cope? Evaluate etc.
  • the external situation as a pattern and execute processing by creating a connection relationship between the image information detected from a visual sensor, a voice detector, etc., the pattern obtained by converting the voice information, and the pattern corresponding to the corresponding word.
  • the confirmation of the current state and the update of the state are implemented by updating the pattern corresponding to the current state and the state. It can be implemented by recording the output result from the detector or monitor system that detects the current state and the state in the area where the corresponding pattern is recorded.
  • the response to the situation can be defined by the pattern corresponding to the conditional operation.
  • Planning, scheduling, execution of action Planning, execution, evaluation, and execution of action (drive system control) are implemented.
  • Planning, execution, evaluation, and execution of the operation can be defined by a pattern corresponding to the conditional operation.
  • FIG. 1 is a diagram showing the configuration of an autonomous knowledge system builder according to an embodiment of the present invention.
  • reference numeral 1 denotes a pattern converter for converting information into a pattern.
  • Reference numeral 2 denotes a pattern recorder which records a history of patterns, connection relationships between patterns, logical relationships between patterns, and excitations of patterns.
  • Reference numeral 3 denotes a pattern registration unit which registers and changes patterns and connection relationships between patterns by human instruction or autonomously.
  • a pattern controller 4 controls processing of the pattern.
  • Reference numeral 5 is a pattern reverse converter which converts a pattern into information.
  • 6 is a pattern analyzer that analyzes patterns and relationships between patterns.
  • 7 is a pattern evaluator that evaluates the value of information.
  • a pattern converter converts information into patterns.
  • the converted patterns are sequentially recorded in the recording module of the two pattern recorders.
  • the pattern registration unit 3 collates the input information with the recording module of the pattern recorder, confirms whether the same pattern as the input information is recorded, and if it is not recorded, records and excites it as a new pattern in the recording module .
  • the excitation refers to a state in which the recording module is activated by collating information or the like. If the same pattern as the input information is already recorded, the corresponding recording module is excited. The history of the activated recording module is sequentially recorded on the pattern recorder.
  • connection information is transferred to the connection information recording unit recording the information.
  • the method of increasing the coupling coefficient which indicates the strength of the transfer and connection of the history of the associated recording module, detects the correlation with the already recorded history information with regard to the individual excitation history as well as a simple superposition. Include methods to increase the appropriate coupling factor. Although this method is somewhat complicated in processing, an increase in the coupling coefficient is caused by the history of the recording module which is not directly related to the excitation of the recording module included in the history information, and it is possible to avoid making an erroneous judgment.
  • FIG. 7 shows an operation of transferring the history of the excitation pattern considered to be related to the excitation of the module to the connection information recording part of the recording module when the recording module is excited and strengthening the connection with the related pattern.
  • FIG. 8 shows an operation in which the connection with the pattern related to the excitation in the connection information recording unit of the recording module is strengthened by the repeated excitation of the recording module.
  • FIG. 9 shows an operation in which the logical relationship between patterns is recorded in the inter-pattern relationship recording section of the pattern recorder.
  • the sentence S1 is a cause
  • the sentence S2 is a result of the sentence S1
  • the sentence S3 is a task
  • the sentence S4 is a solution of the sentence S3.
  • Pattern S1 (pattern showing sentence S1) ⁇ Circle over (6) ⁇ pattern S2 ⁇ (pattern showing sentence S2) 7 7 ⁇ Pattern S3 ⁇ (pattern showing sentence S3) 8 8 ⁇ Pattern S 4 ⁇ (pattern showing sentence S 4)
  • connection relationships between patterns described above have been generated. Specifically, it is assumed that the connection information recording unit of the pattern S2 has a connection relationship with the pattern S1, and the connection information recording unit of the pattern S4 has a connection relationship with the pattern S3.
  • the pattern S2 when the pattern S2 is excited, it is detected that there is a connection relation with the pattern S1, but a human being excites the pattern R2 indicating the result in the logical relation (the result is learned And the pattern S1 and the pattern S2 are recorded as a combination excitation pattern in the connection information recording portion of the pattern R2.
  • the pattern S4 when the pattern S4 is excited, it is also detected that there is a connection relation with the pattern S3, but a human being excites the pattern R4 showing the solution in the logical relation (learning that it is a solution And the pattern S4 and the pattern S4 are recorded as a combination excitation pattern in the connection information recording portion of the pattern R4.
  • connection information recording unit of the recording module indicating each logical relationship
  • the relevant logical relationship The combination of connection information is generated so that the recording module corresponding to ⁇ ⁇ is excited.
  • the present knowledge system builder can autonomously output patterns and pattern identification results. If there is an error in the output result, correction is made such as strengthening the connection with the logical relationship to be originally identified and weakening or prohibiting the erroneously output logical relationship.
  • the knowledge construction machine can autonomously generate information and information connection relation and logical relation.
  • each input information is compared with a pattern (corresponding to knowledge) already recorded, and a logical connection relation (corresponding to a knowledge system) constructed by the collated patterns.
  • the logical validity, novelty, and validity of the input information can also be evaluated by comparing with the logical relationship of the input information. Allowing a knowledge system to grow as a knowledge system rather than recording it in a single state by autonomously incorporating new and valid information into the knowledge system of the knowledge construction machine while evaluating the validity of input information Become an artificial intelligence device that can Next, a method of evaluating the value of input information in the present knowledge construction machine will be described.
  • FIG. 11 shows an operation of identifying what kind of field information is related to input information from words included in the input information.
  • this knowledge construction machine autonomously performs the field identification, and if there is an error in the output result, the connection with the field to be originally identified is strengthened, and the field is erroneously output Make corrections such as weakening or prohibiting bonding with
  • this knowledge construction machine can identify the field, theme, etc. of the input information.
  • the field, theme, etc. of the input information is identified, it is compared with the already recorded knowledge of the present knowledge construction machine to check whether there is related information. This is because, as described above, when a related pattern is detected, a pattern having a logical relationship with that pattern can also be searched, so that it may be judged whether the input information includes new information. Can. FIG.
  • FIG. 12 shows the operation of analysis of the field and theme of the input information and the comparison with the information already recorded in the same field and theme.
  • fields, themes, and logical relationships are recorded in a state of being clarified.
  • comparisons can be made in the same logic hierarchy as the information already recorded.
  • FIG. 13 shows an operation example for evaluating the relationship between already recorded information and newly input information. Relationships such as commonness, differences, similarities between words can be defined between patterns. Difficult words can also be expressed by simpler word combinations. As described above, when the relationship between words is defined in the recording area, comparison between the input information and the information already recorded can be made at the semantic level.
  • FIG. 14 shows operations that are evaluated as valuable to new information and are incorporated as additional knowledge into the already constructed knowledge system.
  • the value of input information is evaluated by evaluating the field of input information, consistency with related knowledge, existence of novelty, reliability, etc., and if it is judged that there is value, knowledge of this knowledge system builder By additionally recording in relevant parts of the system, autonomous knowledge improvement is promoted.
  • FIG. 2 is a diagram showing the configuration of an autonomous knowledge system builder according to an embodiment of the present invention.
  • reference numeral 1 denotes a pattern converter which converts information into a pattern.
  • Reference numeral 2 denotes a pattern recorder which records a history of patterns, connection relationships between patterns, logical relationships between patterns, and excitations of patterns.
  • Reference numeral 3 denotes a pattern registration unit which registers and changes patterns and connection relationships between patterns by human instruction or autonomously.
  • a pattern controller 4 controls processing of the pattern.
  • Reference numeral 5 is a pattern reverse converter which converts a pattern into information.
  • 6 is a pattern analyzer that analyzes patterns and relationships between patterns.
  • the pattern converter converts information into patterns.
  • the converted patterns are sequentially recorded in the recording module of the two pattern recorders.
  • the pattern registration unit 3 collates the input information with the recording module of the pattern recorder, confirms whether the same pattern as the input information is recorded, and if it is not recorded, records and excites it as a new pattern in the recording module .
  • the excitation refers to a state in which the recording module is activated by collating information or the like. If the same pattern as the input information is already recorded, the corresponding recording module is excited. The history of the activated recording module is sequentially recorded on the pattern recorder.
  • connection information is transferred to the connection information recording unit recording the information.
  • the method of increasing the coupling coefficient which indicates the strength of the transfer and connection of the history of the associated recording module, detects the correlation with the already recorded history information with regard to the individual excitation history as well as a simple superposition. Include methods to increase the appropriate coupling factor. Although this method is somewhat complicated in processing, an increase in the coupling coefficient is caused by the history of the recording module which is not directly related to the excitation of the recording module included in the history information, and it is possible to avoid making an erroneous judgment.
  • the book, sentences, conversations and utterances of a certain individual are input to the present knowledge system construction machine, and connection patterns of the converted patterns are generated. While recording in time series, by detecting the logical relationship between the appearing pattern and the pattern, a logical connection relationship is also generated. Books, sentences, conversations and remarks etc. of an individual are also recorded in the recording area classified by field and theme.
  • FIG. 15 shows an operation of identifying and recording the set personal information in the field, the theme, and the logical relationship.
  • the pattern to be recorded is collated with the already recorded pattern, and if collated, the connection relation is generated in consideration of the logical relation with the already recorded pattern. If not collated, it is recorded as a new pattern in a recording area corresponding to a predetermined field and theme.
  • FIG. 3 is a diagram showing the configuration of an autonomous knowledge system builder according to an embodiment of the present invention.
  • reference numeral 1 denotes a pattern converter for converting information into a pattern.
  • Reference numeral 2 denotes a pattern recorder which records a history of patterns, connection relationships between patterns, logical relationships between patterns, and excitations of patterns.
  • Reference numeral 3 denotes a pattern registration unit which registers and changes patterns and connection relationships between patterns by human instruction or autonomously.
  • a pattern controller 4 controls processing of the pattern.
  • Reference numeral 5 is a pattern reverse converter which converts a pattern into information.
  • 6 is a pattern analyzer that analyzes patterns and relationships between patterns.
  • Reference numeral 8 denotes a pattern comparator which analyzes the common points and differences of connection relations among a plurality of patterns and patterns.
  • the pattern converter converts information into patterns.
  • the converted patterns are sequentially recorded in the recording module of the two pattern recorders.
  • the pattern registration unit 3 collates the input information with the recording module of the pattern recorder, confirms whether the same pattern as the input information is recorded, and if it is not recorded, records and excites it as a new pattern in the recording module .
  • the excitation refers to a state in which the recording module is activated by collating information or the like. If the same pattern as the input information is already recorded, the corresponding recording module is excited.
  • the history of the activated recording module is sequentially recorded on the pattern recorder.
  • connection information is transferred to the connection information recording unit recording the information.
  • the method of increasing the coupling coefficient which indicates the strength of the transfer and connection of the history of the associated recording module, detects the correlation with the already recorded history information with regard to the individual excitation history as well as a simple superposition. Include methods to increase the appropriate coupling factor. Although this method is somewhat complicated in processing, an increase in the coupling coefficient is caused by the history of the recording module which is not directly related to the excitation of the recording module included in the history information, and it is possible to avoid making an erroneous judgment.
  • the book, sentences, conversations and utterances of a certain individual are input to the present knowledge system construction machine, and connection patterns of the converted patterns are generated. While recording in time series, by detecting the logical relationship between the appearing pattern and the pattern, a logical connection relationship is also generated. Books, sentences, conversations and remarks etc. of an individual are also recorded in the recording area classified by field and theme.
  • FIG. 15 shows an operation of identifying and recording the set personal information in the field, the theme, and the logical relationship.
  • the pattern to be recorded is collated with the already recorded pattern, and if collated, the connection relation is generated in consideration of the logical relation with the already recorded pattern. If not collated, it is recorded as a new pattern in a recording area corresponding to a predetermined field and theme.
  • FIG. 16 shows an operation of comparing a hierarchy of fields, themes, and logical relationships for a plurality of thinking systems.
  • Thoughts can be expressed as a combination of words, and logical relationships between words (same, similar, opposite, no relationship etc.) allow words to be defined between words, so this is a pattern between patterns that indicate words It is possible to define logical relationships. For example, since the opposite of "good” is "bad", the logical relationship between the words is also defined by the connection expressing the opposite. This can be realized by generating a connection relation so as to excite a pattern indicating “opposite meaning” when a pattern indicating “good” and a pattern indicating “bad” are excited at the same time.
  • connection relation indicating a logical relation between words is defined.
  • the patterns of person A and person B are matched against the recording area defined above in the same field, theme, and logical connection. This collation can identify the collation of person A, the collation of person B, and other persons by using a plurality of independent collation lines. If the pattern matched by the person A matches the pattern matched by the person B, it is detected that they are common, and conversely the pattern matched by the person A and the pattern matched by the person B show opposite patterns Excitation makes it possible to detect differences.
  • FIG. 4 is a diagram showing the configuration of an autonomous knowledge system builder according to an embodiment of the present invention.
  • reference numeral 1 denotes a pattern converter for converting information into a pattern.
  • Reference numeral 2 denotes a pattern recorder which records a history of patterns, connection relationships between patterns, logical relationships between patterns, and excitations of patterns.
  • Reference numeral 3 denotes a pattern registration unit which registers and changes patterns and connection relationships between patterns by human instruction or autonomously.
  • a pattern controller 4 controls processing of the pattern.
  • Reference numeral 5 is a pattern reverse converter which converts a pattern into information.
  • 6 is a pattern analyzer that analyzes patterns and relationships between patterns.
  • 9 is a priority pattern selector for comparing and evaluating patterns generated by information inputted from different objects and relationships between the patterns and selecting a logically superior pattern.
  • the pattern converter converts information into patterns.
  • the converted patterns are sequentially recorded in the recording module of the two pattern recorders.
  • the pattern registration unit 3 collates the input information with the recording module of the pattern recorder, confirms whether the same pattern as the input information is recorded, and if it is not recorded, records and excites it as a new pattern in the recording module .
  • the excitation refers to a state in which the recording module is activated by collating information or the like. If the same pattern as the input information is already recorded, the corresponding recording module is excited. The history of the activated recording module is sequentially recorded on the pattern recorder.
  • connection information is transferred to the connection information recording unit recording the information.
  • the method of increasing the coupling coefficient which indicates the strength of the transfer and connection of the history of the associated recording module, detects the correlation with the already recorded history information with regard to the individual excitation history as well as a simple superposition. Include methods to increase the appropriate coupling factor. Although this method is somewhat complicated in processing, an increase in the coupling coefficient is caused by the history of the recording module which is not directly related to the excitation of the recording module included in the history information, and it is possible to avoid making an erroneous judgment.
  • FIG. 17 shows an operation of evaluating which one is logically superior as compared with the information recorded in the same field, theme, and logical relationship, which has already been recorded, when new information is input.
  • the new information is compared with the information already recorded, the logical relationship is analyzed, and it is recorded in a predetermined recording area, but at that time, the new information is trusted.
  • record as attached information For example, identify the information source (whether from a reliable source or not), the type of information (truth, fact, rumor, present state, future talk, imagination, assumption, etc.), etc. deep.
  • the new information is analyzed and if there is something to match with the already recorded information, the new information is evaluated as how it was combined with the already recorded information.
  • A is B. Because it is C. It is assumed that the new information is collated with the already recorded information and the "C" is collated with the already recorded information. At this time, the above sentence mentions that "It is C" as a ground "A is B". Therefore, it is possible to evaluate the reliability of "being C" from the information source and the type of information and to judge whether or not it is appropriate. There are two major cases in which it is judged that the reliability of new information is high.
  • the first case is a case where the logic is configured by a combination of information that the new information has already been recorded, and the already recorded information is identified as having high reliability.
  • the second case is a case where the logic is not configured by the combination of the information that the new information has already been recorded, but it is judged that the information source of the new information is reliable.
  • the constructed knowledge system can be updated to one with higher reliability. It becomes possible.
  • FIG. 18 when it is evaluated that newly input information is logically superior to already recorded information, the field, theme, and logical relationship are matched to the recording area in which the related knowledge is stored. The operation to be incorporated is shown. New information is recorded as information of higher priority than the original information.
  • FIG. 5 is a view showing the configuration of an autonomous knowledge system builder according to an embodiment of the present invention.
  • reference numeral 1 denotes a pattern converter which converts information into a pattern.
  • Reference numeral 2 denotes a pattern recorder which records a history of patterns, connection relationships between patterns, logical relationships between patterns, and excitations of patterns.
  • Reference numeral 3 denotes a pattern registration unit which registers and changes patterns and connection relationships between patterns by human instruction or autonomously.
  • a pattern controller 4 controls processing of the pattern.
  • Reference numeral 5 is a pattern reverse converter which converts a pattern into information.
  • 6 is a pattern analyzer that analyzes patterns and relationships between patterns.
  • 7 is a pattern evaluator that evaluates the value of information.
  • 10 is a generalizer that performs semantic analysis and generalization of information.
  • a pattern converter converts information into a pattern.
  • the converted patterns are sequentially recorded in the recording module of the two pattern recorders.
  • the pattern registration unit 3 collates the input information with the recording module of the pattern recorder, confirms whether the same pattern as the input information is recorded, and if it is not recorded, records and excites it as a new pattern in the recording module .
  • the excitation refers to a state in which the recording module is activated by collating information or the like. If the same pattern as the input information is already recorded, the corresponding recording module is excited. The history of the activated recording module is sequentially recorded on the pattern recorder.
  • connection information is transferred to the connection information recording unit recording the information.
  • the method of increasing the coupling coefficient which indicates the strength of the transfer and connection of the history of the associated recording module, detects the correlation with the already recorded history information with regard to the individual excitation history as well as a simple superposition. Include methods to increase the appropriate coupling factor. Although this method is somewhat complicated in processing, an increase in the coupling coefficient is caused by the history of the recording module which is not directly related to the excitation of the recording module included in the history information, and it is possible to avoid making an erroneous judgment.
  • FIG. 19 shows an operation in which generalization or generalization is performed autonomously by sequentially inputting a sentence.
  • Generalization or generalization of the logic is possible by expanding the words or meanings constituting the individual sentences to the same word and further exciting the upper concepts or attributes of the words. For example, suppose that Mr. A likes fruits such as apples, mandarins and strawberries. At this time, a fruit is set as a superordinate concept of apple, and when the word apple is excited, the word fruit is also excited. Similarly, fruits are set as superordinate concepts for oranges and strawberries. Mr A likes mikan. Mr. A likes strawberries. Every time the sentence appears, the connection between the word fruit and the pattern "Mr.
  • FIG. 6 is a diagram showing the configuration of an autonomous knowledge system builder according to an embodiment of the present invention.
  • reference numeral 1 denotes a pattern converter for converting information into a pattern.
  • Reference numeral 2 denotes a pattern recorder which records a history of patterns, connection relationships between patterns, logical relationships between patterns, and excitations of patterns.
  • Reference numeral 3 denotes a pattern registration unit which registers and changes patterns and connection relationships between patterns by human instruction or autonomously.
  • a pattern controller 4 controls processing of the pattern.
  • Reference numeral 5 is a pattern reverse converter which converts a pattern into information.
  • 6 is a pattern analyzer that analyzes patterns and relationships between patterns.
  • 7 is a pattern evaluator that evaluates the value of information.
  • 10 is a generalizer that performs semantic analysis and generalization of information.
  • a fixed pattern driver 11 selects and executes a corresponding treatment pattern in accordance with the analysis result of the pattern.
  • a pattern converter converts information into a pattern.
  • the converted patterns are sequentially recorded in the recording module of the two pattern recorders.
  • the pattern registration unit 3 collates the input information with the recording module of the pattern recorder, confirms whether the same pattern as the input information is recorded, and if it is not recorded, records and excites it as a new pattern in the recording module .
  • the excitation refers to a state in which the recording module is activated by collating information or the like. If the same pattern as the input information is already recorded, the corresponding recording module is excited. The history of the activated recording module is sequentially recorded on the pattern recorder.
  • connection information is transferred to the connection information recording unit recording the information.
  • the method of increasing the coupling coefficient which indicates the strength of the transfer and connection of the history of the associated recording module, detects the correlation with the already recorded history information with regard to the individual excitation history as well as a simple superposition. Include methods to increase the appropriate coupling factor. Although this method is somewhat complicated in processing, an increase in the coupling coefficient is caused by the history of the recording module which is not directly related to the excitation of the recording module included in the history information, and it is possible to avoid making an erroneous judgment.
  • FIG. 20 shows an operation in which many thoughts are detected as a fixed pattern and appropriate processing is autonomously performed according to the detected type. Examples of typical types are described below.
  • Control of thought patterns is also defined as a pattern, and control of thought patterns can be implemented by changing from patterns to patterns. So-called, "concept of things” is defined as a pattern, and while changing this pattern, it is an idea of controlling the patterns described so far. When thinking, humans apply and respond to appropriate measures and ideas for individual events. Similarly, it is possible to cause the present knowledge system builder to learn how to cope with individual events, thinking, etc., and to perform the treatment autonomously.
  • a pattern corresponding to each of the types shown above is defined, and the pattern is sequentially changed and implemented.
  • Each control pattern is excited if a certain condition is satisfied, and processing will be executed.
  • detection of the condition is taught by human at the initial stage, and this knowledge system construction machine learns sequentially the pattern of the condition taught by human. Enables autonomous operation. Below, it demonstrates a little concretely.
  • (1) Treatment of input information When information is input, collation with already recorded information, position analysis, recording, updating of already recorded information, etc. are performed.
  • the input information is analyzed in terms of fields and themes, and collated with information already recorded. If there is any collation with the information already recorded, the position of the input information is analyzed. The position is evaluated from the logical connection relation with the recorded information, reliability, etc., and if the input information is useful (newness, logical validity, reliability), the corresponding field, theme, logical connection Add the relevant knowledge to In addition, if necessary, update the already recorded information or change the priority.
  • These actions can be defined by patterns. Analysis of information and themes and analysis of position is the method described in the previous section.
  • FIG. 21 detects input information, field, theme analysis, search of related information already recorded, reliability with related information, evaluation of novelty etc., processing of input information (incorporation into knowledge system, discard etc. An example of the operation of.
  • FIG. 22 shows an operation example for appropriately processing according to the analysis result after analyzing the type of input information (instruction sentence, request sentence, question sentence, ordinary sentence, etc.).
  • Planning, scheduling, execution of action Planning, execution, evaluation, and execution of action are implemented.
  • Planning, execution, evaluation, and execution of the operation can be expressed as a pattern corresponding to the conditional operation.
  • it is possible to generate a control signal for controlling the drive system by defining a number of motion patterns and generating a pattern corresponding to each motion pattern in time series.
  • FIG. 24 shows an operation example such as generating a control signal according to the situation from an image pattern or the like.
  • the first aspect of the present invention while converting the input information into a pattern and sequentially recording, analyzing and recording the information and the structure of the information by a human instruction, and autonomously learning the value of the input information It is possible to evaluate and add new knowledge or update knowledge in the internally built knowledge system, and autonomously carry out the expansion of knowledge and the systematization of knowledge.
  • Conventionally when making a machine process input information, it is necessary to set a program in advance in a computer. It was necessary to create a program for judging the situation from the input information and a program to operate the machine according to each condition, and install and execute it on a computer mounted on the machine. The program needs to be created in a dedicated programming language, and has the disadvantage of requiring a lot of time for development.
  • the input information is converted into a pattern and sequentially recorded, and the information and structure of the information are analyzed and recorded by a human instruction, and the set of individual books, sentences, conversations and utterances are set.
  • a human instruction By inputting, it is possible to construct the thinking system.
  • the third invention by inputting books, sentences, conversations and utterances of a plurality of persons, it is possible to analyze the common points and differences of the thoughts and to compare the thinking systems.
  • Conventionally when making a machine process input information, it is necessary to set a program in advance in a computer. It was necessary to create a program for judging the situation from the input information and a program to operate the machine according to each condition, and install and execute it on a computer mounted on the machine.
  • the program needs to be created in a dedicated programming language, and has the disadvantage of requiring a lot of time for development. If the condition detection and the corresponding operation are not appropriate, the program installed in the computer needs to be corrected by a human, and there is a disadvantage that the correction takes a lot of time.
  • the present invention since it is not necessary to design and execute a program etc. in comparison of thinking systems, development effort can be greatly reduced. [Advantage of the Invention 4] (Corresponding to Claim 4)
  • the fourth invention by inputting the books, sentences, conversations and remarks of a plurality of persons, the superiority of the logical thinking is analyzed, and the knowledge is constructed based on the superior logical thinking.
  • Conventionally when making a machine process input information, it is necessary to set a program in advance in a computer. It was necessary to create a program for judging the situation from the input information and a program to operate the machine according to each condition, and install and execute it on a computer mounted on the machine.
  • the program needs to be created in a dedicated programming language, and has the disadvantage of requiring a lot of time for development.
  • the value of input information is evaluated by learning autonomously, addition of new knowledge, updating of knowledge and generalization of knowledge are performed to the internally built knowledge system, and knowledge is expanded. And systematization of knowledge can be implemented autonomously.
  • Conventionally when making a machine process input information, it is necessary to set a program in advance in a computer. It was necessary to create a program for judging the situation from the input information and a program to operate the machine according to each condition, and install and execute it on a computer mounted on the machine. The program needs to be created in a dedicated programming language, and has the disadvantage of requiring a lot of time for development.
  • the value of input information is evaluated by learning autonomously, addition of new knowledge, updating of knowledge and generalization of knowledge are performed to the internally built knowledge system, and the information input Appropriate action can be carried out autonomously.
  • Conventionally when making a machine process input information, it is necessary to set a program in advance in a computer. It was necessary to create a program for judging the situation from the input information and a program to operate the machine according to each condition, and install and execute it on a computer mounted on the machine. The program needs to be created in a dedicated programming language, and has the disadvantage of requiring a lot of time for development.
  • the program installed in the computer needs to be corrected by a human, and there is a disadvantage that the correction takes a lot of time.
  • the present invention since it is not necessary to design and execute a program or the like for autonomously performing an appropriate process on input information, it is possible to significantly reduce development effort.
  • Configuration example of an autonomous knowledge system builder (corresponding to claim 1) Configuration example of an autonomous knowledge system builder (Claim 2) Configuration example of an autonomous knowledge system builder (Claim 3) Configuration example of an autonomous knowledge system construction machine (corresponding to claim 4) Configuration example of an autonomous knowledge system construction machine (corresponding to claim 5) Configuration example of an autonomous knowledge system builder (corresponding to claim 6)
  • Example of operation that connection with related pattern is strengthened Operation example in which the connection with the repetitively excited pattern is strengthened
  • Example of operation in which the logical relationship between patterns and patterns is recorded
  • Logical relationship identification example Operation example of field identification Matching example with information in same field, theme Operation example of relationship evaluation with already recorded information Operation example incorporated into the knowledge system Operation example of identifying and recording personal information in fields, themes, and logical relationships
  • Example of operation to compare multiple thinking systems Operation example to evaluate whether it is logically superior Operation example of updating to a logically superior pattern Operation example of generalization and generalization Example of operation in which appropriate measures are performed autonomously

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Mathematical Physics (AREA)
  • Physics & Mathematics (AREA)
  • Computing Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Software Systems (AREA)
  • Data Mining & Analysis (AREA)
  • Evolutionary Computation (AREA)
  • Computational Linguistics (AREA)
  • Artificial Intelligence (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
  • Machine Translation (AREA)

Abstract

Le problème décrit par la présente invention est que, classiquement, il est nécessaire de régler un programme dans un ordinateur à l'avance lors de l'exécution d'un traitement et similaire d'informations d'entrée dans une machine. Par extension, il est classiquement difficile de construire automatiquement une logique et un système de connaissances à partir d'informations d'entrée. La solution selon l'invention porte sur un dispositif pour construire un système de connaissances autonome, ledit dispositif étant pourvu : d'un convertisseur de motif qui convertit des informations en un motif ; d'un dispositif d'enregistrement de motif qui enregistre un motif, une relation de correspondance entre des motifs, une relation logique entre des motifs, et un historique d'excitation de motif ; d'un dispositif d'inscription de motif qui inscrit et change une relation de correspondance entre des motifs conformément à une instruction provenant d'une personne ou de manière autonome ; d'un dispositif de commande de motif qui commande un traitement de motif ; d'un convertisseur inverse de motif qui convertit un motif en informations ; et d'un dispositif d'analyse de motif qui analyse un motif et la relation entre des motifs.
PCT/JP2015/062160 2014-12-17 2015-04-14 Dispositif pour construire un système de connaissances autonome WO2016098366A1 (fr)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
JP2014267193A JP5821141B2 (ja) 2014-12-17 2014-12-17 自律型知識体系構築機
JP2014-267193 2014-12-17

Publications (1)

Publication Number Publication Date
WO2016098366A1 true WO2016098366A1 (fr) 2016-06-23

Family

ID=53439349

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/JP2015/062160 WO2016098366A1 (fr) 2014-12-17 2015-04-14 Dispositif pour construire un système de connaissances autonome

Country Status (2)

Country Link
JP (1) JP5821141B2 (fr)
WO (1) WO2016098366A1 (fr)

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP5854250B2 (ja) * 2015-06-17 2016-02-09 洋彰 宮崎 人工知能装置
JP5854251B2 (ja) * 2015-07-08 2016-02-09 洋彰 宮崎 人工知能装置

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPH0520074A (ja) * 1991-07-16 1993-01-29 Fujitsu Ltd 知識ベース編集装置及びエキスパートシステム
JP2010271819A (ja) * 2009-05-20 2010-12-02 Nec Corp 語句関係抽出装置、語句関係抽出方法及びプログラム
JP2014222504A (ja) * 2014-05-24 2014-11-27 洋彰 宮崎 自律型思考パターン生成機

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPH0520074A (ja) * 1991-07-16 1993-01-29 Fujitsu Ltd 知識ベース編集装置及びエキスパートシステム
JP2010271819A (ja) * 2009-05-20 2010-12-02 Nec Corp 語句関係抽出装置、語句関係抽出方法及びプログラム
JP2014222504A (ja) * 2014-05-24 2014-11-27 洋彰 宮崎 自律型思考パターン生成機

Also Published As

Publication number Publication date
JP2015109098A (ja) 2015-06-11
JP5821141B2 (ja) 2015-11-24

Similar Documents

Publication Publication Date Title
WO2015182007A1 (fr) Générateur de modèle de pensée autonome
US10810509B2 (en) Artificial intelligence apparatus autonomously expanding knowledge by inputting language
JP5737642B2 (ja) 自律型知識向上装置
WO2016067478A1 (fr) Machine de construction logique automatique
JP5737643B2 (ja) 自律学習型知識構築機
JP5807831B2 (ja) 自律型問題解決機
JP5810476B2 (ja) 自律型知識抽出機
WO2016098366A1 (fr) Dispositif pour construire un système de connaissances autonome
JP5881030B2 (ja) 自己組織的に知識を拡大する人工知能装置
JP5821142B2 (ja) 自律型知能体系構築機
JP5807830B2 (ja) 自律型知能生成機
WO2016125320A1 (fr) Analyseur de connaissances autonome
JP5854392B2 (ja) 暴走しない人工知能装置
Yared et al. Plan recognition and generalization in command languages with application to telerobotics
JP5854251B2 (ja) 人工知能装置
Mrasek et al. User-friendly property specification and process verification–a case study with vehicle-commissioning processes
JP5828437B2 (ja) 自律型処理手順生成および実行機
JP5828438B2 (ja) 自律型知識体系構築機
JP5854250B2 (ja) 人工知能装置
Pearson et al. Incremental learning of procedural planning knowledge in challenging environments
Connell Verbal Programming of Robot Behavior
Wong et al. A Hybrid Approach and Implementation on Root Cause Analysis Logic Model for Enterprise Software Application
Kersting et al. Learning to transfer optimal navigation policies
JP2015207298A (ja) 人工知能装置
Galetta et al. A relative identification method for reactive systems

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 15869575

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 15869575

Country of ref document: EP

Kind code of ref document: A1