System and method for generating candidate idea
A system converts a first input idea vector representing an idea into a first contracted vector. The system generates one or more second contracted vectors in a multivariate space to which the first contracted vector belongs, based on a value of a first predetermined component of the first contracted vector in the multivariate space. The system generates, respectively from the one or more second contracted vectors, one or more first output idea vectors representing a candidate idea to be proposed to a user. The multivariate space is configured to maintain a similarity between an input idea vector for generating a contracted vector and an output idea vector generated from the contracted vector and a similarity between the first predetermined component of the contracted vector and a first predetermined index value.
1 . A system for generating a candidate idea, the system comprising:
one or more processors; and
one or more storage devices configured to store programs to be executed by the one or more processors, wherein
the one or more processors
convert a first input idea vector into a first contracted vector belonging to a multivariate space using a trained model,
extract one or more second contracted vectors from the multivariate space, based on a value of a first predetermined component of the first contracted vector in the multivariate space, and
output, respectively based on the one or more second contracted vectors, one or more first output idea vectors representing a candidate idea to be proposed to a user using the trained model, and
wherein the trained model is trained using an objective function comprising a term representing similarity between an input idea vector and an output vector for a contracted vector and a term representing similarity between the first predetermined component and a first predetermined index value.
2 . The system according to claim 1 , wherein
the one or more processors generate the one or more second contracted vectors based on the first predetermined component and a second predetermined component of the first contracted vector, and
the objective function further comprises a term representing similarity between the second predetermined component and a second predetermined index value.
3 . The system according to claim 1 , wherein
the similarity between the first predetermined component and the first predetermined index value indicates a correlation between the first predetermined component and the first predetermined index value, and
the similarity between the input idea vector and the output idea vector is represented by a Euclidean distance.
4 . The system according to claim 1 , wherein
the one or more processors apply one or more predetermined biases to the first predetermined component of the first contracted vector in generating of the one or more second contracted vectors.
5 . The system according to claim 1 , wherein
the one or more processors select, from a plurality of ideas, an idea in which a difference between the first predetermined index value and a target value is within a predetermined range, and generate the first input idea vector from the selected idea.
6 . The system according to claim 1 , wherein
the one or more storage devices include a variable auto-encoder including an encoder and a decoder, and
the one or more processors
generate the one or more second contracted vectors using the encoder, and
generate the one or more first output idea vectors using the decoder.
7 . The system according to claim 1 , wherein
the one or more second contracted vectors are a plurality of second contracted vectors, and
in generating of the plurality of second contracted vectors, the one or more processors apply different biases to the first predetermined component of the first contracted vector for the plurality of second contracted vectors.
8 . The system according to claim 1 , wherein
the one or more storage devices manage a proposed idea,
the one or more second contracted vectors are a plurality of second contracted vectors, and
the one or more processors select the second contracted vector of the candidate idea to be proposed to the user, based on a similarity between a contracted vector of the proposed idea in the multivariate space and the plurality of second contracted vectors.
9 . The system according to claim 1 , wherein
the idea includes at least one of a material and a phrase, and
the one or more processors convert the one or more first output idea vectors into an expression representing an idea and displays the expression.
10 . A method in which a system generates a candidate idea, the method comprising:
converting, by the system, a first input idea vector into a first contracted vector belonging to a multivariate space using a trained model;
extracting, by the system, one or more second contracted vectors from the multivariate space, based on a value of a first predetermined component of the first contracted vector in the multivariate space; and
outputting, by the system, respectively based on the one or more second contracted vectors, one or more first output idea vectors representing a candidate idea to be proposed to a user using the trained model,
wherein the trained model is trained using an objective function comprising a first term representing similarity between an input idea vector and an output vector for a contracted vector and a second term representing similarity between the first predetermined component and a first predetermined index value of quality,
wherein the second term represents similarity via correlation and the object function reduces loss as correlation increases.