Fujitsu Patente

A processing unit identifies a first joint element from a first patch element group included in a first patch that appears in the course of lattice surgery, performs an optimization process for optimizing an application order of controlled NOT (CNOT) gates for a syndrome measurement circuit in the first joint element with respect to a direction of a logical operator defined for the first joint element, and stores, in a storage unit, optimal circuit information associating the first joint element with the application order of CNOT gates. The processing unit identifies, upon detecting a second path during execution of lattice surgery, a second joint element corresponding to the first joint element from a second patch element group included in the second patch, and assigns the application order of CNOT gates associated with the first joint element to the second joint element, based on the optimal circuit information.

An information processing apparatus (100) classifies time-series frames into a plurality of clusters by executing clustering on the time-series frames based on feature amounts of the time-series frames, the time-series frames being included in a moving image belonging to a class of a real moving image or a class of a fake moving image, extracts a concept from each of the plurality of clusters, determines whether the concept extracted from the plurality of clusters is a concept specific to the class or a concept specific to the cluster, sets a weight to the plurality of clusters based on the determination result, and trains a model for estimating a class of a target moving image based on a frame selected based on weights set for the plurality of clusters.

An information processing program causes a computer to execute a process including, in a machine learning model configured to have layers, an input of quantum data being accepted at a first layer, a preceding layer of two consecutive layers included in the layers inputting an output generated by the generation function based on an output of the quantum circuit to the quantum circuit included in a subsequent layer, inputting first quantum data to the first layer and causing first output data to be generated, and inputting, in each second layer arranged in a subsequent stage to the first layer, second quantum data to a quantum circuit that has generated a quantum feature map for an output from a layer arranged in a preceding stage of each second layer, and causing second output data to be generated.

A method comprising: generating subgraphs graph based on classification scores corresponding to nodes and obtained by inputting input graph into a graph neural network, GNN, wherein each node in each input graph comprises a plurality of values corresponding respectively to a plurality of features; determining statistical measures of each feature's values in the subgraphs to generate statistical measures for each subgraph, and obtaining for each subgraph classification probabilities corresponding respectively to a plurality of classes; performing k-means clustering on the subgraphs based on the statistical measures to generate a plurality of clusters of subgraphs; generating, for each cluster of subgraphs, a set of surrogate models to approximate the classification probabilities, respectively, of the subgraphs in the cluster, in terms of the statistical measures; and determining, based on the surrogate models of each set of surrogate models, feature importance scores.

A computer-implemented method comprising: generating, using a large language model, LLM, and based on information describing a threat to a network, an attack graph, wherein the attack graph comprises nodes corresponding to attack steps and directed edges connecting the nodes; obtaining preliminary attack implementations based on the attack steps in the attack graph, respectively, each preliminary attack implementation comprising code for executing the attack step concerned; assigning, based on information about the network, values to parameters in the preliminary attack implementations to generate objective attack implementations; running an attack on the network or on a simulation of the network by carrying out the objective attack implementations; and outputting a report comprising information indicating outcomes of carrying out the objective attack implementations.

A method for 3D human-object interaction generation using neural language models is disclosed. The method includes receiving an input comprising an interaction location and a text prompt. A text-to-image generation model is applied on the text prompt. Initial differential parameters for a 3D parametric model representing the person in an interaction image are determined and a plurality of 2D views of the 3D environment is extracted. A sequence of prompts is prepared based on the plurality of 2D views and a neural language model is applied to the sequence of prompts. Loss function is initialized based on a set of body parts and a set of locations. The initial differential parameters are iteratively updated based on the loss function to obtain final differential parameters. The 3D parametric model is rendered in the 3D environment based on the final differential parameters.

Provided is an estimation method including generating a probability model for estimating a probability that a prospective purchaser changes purchase from a first commodity to a second commodity based on prices of the first commodity and the second commodity, a price change amount of the second commodity, and information indicating whether a purchaser has actually changed the purchase from the first commodity to the second commodity according to price change of the second commodity (S5), and calculating an estimated number of purchases of each of the first commodity and the second commodity based on the probability model (S6).

A computer-implemented method for traffic simulation, the method comprising: accepting input of traffic data for a geographic area; executing a traffic simulation, wherein the traffic simulation comprises a main model configured to progress a traffic system through timesteps, and a sub model configured to perform an ancillary traffic simulation function for a later timestep using traffic data corresponding to an earlier timestep; and outputting a status of the traffic system.

An information processing program causes a computer to execute a process including acquiring video having a scan area of a self-checkout terminal for scanning a barcode of a product analyzing the acquired video to detect an action of a user holding a plurality of products in the scan area of the self-checkout terminal setting either the self-checkout terminal or a management terminal that monitors the self-checkout terminal as a terminal to be notified based on an analysis result of a type of an overlapping state of the plurality of products when the action of the user holding the plurality of products is detected generating alert image information corresponding to the type of the overlapping state of the plurality of products; and causing the generated alert image information to be displayed on a display screen of the terminal set as a notification destination.

Methods and systems for geospatial climate profiling are disclosed. According to an aspect of an embodiment, a method includes obtaining first real-time observational data corresponding to current weather conditions in or near a specified location. The first real-time observational data is obtained based on the first real-time observational data corresponding to surface observations in, on or near the specified location. A virtual weather surveillance image (WSR) image of the specified location is generated using an artificial intelligence (AI) radar image generation model and based on the first real-time observational data. The method can further include generating one or more climate profiles for the specified location based on the virtual WSR image, the first real-time observational data, and one or more contemporaneous satellite images of the specified location.

According to an aspect of an embodiment, operations may include generating a first representation of a first event of a process flow corresponding to an incident. A second representation of a second event that may be a prediction of a future event to follow the first event in the process flow may be generated based on the first representation using an artificial intelligence model. A third representation of a third event that may be a prediction of a future event to follow the second event in the process flow may be generated based on the first representation and the second representation using the artificial intelligence model. A predicted event sequence for the entire process flow may be generated based on an order of the first event, the second event, and the third event as predicted.

A method for synthetic data generation, the method comprising: accepting input of a target geographic region; acquiring aggregated demographic data for the target geographic region; generating a synthetic population based on the aggregated demographic data using iterative proportional fitting; acquiring infrastructure information for the target geographic region; assigning the synthetic population to a plurality of households using the infrastructure information, thereby forming synthetic population data; outputting the synthetic population data for further use.

A computer-implemented method of automatically determining optimum electric vehicle (EV) charging station real-world locations, based on a simulation of a particular real-world target area, comprising: receiving real-world EV data, comprising data for a plurality of target EVs associated with the target area, and for each target EV comprising at least one start location and associated start time, at least one finish location and associated finish time, and a distance between the start location and finish location; determining charging demand locations in the target area, being locations in which the target EVs need to charge, comprising performing a simulation of the routes taken by the target EVs based on the EV data, and plotting the charging demand locations onto a map of the target area; retrieving population density data for the target area and plotting the population density data onto the map of the target area; clustering, based on the map of the target area, the charging demand locations to determine candidate EV charging station locations; and determining optimum EV charging station locations, comprising weighting the candidate EV charging station locations based on the population density data, and plotting the optimum EV charging station locations onto the map of the target area.

A computer implemented method for training a graph neural network, GNN, to classify a medical condition of samples using a biological dataset, comprising loading the biological dataset, the biological dataset comprising samples and the samples comprising features and a classification of the medical condition, extracting relationships between the features from a knowledge base, constructing, for the samples, graphs comprising nodes and edges between the nodes, wherein the nodes correspond to the features and the edges correspond to the relationships; and training the GNN with the graphs to provide the classification of the medical condition for the samples.

An information processing apparatus extracts a subset of vector data from a vector database. The information processing apparatus calculates the value of a predetermined statistical indicator using the extracted subset of the vector data. The information processing apparatus generates a function for obtaining the amount of resources to be used for calculating the value of the statistical indicator for the vector database, based on a first resource amount used for calculating the value of the statistical indicator using the subset of the vector data. Then, the information processing apparatus calculates, using the generated function, a second resource amount to be used for calculating the value of the statistical indicator for the vector database.

An explanatory material generating method executed by a computer, includes: receiving personal information representing an attribute of a user; specifying for each of a plurality of factors included in a first causal graph, a weight corresponding to the attribute represented by the received personal information; and generating explanatory material of a causal relationship expressed in the first causal graph, based on the first causal graph and the specified weight of each factor, using a machine learning model configured to generate information for explaining a causal relationship expressed in an input causal graph.

According to an aspect of an embodiment, operations may include obtaining a dataset including a plurality of sequential events and supplemental data indirectly indicating an order of sequential events. The operations may additionally include determining a portion of the supplemental data is corrupted. In some embodiments, the operations may include obtaining a validation dataset including potential values that may be included in the supplemental data. In some embodiments, the operations may further include restoring the corrupted portion of the supplemental data based on comparing the supplemental data with one or more of the potential values included in the validation dataset.

Provided is a causal model generation method that causes a computer to execute a process. The process includes identifying a timing at which the specific event occurred in each of a plurality of second data acquisition ranges in which the specific event previously occurred (S14), evaluating a similarity between a plurality of types of data acquired in each of the plurality of second data acquisition ranges in a focused period before an identified timing and a plurality of types of data acquired in the first data acquisition range (S30 - S42), identifying a second data acquisition range that is most similar to the first data acquisition range in the plurality of second data acquisition ranges based on a result of the evaluating (S44), and generating a causal model when the specific event occurs in the first data acquisition range using data acquired after the specific event occurs in an identified second data acquisition range (S50 - S56).

A processing unit generates a plurality of first quantum circuits each including a quantum gate serving as a target of noise measurement and a plurality of single-qubit gates that perform a twirling operation, the first quantum circuits differing from each other in the number of repetitions of the gate operation of the quantum gate. The processing unit generates, for each of the plurality of first quantum circuits, a plurality of second quantum circuits that are quantum circuits that implement the same unitary operation and differ from each other in the number of first single-qubit gates among the plurality of single-qubit gates. The processing unit causes a quantum computer to execute the plurality of second quantum circuits, and estimates noise of the quantum gate in a state where no first single-qubit gate is included, by extrapolation based on the execution results of the plurality of second quantum circuits.

A computer-implemented method is provided for performing automatic gate-cutting of a target quantum circuit into a set of output subcircuits, comprising: identifying a target quantum circuit comprising a number of qubits and entanglement gates; identifying a subcircuit constraint defining the maximum number of qubits available per output subcircuit of the set of output subcircuits; performing a cutting algorithm to determine a set of optimum subcircuits, the cutting optimisation algorithm comprising: iteratively evaluating a cost function which determines candidate sets of subcircuits based on an assessment of the entanglement gates of the target quantum circuit, comprising determining for each candidate set of subcircuits: whether the subcircuit constraint is satisfied; and a subset of the entanglement gates to be cut to achieve the candidate set of subcircuits; minimising the cost function and defining the set of optimum subcircuits as the respective candidate set of subcircuits for which the cost function is minimised; creating the set of output subcircuits, comprising cutting each of the entanglement gates of the respective subset of the entanglement gates to be cut to achieve the set of optimum subcircuits.

A method may include generating directed graphs, each of the directed graphs representing a quantum circuit. The method may also include evaluating each of the directed graphs according to operation of the quantum circuit represented by each of the directed graphs. The method may include selecting one of the directed graphs based on the evaluations. The method may further include transforming the selected directed graph to generate a second directed graph.

According to an aspect of an embodiment, operations include obtaining initial quantum circuit comprising entangling gates to generate first randomized quantum circuits by applying RC protocol on initial quantum circuit. The operation further includes obtaining first combined measurement results by executing first plurality of randomized quantum circuits on quantum computer. The operation includes generating plurality of random noise-magnified quantum circuits by applying ZNE protocol on initial quantum circuit. The operation further includes generating second randomized quantum circuits by applying RC protocol on each random noise-magnified quantum circuit and obtaining second combined measurement results by executing second plurality of randomized quantum circuits on quantum computer. Finally, operation includes generating final measurement results for initial quantum circuit by applying extrapolation method on first combined measurement results and second combined measurement results.

According to an aspect, operations include acquiring a sequence of video frames associated with video data and a query. The operations further include acquiring a temporal graph associated with the sequence of video frames . The temporal graph includes a set of subgraphs corresponding to the sequence of video frames, with a set of nodes and edges in each subgraph representing the set of entities and relationships between the set of nodes, respectively. The operations further include converting the query into a set of retrieval functions and applying the set of retrieval functions on the temporal graph to retrieve a query response including at least one video frame of the set of video frames.

A computer implemented method for forecasting traffic flow using a pretrained recursive model, comprising, inputting multivariate data at a forecast time-point into the pretrained recursive model, and predicting, using the pretrained recursive model, a traffic flow at the forecast time-point, wherein the pretrained recursive model is trained using a multivariate dataset comprising a time-series traffic flow variable and an additional time-series variable at times before the forecast time-point, and wherein the multivariate data at the forecast time-point input into the pretrained recursive model comprises the additional time-series variable at the forecast time-point.

According to an aspect of an embodiment, operations may include obtaining data corresponding to a completed process flow, the completed process flow being associated with an incident that has been resolved. The operations may further include generating one or more labels that serve as target variables for training a machine learning model based on the data corresponding to the completed process flow. The operations may additionally include extracting a subset of the data corresponding to a predefined time range cutoff. Further, the operations may include generating a plurality of features derived from the data corresponding to the completed process flow for training the machine learning model and training the machine learning model to predict future behavior of ongoing process flows using the plurality of features and the one or more labels associated with the data corresponding to the completed process flow.

A processing unit causes a quantum computer to execute a quantum circuit including a first-polarity first control gate and a second-polarity second control gate, each controlled by an auxiliary qubit. The first control gate applies a first time-evolution operator corresponding to a Hamiltonian H(λ) to a qubit group corresponding to a quantum state |ψ>. The second control gate applies a second time-evolution operator corresponding to H(λ + δλ) to the qubit group. The processing unit acquires, from the quantum computer, a probability amplitude corresponding to an event that a result of applying the two types of time-evolution operators to the quantum state |ψ> matches |ψ>. The processing unit acquires a cumulative distribution function for an energy level difference between H(λ) and H(λ + δλ) by executing discrete Fourier transform using the probability amplitude, and estimates the difference based on the cumulative distribution function.

An information processing apparatus generates a waveform function that defines a temporal change in amplitude in a first interval of a rise time of a pulse signal using a first polynomial having a first parameter as a coefficient and defines a temporal change in the amplitude in a second interval of the rise time using a second polynomial having a second parameter as a coefficient. The information processing apparatus determines a waveform using the waveform function by assigning a first parameter value group to the parameter group. The information processing apparatus changes a parameter value group assigned to the parameter group, from the first parameter value group to a second parameter value group, based on a result of quantum computation performed by a quantum computer using a pulse signal having the waveform.

A machine learning device includes a processor that executes a procedure. The procedure including: specifying, among a plurality of points that are present within a predetermined distance from a first point in a latent space of an autoencoder, a second point that satisfies a predetermined relationship with the first point; and updating at least a parameter of a decoder of the autoencoder by optimization of an objective function including a regularization term to reduce a difference between the first point and the second point in the latent space.

An information processing apparatus determines, based on target information regarding an imitation target, a first knowledge domain in which the imitation target has knowledge and a first knowledge level of the imitation target regarding the first knowledge domain. Upon receiving a first question, the information processing apparatus instructs an agent, which responds to a question by using a language model, to generate a response to the first question. In instructing the response, the information processing apparatus instructs a response generated by imitating thoughts of the imitation target that are referable to information restricted based on the first knowledge domain and the first knowledge level.

An information processing apparatus acquires error data indicating an error of a first quantum gate. The information processing apparatus generates, using an approximation function that linearly approximates an influence of an additional quantum gate, which is to be added to the first quantum gate, on an operation of the first quantum gate, a linear equation which includes a variable corresponding to the additional quantum gate and indicates a relationship in which the error is canceled by the additional quantum gate. The information processing apparatus determines a second quantum gate corresponding to the additional quantum gate by solving the linear equation for the variable. The information processing apparatus determines a second quantum gate corresponding to the additional quantum gate by solving the linear equation for the variables.

An information processing apparatus performs, a plurality of times, an update process of updating a value of a first parameter, which is a variable included in a cost function applied to a variational quantum circuit used for variational quantum eigenvalue computation. The information processing apparatus determines a value of a second parameter representing a weight for an amount of change in the value of the first parameter in each update process, by adding a constant term and a variable term expressed using a ratio between the values of the cost function calculated through the variational quantum eigenvalue computation using the values of the first parameter obtained in the k-th and (k-1)-th update processes. The information processing apparatus performs the (k+1)-th update process using the amount of change weighted by the determined value of the second parameter.

A processing unit acquires, for a plurality of terms included in a Hamiltonian corresponding to a problem to be solved by a variational quantum eigensolver and capable of being simultaneously measured by a quantum computer, first information indicating the variance of each term and the covariance of each pair of terms, upon performing a simultaneous measurement, generates, based on the first information, an optimization problem that minimizes a total sum of the variance of each term and the covariance of each pair based on the constraint that the amount of a change with respect to an average value of the sum of the terms is equal to or less than a predetermined value, determines a plurality of first terms to be simultaneously measured by the quantum computer by solving the optimization problem, and computes the number of samples of the simultaneous measurement based on the minimized total sum.

Provided is an improvement proposal presentation method including estimating a dialogue stage for each dialogue of a plurality of dialogues included in a conversation based on the plurality of the dialogues and a first large language model (S4), estimating a final dialogue stage among dialogue stages as a dropout stage of the conversation (S5), and outputting an improvement proposal according to the dropout stage based on a dialogue content of the plurality of the dialogues, the dropout stage, and the first large language model (S7).

In an embodiment, a parameterized quantum circuit is initialized on a quantum computer for a Quantum Neural Network (QNN) with an ansatz architecture. Input data, which includes a quantum representation of a multi-dataset and an initial state of the QNN, and labels, is received. The quantum representation of the multi-datasets on the QNN is loaded and a first dataset of the multi-dataset is labelled. A label-controlled circuit associated with label-controlled input qubits of the ansatz architecture for the QNN is determined, based on the labelled first dataset. A universal circuit associated with the ansatz architecture for a prediction circuit associated with the QNN is determined, based on labelled first dataset. The prediction circuit is determined based on the label-controlled circuit and the universal circuit. The prediction circuit is trained and configured to generate predictions associated with the loaded fractionally weighted data portfolios.

A quantum device having a plurality of quantum bits (1) connected to each other via coupling wiring lines (30), wherein each of the plurality of quantum bits includes: a Josephson junction device (4); and a quantum bit electrode (50) connected to one end of the Josephson junction device; and the quantum device including: a ground electrode (51) to which the other end of the Josephson junction device is connected; an insulating region (52) between the quantum bit electrode and the ground electrode; and a guard structure (54) that is provided along a portion of the coupling wiring line extending in the insulating region and is fixed to a ground potential.

A computer implemented method for determining an origin of indication information in a networked connection between a client device and a server, comprising: receiving, at the server, the indication information of the client device, determining a client-server value based on a transmission time of a connection signal transmitted between the client device and the server, transmitting a tracing signal addressed to the indication information via a particular router in the networked connection between the client device and the server, determining a tracing value based on a transmission time of the tracing signal to the particular router, establishing whether the indication information originates from an intermediate connection in the networked connection or from the client device by comparing a difference value with a threshold value, the difference value being based on the client-server value and tracing value.

A computer obtains a causal relationship including a plurality of explanatory variables and a response variable by performing causal discovery using a subset extracted from a data aggregate including a value of each of the explanatory variables and a value of the response variable on the basis of each of a plurality of conditions related to ranges of the values of the explanatory variables. The computer obtains a causal effect of any explanatory variable of the explanatory variables on the response variable on the basis of the causal relationship obtained on the basis of each of the conditions. The computer obtains a determination result of similarity in the causal effects obtained on the basis of the respective conditions, by determining the similarity. The computer obtains an identified range of the value of each of the explanatory variables in which the causal effects have a predetermined sign on the basis of the determination result of the similarity and outputs range information indicating the identified range.

An information processing device executes a quantum circuit multiple times and classifies combinations of determination results of first conditional branches in a first portion of the quantum circuit into multiple groups, based on the similarity of the combinations of the determination results of a second conditional branch in the second portion. The information processing device generates first information indicating a method of allocating physical qubits to the first portion, and generates and associates with each group, second information indicating a method of allocating physical qubits to the second portion. After executing the first portion according to the generated first information, the information processing device controls an executing unit to execute the second portion according to any of the second information, according to the current combination of the determination results of the first conditional branch.

An information processing apparatus divides a quantum circuit into subcircuits each assigned to one of a plurality of quantum processors. The information processing apparatus predicts, for each of the subcircuits, fidelity when the subcircuit is executed by the quantum processor assigned thereto. The information processing apparatus repeats the process of dividing the quantum circuit into the subcircuits by increasing the number of subcircuits and the process of predicting the fidelity until the fidelity satisfies a predetermined criterion. When the criterion is satisfied, the information processing apparatus instructs a quantum computer to execute each of the subcircuits by the quantum processor assigned thereto. The information processing apparatus calculates, by classical computation, a computation result of the quantum circuit based on execution results of the subcircuits.

A vectorization unit (12) converts a first sentence and a second sentence into vector values using a sentence embedding model. A similarity calculation unit (13) calculates a first similarity of the first sentence to the second sentence based on the vector value of the first sentence and the vector value of the second sentence. A feature value extraction unit (14) extracts feature values of the first sentence and the second sentence. A corrected similarity calculation unit (15) calculates a second similarity by correcting the first similarity by weighting based on a difference between the feature values of the first sentence and the second sentence.