Abstract: In an embodiment, a method is provided for time series anomaly detection using temporal correlation weights. The method involves receiving a dataset to prepare a data input for a neural network with first decoder, second decoder, and encoder. Temporal embeddings and sampling temporal correlation weights are generated from the data input. Further, anomalous time series data is reconstructed by applying the first decoder to the temporal embeddings. The method further involves generating a first discriminator output by applying the second decoder to the temporal embeddings and feeding the anomalous time series data to the encoder to produce anomalous temporal embeddings and correlation weights. A second discriminator output is generated by applying the second decoder to the anomalous temporal embeddings.

Abstract: Techniques are disclosed relating to feature selection based on feedback-assisted optimization models. In various embodiments, for example, the disclosed techniques include accessing a training dataset that includes a plurality of data samples that include data values for a plurality of features, and a set of labels corresponding to the plurality of data samples. In some embodiments, a computer system performs feature-selection operations to select, from the plurality of features, a subset of features to include in a reduced feature set. For example, in some embodiments the feature-selection operations include processing the training dataset based on an optimization model, where an objective function utilized in the optimization model utilizes performance feedback information corresponding to machine learning models that are trained based on candidate feature sets.

Abstract: There is disclosed a unit cell-based DAC configured to operate in N consecutive phases, wherein the unit cell-based DAC comprises a plurality of unit cells and wherein a single transition between a first state and a second state of the unit cell-based DAC is caused by M transitions of the plurality of unit cells between corresponding first states and second states, where M is smaller than N, wherein each transition of a unit cell is configured to occur during one of the N phases and wherein the interconnections of the unit cells are reconfigurable such that the unit cell-based DAC is configured to switchably operate in one of a PAM3 mode, in which the unit cell-based DAC converts a two-bit input signal into a three-level output signal, and a PAM4 mode in which the unit cell-based DAC converts the two-bit input signal into a four-level output signal.

Abstract: A system includes an input interface receiving conversation input from at least one system participant. A natural language processor operates on the conversational input. A sentiment processor operates on the pre-processed input using a sentiment classification model to predict sentiment, the sentiment processor predicting positive, neutral, and negative sentiment for the pre-processed input. An emotion processor receiving the pre-processed input from the natural language processor and a corresponding sentiment from the sentiment processor and operates to predict an emotion class and a strength of the emotion based on the corresponding sentiment. An output mechanism generates sentiment strength for the conversation input based on the emotion class and strength of the emotion.

Abstract: The technology generally relates to providing search results that include a recommendation to install an application that can provide content relevant to the search query. A search query for content submitted by a user device may be received by a search server. The search server may identify an indexed web resource that satisfies the search query. The identified indexed web resource may be identified as being accessible by an application. The search server may determine the application is not installed on the user device. Based on determining the application is not installed on the user device, provide a link to an installation location for the application to the user device in response to the search query.

Abstract: A method may include: generating, by a plurality of auditor computer programs and a data curator computer program, a first random string using a first distributed randomness protocol; sending the first random string to a quantum party computer program; executing, by the quantum party computer program, a certified randomness protocol with a quantum randomness source using the first random string; receiving, from the quantum randomness source, a first quantum randomness; generating, by the auditor computer programs and the data curator computer program, a second random string using a second distributed randomness protocol; extracting, by the data curator computer program and using a randomness extractor, a second quantum randomness from the first quantum randomness using the second random string, wherein the second quantum randomness comprises a near-uniformly random string; and executing, by the data curator computer program, a differential privacy protocol using the second quantum randomness as additive noise

Abstract: A method may include: selecting, by a plurality of classical parties, each of the classical parties using a classical party computer program, a distributed randomness protocol; generating, by the plurality of classical parties, a random string using the selected distributed randomness protocol; providing, by one of the classical parties, the random string to a quantum party, wherein the quantum party executes a quantum party computer program in communication with a quantum randomness source; executing, by the quantum party, a certified randomness protocol with the quantum randomness source using the random string as an input; receiving, by the quantum party, quantum randomness comprising a sequence of random bits from the quantum randomness source; and verifying, by the classical parties, that the random string was randomly selected, and that the quantum randomness is a valid output of the certified randomness protocol using the random string as input.

Abstract: A method may include: generating, by a plurality of auditor computer programs and a data curator computer program, a first random string using a first distributed randomness protocol; sending the first random string to a quantum party computer program; executing, by the quantum party computer program, a certified randomness protocol with a quantum randomness source using the first random string; receiving, from the quantum randomness source, a first quantum randomness; generating, by the auditor computer programs and the data curator computer program, a second random string using a second distributed randomness protocol; extracting, by the data curator computer program and using a randomness extractor, a second quantum randomness from the first quantum randomness using the second random string, wherein the second quantum randomness comprises a near-uniformly random string; and executing, by the data curator computer program, a differential privacy protocol using the second quantum randomness as additive noise

Abstract: Quantum computers with a limited number of input qubits are used to perform machine learning processes having a far greater number of trainable features. A list of features of a field are divided into a plurality of feature groups. Each of the feature groups includes a respective group of some, but not all, of the features. A first machine learning process is performed to train a first instance of a quantum computer model, where the feature groups are used as inputs. Based on the first machine learning process being performed, a subset of the feature groups is selected for a second machine learning process. Thereafter, the second machine learning process is performed to train one or more second instances of the quantum computer model. The individual features of the selected subset of the feature groups are used as inputs for the second instances of the quantum computer model.

Abstract: Systems and methods for approximation of Shapley values in memory-constrained environments are disclosed. According to an embodiment, a method may include: (1) receiving, by a training computer program executed on a training electronic device, a training dataset; (2) training, by the training computer program, a machine learning model on the training dataset; (3) deploying, by the training computer program, the machine learning model to a deployment electronic device; (4) receiving, by the deployment electronic device, an incoming data query; (5) generating, using the machine learning model, a prediction for the incoming data query; and (6) estimating, by a model explanation computer program, a Shapley value for the prediction.

Abstract: Systems and methods for construction and retraining of decision trees ensemble using hybrid classical-quantum algorithms are disclosed. A method may include a classical computer program: receiving a dataset; calculating feature-weights for original examples using a feature-weight calculation method; updating the feature-weights for the original examples with feature-weights for the original examples and the new examples; loading the feature-weights for the dataset and new data into a first quantum-accessible data structure and loading overwritten values into a second quantum-accessible data structure; instructing a quantum computer to query quantum states for the first and second quantum-accessible data structures using random sampling with replacement, to execute quantum-supervised clustering with the quantum states and the feature-weights, to grow a depth for the tree, and to calculate labels for a regression task and/or a classification task; and receiving the labels from the quantum computer.

Abstract: A schedule data structure includes a schedule of visits to a plurality of destinations over a plurality of weeks, the visits being associated with respective rider/vehicle identification (RVID) parameters corresponding to one of the plurality of riders/vehicles. The schedule may be optimized for multiple parameters but non-optimized for mirroring. The schedule data structure may be broken into a first part and a second part. A permutation of the first part may be determined and appended with the second part. A mirroring-improved schedule data structure may be determined in response to iterated permutation and appending.

Abstract: A computer-implemented method for training a neural network to forecast multivariate data involves inputting datasets from a forecast location and adjacent locations. These datasets include spatio-temporal characteristics and recorded multivariate data. The method determines the longest time-series sequence length with a high occurrence frequency in the dataset. It then trains a neural network for the forecast location based on this sequence length to encode the multivariate data into a vector. Similarly, it trains neural networks for adjacent locations to encode their data into vectors. These adjacent location vectors are combined into a single vector. The forecast location vector and the combined adjacent location vector are then composed into a final vector, which is decoded to generate a forecast for the forecast location.

Abstract: The technology generally relates to providing search results that include a recommendation to install an application that can provide content relevant to the search query. A search query for content submitted by a user device may be received by a search server. The search server may identify an indexed web resource that satisfies the search query. The identified indexed web resource may be identified as being accessible by an application. The search server may determine the application is not installed on the user device. Based on determining the application is not installed on the user device, provide a link to an installation location for the application to the user device in response to the search query.

Abstract: Systems and methods for decision tree construction and update using quantum algorithms are disclosed. A method may include: receiving, by a classical computer program, a dataset comprising a plurality of training examples, each of the plurality of training examples having a plurality of features; loading, by a classical computer program, the dataset into a quantum accessible data structure; providing, by the classical computer program, the quantum accessible data structure to a quantum computer, wherein the quantum computer is configured to perform quantum estimation of a Pearson correlation coefficient on the dataset in the quantum accessible data structure to create a weighted dataset; clustering, by the classical computer program and the quantum computer, each of the plurality of training examples in the weighted dataset into one of a plurality of clusters; and selecting, by the classical computer program, a label for each of the plurality of clusters.

Abstract: A computer-implemented method comprising predicting 5G usage data including generating at least one of first to third 5G usage data predictions, wherein generating the first 5G usage data prediction comprises using a third model to generate the first 5G usage data prediction based on predicted non-network data of a first intermediate prediction and predicted non-5G network usage data of a second intermediate prediction, wherein generating the second 5G usage data prediction comprises using the third model to generate the second 5G usage data prediction based on predicted non-network data of the first intermediate prediction and predicted non-5G network usage data of a third intermediate prediction, and wherein generating the third 5G usage data prediction comprises using the third model to generate the third 5G usage data prediction based on the predicted non-network data of the first intermediate prediction and predicted non-5G network usage data of a sixth intermediate prediction.

Abstract: The system includes a parsing module to receive one or more input constructs from a source database in a first syntactical dialect, parse the one or more input constructs received in the first syntactical dialect using an external domain specific language parser, converts the one or more input constructs upon parsing into an abstract syntax tree corresponding to the first syntactical dialect of the source database. A canonicalization module transforms the abstract syntax tree into a canonical abstract syntax tree by utilizing a library of a plurality of operators and expressions corresponding to the first syntactical dialect. A tree optimization module optimizes the canonical abstract syntax tree by utilizing a set of optimization rules. An input translation module converts the canonical abstract syntax tree into an output code corresponding to a second syntactical dialect associated with a target database for compatibility using a translation layer.

Abstract: A system and methods include a negative certificate authority for distributed management of negative certificates. An authorization restriction is associated with an untrusted user. A negative certificate generated for the untrusted user includes a public key associated with the untrusted user and an authorization restriction. The authorization restriction includes at least one global restriction, which is applicable to each consortium member that subscribes to the negative certificate. The authorization restriction includes at least one local restriction, which allows individual consortium members to further define their own locally applicable restrictions using the negative certificate authority. The negative certificate is accessible to each member of the consortium to enforce the authorization restriction against a transaction request. A secure contributor record including a unique cryptographically generated address is generated for each contributor.

Abstract: A system for data warehouse workload transformation is disclosed. A data extraction module receives job metadata from one or more extract transform load applications, extracts domain logic information from the job metadata. A canonical model generation module creates one or more relational models from the domain logic information, generates a canonical model for the job metadata by merging the one or more relational models. A model optimization module creates a query execution plan using an optimization technique, passes the canonical model through one or more optimization layers, and optimizes the query execution plan to execute common logic at a single point of time for removing one or more redundant operations. A code generation module generates an output code corresponding to the predefined target environment from an optimized canonical model, upon mapping of the one or more entities, for the workload transformation.

Abstract: A method for generating a dynamic interface to display real-time information about a plurality of tradeable assets within a portfolio. The method includes receiving real-time data for each of the plurality of tradeable assets and generating a node for each tradeable asset. The nodes include a first characteristic indicative of a volume relative to the portfolio and a second characteristic indicative of a trading momentum. The method includes generating the dynamic interface which includes positioning each node within a circular common area, with the position between a center of the circular common area and an outer edge of the circular common area, the position based on the first characteristic of each node. One or more nodes has an angular movement based on the second characteristic, and the dynamic interface updates with real-time changes in the data of any of the plurality of tradeable assets.