Abstract: Generally discussed herein are devices, systems, and methods for encoding input-output examples. A method of generating a program using an encoding of input-output examples, may include processing an input example of the input-output examples, using a first long short term memory (LSTM) neural network, one character at a time to produce an input feature vector, processing an output example associated with the input example in the input-output examples, using the LSTM neural network, one character at a time to produce an output feature vector, determining (a) a cross-correlation between the input feature vector and the output feature vector or (b) previously computed feature vectors for a different input-output example that are sufficiently close to the input feature vector and the output feature vector, respectively, and using the determined cross-correlation or previously computed vector, generating a program consistent with the input example and the output example.

Abstract: According to an exemplary embodiment of the present disclosure, disclosed is a computer program stored in a computer readable storage medium. When the computer program is executed in one or more processors, the computer program performs the following method for anomaly detection of data using a network function, and the method includes: generating an anomaly detection model including a plurality of anomaly detection sub models including a trained network function using a plurality of training data sub sets included in the training data set; calculating input data using at least one of the plurality of generated anomaly detection sub models; and determining whether there is an anomaly in the input data based on output data for input data of at least one of the plurality of generated anomaly detection sub models and the input data.

Abstract: Architecture introduces a new pattern operator referred to as called an augmented transition network (ATN), which is a streaming adaptation of non-reentrant, fixed-state ATNs for dynamic patterns. Additional user-defined information is associated with automaton states and is accessible to transitions during execution. ATNs are created that directly model complex pattern continuous queries with arbitrary cycles in a transition graph. The architecture can express the desire to ignore some events during pattern detection, and can also detect the absence of data as part of a pattern. The architecture facilitates efficient support for negation, ignorable events, and state cleanup based on predicate punctuations.

Abstract: Methods, systems and apparatuses, including computer programs encoded on computer storage media, are provided for learning or optimizing an indirect encoding of a mapping from digitally-encoded input arrays to digitally-encoded output arrays, with numerous technical advantages in terms of efficiency and effectiveness.

Abstract: A neural network unit includes a register programmable with a control value, a plurality of neural processing units (NPU), and a plurality of activation function units (AFU). Each NPU includes an arithmetic logic unit (ALU) that performs arithmetic and logical operations on a sequence of operands to generate a sequence of results and an accumulator into which the ALU accumulates the sequence of results as an accumulated value. Each AFU includes a first module that performs a first function on the accumulated value to generate a first output, a second module that performs a second function on the accumulated value to generate a second output, the first function is distinct from the second function, and a multiplexer that receives the first and second outputs and selects one of the two outputs based on the control value programmed into the register.

Abstract: An answer to a question may selected from answers from a set of answering pipelines. Question answer data can be generated for a question, using a first answering pipeline. Another set of question answer data can be generated for the second question, using the second answering pipeline. The question answer data can include answers and confidence values for each answer. Using a weighting formula and a blending profile for the first answering pipeline, a vote weight can be determined for an answer with the highest confidence value. The same weighting formula and a second blending profile may be used to determine a vote weight for another answer with the highest confidence value. An answer to the question may be selected from the answers, based on the overall highest vote weight.

Abstract: Disclosed are systems and methods that implement efficient engines for computation-intensive tasks such as neural network deployment. Various embodiments of the invention provide for high-throughput batching that increases throughput of streaming data in high-traffic applications, such as real-time speech transcription. In embodiments, throughput is increased by dynamically assembling into batches and processing together user requests that randomly arrive at unknown timing such that not all the data is present at once at the time of batching. Some embodiments allow for performing steaming classification using pre-processing. The gains in performance allow for more efficient use of a compute engine and drastically reduce the cost of deploying large neural networks at scale, while meeting strict application requirements and adding relatively little computational latency so as to maintain a satisfactory application experience.

Abstract: Systems, methods, and non-transitory computer-readable media can determine a respective embedding for each entity in a set of entities that are accessible through the social networking system, wherein each embedding is learned based at least in part on one or more sessions of connections formed between users and entities of the social networking system. One or more candidate entities that are related to a first entity can be determined based at least in part on the respective embeddings for the candidate entities and the first entity. At least a first candidate entity from the one or more candidate entities can be provided as a recommendation to a user that formed a connection with the first entity.

Abstract: Described is a neuromorphic system implemented in hardware that implements neuron membrane potential update based on the leaky integrate and fire (LIF) model. The system further models synapse weights update based on the spike time-dependent plasticity (STDP) model. The system includes an artificial neural network in which the update scheme of neuron membrane potential and synapse weight are effectively defined and implemented.

Abstract: Methods, systems, and computer-storage media are provided for generating and populating a feature catalog for use in predictive modeling applications. The feature catalog may be populated with features extracted from data warehoused in a nested, hierarchical data structure. Extraction of features may result from applying a set of rules-based algorithms to warehoused data. Updated features may be extracted from a queue of data updates by applying the same rules-based algorithms. Predictions of future outcomes may be generated by applying predictive models to features.

Abstract: An online system generates one or more models that determine a likelihood of a user interacting with an application over a particular time interval after installing the application. To generate the one or more models, the online system obtains information describing a user's interaction with the application that occurred greater than a threshold time period prior to a time for which user interaction with the application is to be determined. Example user interactions with the application include: usage of the application, numbers of particular interactions with the application, an amount of compensation the application receives from the user, interactions with other users of the application via the application, and any other suitable interactions. Various engagement metrics may be predicted by the one or more models such as an amount of time spent using the application, particular actions taken in the application, and revenue generated by the user in the application.

Abstract: A voice user interface (VUI) system use collaborative filtering to expand its own knowledge base. The system is designed to improve the accuracy and performance of the Natural Language Understanding (NLU) processing that underlies VUIs. The system leverages the knowledge of system users to crowdsource new information.

Abstract: This invention relates to an apparatus, system, and method for computing with neuromorphic circuit architectures that have neurons with interconnected internal state information. The interconnected internal state information allows the neurons to enable or strengthen the input to other neurons. Furthermore, neuron internal state information provides insights on the characteristics of the input data that can be used to enhance the performance of the neuromorphic system. The neuromorphic system can be implemented with an artificial phase-change-based neurons.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for training and applying a machine learning model. One of the methods includes the actions of obtaining a collection of training data, the training data comprising collection of data points associated with a labeled set of real property parcels; training a machine learning model using the training data, the machine learning model being trained to generate a likelihood with respect to a parameter from input data associated with a specific parcel of real property, wherein training includes optimizing the model using a Markov chain optimization that seeks to minimize error in the model where the model is underpinned by one or more non-differentiable functions; receiving a plurality of data points associated with an input parcel of real property; and using the optimized model to generate a likelihood for the parameter for the input parcel of real property.

Abstract: Systems, methods, and non-transitory computer readable media are configured to determine a feature set for a model to be trained by machine learning. A subset of features from the feature set can be associated with entities having relationship types and corresponding to pages on a social networking system. The feature set can be reduced based on at least one rule applied to the relationship types.

Abstract: Systems and methods for recognizing an analogous case from a knowledge base including a plurality of candidate cases are disclosed. The system may include a data receiver to receive a representation of a target case, and a microprocessor coupled to the data receiver to generate a plurality of target attributes from the representation of the target case. The microprocessor may query the knowledge base to recognize the analogous case similar to the target case in information content using the target attributes and a plurality of attribute memories each including first and second attributes and an inter-attribute contextual representation of interactions between the first and second attributes in a context of the plurality of candidate cases. The system may include an output circuit to output the recognized analogous case to a user or a process.

Abstract: A method for implementing a convolutional neural network (CNN) accelerator on a target includes identifying characteristics and parameters for the CNN accelerator. Resources on the target are identified. A design for the CNN accelerator is generated in response to the characteristics and parameters of the CNN accelerator and the resources on the target.

Abstract: A computerized efficient machine learning method for classification of data and new class discovery inputs labeled data and unlabeled data into a computer memory for a computerized machine tool to perform (a) initial supervised learning using the labeled data to generate a classifier, (b) semi-supervised learning using the labeled data, the classifier and the unlabeled data to generate an updated classifier and high confidence data, (c) active learning using the updated classifier and the unlabeled data to generate a data label request and receive new class labeled data to generate augmented labeled data, (d) new class discovery using the updated classifier and the data label request to generate data of potential new classes and receive labels for potential new class data to generate new class labeled data, and (e) supervised learning using the high confidence data, the labeled data and the augmented labeled data to generate an output classifier.

Abstract: A system and computer-readable medium for performing a cognitive learning operation comprising monitoring a user interaction of a user; generating user interaction data based upon the user interaction; receiving data from a plurality of data sources; processing the user interaction data and the data from the plurality of data sources to perform a cognitive learning operation, the processing being performed via a cognitive inference and learning system, the cognitive learning operation comprising analyzing the user interaction data, the cognitive learning operation generating a cognitive learning result based upon the user interaction data; and, associating a cognitive profile with the user based the cognitive learning result.

Abstract: A cognitive learning method comprising: monitoring a user interaction of a user; generating user interaction data based upon the user interaction; receiving data from a plurality of data sources; processing the user interaction data and the data from the plurality of data sources to perform a cognitive learning operation, the processing being performed via a cognitive inference and learning system, the cognitive learning operation comprising analyzing the user interaction data, the cognitive learning operation generating a cognitive learning result based upon the user interaction data; and, associating a cognitive profile with the user based the cognitive learning result.