Abstract: Systems and methods for managing conversations between a robot computing device and a user are disclosed. Exemplary implementations may: initiate a first-time user experience sequence with the user; teach the user the robot computing capabilities and/or characteristics; initiate, utilizing a dialog manager, a conversation with the user; receive, one or more command files from the user via one or more microphones; and generate conversation response files and communicating the generated conversation files to the dialog manager in response to the one or more received user global command files to initiate an initial conversation exchange.

Abstract: Devices and techniques are generally described for nearest device arbitration. In various examples, a first device may receive first audio data representing a wakeword spoken by a first speaker at a first time. In some examples, a second device may receive second audio data representing the wakeword spoken by the first speaker at the first time. In some cases, the first device may generate first feature data representing the first audio data and the second device may generate second feature data representing the second audio data. In various examples, a machine learning model may use the first feature data and the second feature data to generate first prediction data representing a prediction that the first device is closer to the first speaker than the second device.

Abstract: Dynamically adjusting, using artificial intelligence (AI), sensors and models of an autonomous roaming robotic device, which includes receiving data regarding an asset at a computer of a roaming robotic device from sensors on the robotic device. The robotic device identifies an asset at a location using the sensors, and the robotic device has instructions, received from a control system, to inspect the location or items at the location. The data is analyzed using the computer of the robotic device, and the analysis includes using historical data for the asset. An AI model is loaded using the computer of the robotic device, based on the identification of the asset. A sensor is selected using the computer of the robotic device, for conducting an inspection of the asset based on the analysis of the data and the AI model.

Abstract: This disclosure describes a machine learning system that includes a contrastive learning based two-tower model for retrieval of relevant chemical reaction procedures given a query chemical reaction. The two-tower model uses attention-based transformers and neural networks to convert tokenized representations of chemical reactions and chemical reaction procedures to embeddings in a shared embedding space. Each tower can include a transformer network, a pooling layer, a normalization layer, and a neural network. The model is trained with labeled data pairs that include a chemical reaction and the text of a chemical reaction procedure for that chemical reaction. New queries can locate chemical reaction procedures for performing a given chemical reaction as well as procedures for similar chemical reactions. The architecture and training of the model make it possible to perform semantic matching based on chemical structures. The model is highly accurate providing an average recall at K=5 of 95.9%.

Abstract: The objective of the present invention is to realize an operating state classification system having a classification accuracy that continuously improves through additional learning, and which is appropriately protected from unauthorized duplication of a classification function.

Abstract: A method for analyzing spectra comprises identifying a set of sample peaks in a sample spectrum, where the sample peaks are associated with fragments of a sample, each having a sample fragment mass. A reference spectrum is selected with one or more reference peaks corresponding to fragments of a reference, each having a reference fragment mass. A mass difference can be determined between selected sample and reference peaks, and a group exchange can be selected based on the mass difference; e.g., where the group exchange represents a change in the sample or reference fragment masses associated with the selected peaks. The selected peaks can be shifted by the mass difference, and a fit value can be determined with respect to the reference spectrum. The fit value characterizes similarity between the respective sets of sample and reference peaks, responsive to the group exchange and corresponding peak shift.

Abstract: A text processing model training method, and a text processing method and apparatus in the natural language processing field in the artificial intelligence field are disclosed. The training method includes: obtaining training text; separately inputting the training text into a teacher model and a student model to obtain sample data output by the teacher model and prediction data output by the student model; the sample data includes a sample semantic feature and a sample label; the prediction data includes a prediction semantic feature and a prediction label; and the teacher model is a pre-trained language model used for text classification; and training a model parameter of the student model based on the sample data and the prediction data, to obtain a target student model. The method enables the student model to effectively perform knowledge transfer, thereby improving accuracy of a text processing result of the student model.

Abstract: An automotive assistant that is connected to microphones and loudspeakers that are associated with different seats in a passenger vehicle includes a dialog manager that is configured to initiate a dialog based on an utterance received at a first one of the microphones and to advance that dialog based on an utterance received from another of the microphones.

Abstract: A remote distress monitor includes a steerable microphone array, a memory, and a control system. The steerable microphone array is configured to detect audio data and generate sound data. The memory stores machine-readable instructions. The control system includes one or more processors configured to execute the machine-readable instructions. The generated sound data from the steerable microphone array is analyzed. Based at least in part on the analysis, a token associated with the audio data detected by the steerable microphone array is generated. The audio data is representative of one or more sounds associated with a distress event. The generated token is transmitted, via a communications network, to an electronic device to cause a distress response action to occur. The distress response action is associated with the distress event.

Abstract: A temperature estimation apparatus includes an estimation unit configured to successively estimate temperature data by successively inputting given time series process data relating to conditions inside a processing space in which plasma processing is performed, into a time series model generated in advance that correlates data values, in each time period, of time series process data relating to conditions inside the processing space, with a data value, at a respective time point, of time series temperature data measured inside the processing space.

Abstract: A radio frequency connection between a far field voice detection device and a further device is used to determine a first angular direction from the far field voice detection device to the further device. The determined first angular direction is then used to emphasize, during a noise processing of a plurality of sounds received via use of a plurality of microphones of the far field voice detection device, a first one of the plurality of sounds relative to a remainder of the plurality of sounds.

Abstract: A sensor system can include a sensor configured to output raw sensor data, a plurality of processing modules configured to process the raw sensor data from the sensor to output processed sensor data, a state module for each processing module operative to cause a respective processing module to receive and/or process the sensor data, and a control module configured to receive a coded bitstring and activate or deactivate a predefined set of state modules based on the coded bitstring to control which processing modules process the sensor data and/or an order of processing. The coded bitstring can include a plurality of discrete bits less than the amount of processing modules and/or state modules. A plurality of the state modules and/or processing modules can be associated with at least one discrete bit of the plurality of discrete bits such that each coded bitstring corresponds to a predetermined group of state modules and/or processing modules.

Abstract: A method for detection of a keyword in a continuous stream of audio signal, by using a dilated convolutional neural network, implemented by one or more computers embedded on a device, the dilated convolutional network comprising a plurality of dilation layers, including an input layer and an output layer, each layer of the plurality of dilation layers comprising gated activation units, and skip-connections to the output layer, the dilated convolutional network being configured to generate an output detection signal when a predetermined keyword is present in the continuous stream of audio signal, the generation of the output detection signal being based on a sequence of successive measurements provided to the input layer, each successive measurement of the sequence being measured on a corresponding frame from a sequence of successive frames extracted from the continuous stream of audio signal, at a plurality of successive time steps.

Abstract: A method and apparatus are provided for sensor fault prediction. A time-sequence of output values is received from a sensor. A plurality of features are extracted from the received values. A model is applied to the features to obtain a health score for the sensor. The trend of the health score over time is calculated to detect degrading performance of the sensor, and a time at which the sensor will become faulty is predicted.

Abstract: Methods and systems are described for detecting a wide variety of fault conditions in a HVAC system based on analysis of current and historical energy consumption patterns, and on comparison with energy consumption patterns of other similarly situated RTUs. Energy consumption comparisons are preferably made in regard to a normalized load ratio or NLR or more preferably to a daily maximum normalized load ratio or MDNLR, which provide more robust and reliable bases for comparison of faulty and fault-free RTUs, and hence for generalized fault detection, than other previously known metrics or criteria.

Abstract: To automatically determine a more memorable macro name. Provided is an information processing device that comprises an utterance learning adaptation unit that executes clustering pertaining to a plurality of function execution instructions by a user and estimates, as a macro, a cluster that includes the plurality of function execution instructions and a response control unit that controls the presentation of information pertaining to the macro, wherein the utterance learning adaptation unit determines a name for the estimated macro on the basis of a context acquired at the time of issuing the plurality of function execution instructions included in the cluster, the response control unit controls a notification of the macro name to the user, and the plurality of function execution instructions include at least one function execution instruction issued via an utterance.

Abstract: A method for speech enhancement, the method may include receiving or generating sound samples that represent sound signals that were received during a given time period by an array of microphones; frequency transforming the sound samples to provide frequency-transformed samples; clustering the frequency-transformed samples to speakers to provide speaker related clusters, wherein the clustering is based on (i) spatial cues related to the received sound signals and (ii) acoustic cues related to the speakers; determining a relative transfer function for each speaker of the speakers to provide speakers related relative transfer functions; applying a multiple multiple output (MIMO) beamforming operation on the speakers related relative transfer functions to provide beamformed signals; and inverse-frequency transforming the beamformed signals to provide speech signals.

Abstract: In an exemplary technique, audio information responsive to received input is provided. While providing the audio information, one or more conditions for stopping the provision of audio information are detected, and in response, the provision of the audio information is stopped. After stopping the provision of the audio information, if the one or more conditions for stopping the provision of audio information have ceased, then resumed audio information is provided, where the resumed audio information includes a rephrased version of a previously provided segment of the audio information.

Abstract: Apparatus and methods for a pattern identification transformer neural network is provided. The pattern identification transformer neural network may be able to learn from relatively small numbers of data elements. The pattern identification transformer neural network may function in similar method to the way humans transform data points. As such, the pattern identification transformer neural network may be able to learn patterns from a small number of examples and determine what attributes are helpful from a single experience. The pattern identification transformer neural networks may include a multi-head attention module, a normalize module and a feed forward neural network. The multi-head attention module may receive vectors that correspond to experiences. The normalize module may normalize the received vectors. The feed forward neural network may incorporate the received vectors into the neural network.

Abstract: A method for performing data recovering operation is provided. The method includes: identifying a plurality of first values and a plurality of first indexes of a plurality of first entries of an incomplete matrix in a received object data, and one or more second values and one or more second indexes of one or more second entries of the incomplete matrix; inputting the first values, the first indexes, a preset first parameter, a preset second parameter and a preset third parameter into an analysis model using Adaptive Rank-One Matrix Completion (AROMC) algorithm; and obtaining a recovered complete matrix corresponding to the incomplete matrix from the analysis model, so as to obtain optimized one or more second values of the second entries, wherein the optimized one or more second values are determined as original values of the second entries, such that the incomplete matrix is recovered to the recovered complete matrix.