Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for generating images of a new subject using a diffusion neural network.

Abstract: A system for classifying touch data is provided. The system comprises: a plurality of sensing elements; and a processing system. The processing system is configured to: receive touch data from a current user via the plurality of sensing elements; determine a first set of classifier parameters corresponding to a current user based on usage data, wherein the usage data comprises the touch data from the current user, and apply the first set of classifier parameters to classify subsequent touch data.

Abstract: Methods, systems, and apparatuses, including computer programs encoded on computer storage media, for identifying relevant documents to a query by using only retrieved token vectors of candidate documents rather than all token vectors of the documents. That is, by using only retrieved token vectors of candidate documents rather than all token vectors of the documents, the described techniques dramatically increase the speed and accuracy of identifying relevant documents to a query.

Abstract: An example method for prompt-based query generation is provided. The method includes receiving, by a computing device, at least two prompts associated with a retrieval task to be performed on a corpus of documents associated with the task. The method includes applying, based on the at least two prompts and the corpus of documents, a large language model to generate a synthetic training dataset comprising a plurality of query-document pairs, wherein each query-document pair comprises a synthetically generated query and a document from the corpus of documents. The method includes training, on the plurality of query?document pairs from the synthetic training dataset, a document retrieval model to take an input query associated with the retrieval task and predict an output document retrieved from the corpus of documents. The method includes providing, by the computing device, the trained document retrieval model.

Abstract: Methods, systems, and apparatuses, including computer programs encoded on computer storage media, for identifying relevant documents to a query by using only retrieved token vectors of candidate documents rather than all token vectors of the documents. That is, by using only retrieved token vectors of candidate documents rather than all token vectors of the documents, the described techniques dramatically increase the speed and accuracy of identifying relevant documents to a query.

Abstract: A system for classifying touch data is provided. The system comprises: a plurality of sensing elements; and a processing system. The processing system is configured to: receive touch data from a current user via the plurality of sensing elements; determine a first set of classifier parameters corresponding to a current user based on usage data, wherein the usage data comprises the touch data from the current user, and apply the first set of classifier parameters to classify subsequent touch data.

Abstract: An example method includes providing, to a sequence model (i) a plurality of few-shot prompts, wherein each prompt comprises a demonstration passage, a demonstration task, and a demonstration query, wherein the demonstration task describes a type of retrieval, and wherein the demonstration query is relevant to the demonstration task, and (ii) a plurality of passages sampled from a corpus of passages. The method also includes receiving, from the sequence model and for the plurality of passages and based on the plurality of few-shot prompts, a respective plurality of predicted task-query pairs, the sequence model having been prompted to predict a task based on an input passage, and predict an output query relevant to the predicted task. The method further includes generating a synthetic training dataset comprising the plurality of passages and the respective plurality of predicted task-query pairs. The method also includes providing the synthetic training dataset.

Abstract: Provided are computing systems, methods, and platforms that train query processing models, such as large language models, to perform query intent classification tasks by using retrieval augmentation and multi-stage distillation. Unlabeled training examples of queries may be obtained, and a set of the training examples may be augmented with additional feature annotations to generate augmented training examples. A first query processing model may annotate the retrieval augmented queries to generate inferred labels for the augmented training examples. A second query processing model may be trained on the inferred labels, distilling the query processing model that was trained with retrieval augmentation into a non-retrieval augmented query processing model. The second query processing model may annotate the entire set of unlabeled training examples. Another stage of distillation may train a third query processing model using the entire set of unlabeled training examples without retrieval augmentation.

Abstract: Provided are computing systems, methods, and platforms that train query processing models, such as large language models, to perform query intent classification tasks by using retrieval augmentation and multi-stage distillation. Unlabeled training examples of queries may be obtained, and a set of the training examples may be augmented with additional feature annotations to generate augmented training examples. A first query processing model may annotate the retrieval augmented queries to generate inferred labels for the augmented training examples. A second query processing model may be trained on the inferred labels, distilling the query processing model that was trained with retrieval augmentation into a non-retrieval augmented query processing model. The second query processing model may annotate the entire set of unlabeled training examples. Another stage of distillation may train a third query processing model using the entire set of unlabeled training examples without retrieval augmentation.

Abstract: A pharmaceutical delivery device, comprising a cylindrical body formed from a plurality of concentrically arranged layers, each layer being formed from a biodegradable material and incorporating at least one active pharmaceutical agent. Optionally, the device comprises an outer layer, and inner layer and one or more intermediate layers, wherein at least one of the one or more intermediate layers is formed from a material having a greater rate of degradation that the inner and outer layers such that the inner and outer layers separate in use.

Abstract: Systems and methods for pre-training and fine-tuning of neural-network-based language models are disclosed in which a neural-network-based textual knowledge retriever is trained along with the language model. In some examples, the knowledge retriever obtains documents from an unlabeled pre-training corpus, generates its own training tasks, and learns to retrieve documents relevant to those tasks. In some examples, the knowledge retriever is further refined using supervised open-QA questions. The framework of the present technology provides models that can intelligently retrieve helpful information from a large unlabeled corpus, rather than requiring all potentially relevant information to be stored implicitly in the parameters of the neural network. This framework may thus reduce the storage space and complexity of the neural network, and also enable the model to more effectively handle new tasks that may be different than those on which it was pre-trained.

Abstract: A user input method includes the following steps. A virtual keyboard layout and a control region are displayed. The virtual keyboard layout includes a plurality of key subgroups each mapped to a respective one of a plurality of regions of the control region. Locations of an object from at least one captured image are extracted to identify a location of a feature point of the object. A target region in which the feature point is located is determined. Keys mapped to the target region are determined. Movements of the object are translated as input data to the user interface system.

Abstract: The subject disclosure is directed towards automatically labeling location-related information such as corresponding to GPS data or the like with a semantic label. A classifier trained with machine learning is provided with feature data corresponding to the location-related information and other features, such as user demographics data of a person associated with location-related information. The semantic label is received from the classifier, and associated with the location-related information. Other features may be used, such as other egocentric features corresponding to a person's particular visit, features from a sequence of visits, and/or features from other user information. The semantic label may be used to trigger an action, label a location on a map or the like, and so on.

Abstract: The subject disclosure is directed towards automatically labeling location-related information such as corresponding to GPS data or the like with a semantic label. A classifier trained with machine learning is provided with feature data corresponding to the location-related information and other features, such as user demographics data of a person associated with location-related information. The semantic label is received from the classifier, and associated with the location-related information. Other features may be used, such as other egocentric features corresponding to a person's particular visit, features from a sequence of visits, and/or features from other user information. The semantic label may be used to trigger an action, label a location on a map or the like, and so on.

Abstract: A user input method includes the following steps. A virtual keyboard layout and a control region are displayed. The virtual keyboard layout includes a plurality of key subgroups each mapped to a respective one of a plurality of regions of the control region. Locations of an object from at least one captured image are extracted to identify a location of a feature point of the object. A target region in which the feature point is located is determined. Keys mapped to the target region are determined. Movements of the object are translated as input data to the user interface system.

Abstract: A layered body comprising: a core region; at least one intermediate layer disposed around the core region; and an outer layer disposed around the at least one intermediate layer, wherein at least one of the at least one intermediate layers comprises a gas, the layered body having at least one dimension, measured across the body and through the core region, of 100 ?m or less.

Abstract: A capacitive ultrasonic transducer includes a first electrode, an insulating layer formed on the first electrode, at least one support frame formed on the insulating layer, and a second electrode formed space apart from the first electrode, wherein the first electrode and the second electrode define an effective area of oscillation of the capacitive ultrasonic transducer, and the respective length of the first electrode and the second electrode defining the effective area of oscillation is substantially the same.

Abstract: The invention is to provide a ring body and supporting structure of a vibratile gyroscope. The ring body is a thin sheet ring body having a height. The supporting structure is provided for supporting the ring body. The supporting structure is located on two opposing edges of the ring body. The supporting structures provide axial and radial supporting forces to restrain the ring body, providing better sensitivity and capability to resist environmental vibration and noise. Additionally, a reinforcing structure surrounding the ring body is arranged at an interior surface of the ring body to raise the rigidity of the ring body and maintain an elliptical resonance mode. If the reinforcing structure si arranged as high as the ring body, then it is possible to arrange electrodes at both inner and outer sides of the ring body to raise the effective area of driving and/or sensing electrodes.

Abstract: A capacitive ultrasonic device is capable of detecting an object and providing information regarding an orientation of the object. The capacitive ultrasonic device includes a power source configured to generate a voltage signal, an array of sensor elements, each of the sensor elements being configured to generate an ultrasonic wave during a first period of the voltage signal and detect whether a wave is reflected from the object during a second period of the voltage signal when the voltage signal is applied thereto, and a control unit configured to activate a first portion of the array of sensor elements through the power source, and activate a second portion of the array of sensor elements when at least one of the first portion of the sensor elements detects a wave reflected from the object.

Abstract: Techniques and systems are described that utilize a scalable, “light-weight” user model, which can be combined with a traditional global email spam filter, to determine whether an email message sent to a target user is a desired email. A global email model is trained with a set of email messages to detect desired emails, and a user email model is also trained to detect desired emails. Training the user email model may comprise one or more of: using labeled training emails; using target user-based information; and using information from the global email model. Global and user model scores for an email sent to a target user can be combined to produce an email score. The email score can be compared with a desired email threshold to determine whether the email message sent to the target user is desired or not.