Abstract: Implementations are described herein for improving unsupervised domain adaptation (UDA) by using improved adaptive teacher for object detection with cross-domain mix-up. In various implementations, cross-domain training of an object detection machine learning model may include: performing weak augmentation on images from a target domain DT to generate a first set of weakly augmented target domain images; perform strong augmentation on images from the source domain DS and images from the target domain DT to generate a second set of strongly augmented images; processing the second set of strongly augmented images to generate a third set of inter-domain mixes of the images from DS and DT; and jointly train the object detection machine learning model, as a student machine learning model, with a teacher machine learning model using the first and third sets.

Abstract: A method includes accessing document including sentences, document being associated with configuration flag indicating whether ABSA, SLSA, or both are to be performed; inputting the document into language model that generates chunks of token embeddings for the document; and, based on the configuration flag, performing at least one from among the ABSA and the SLSA by inputting the chunks of token embeddings into a multi-task model. When performing the SLSA, a part of token embeddings in each of the chunks is masked, and the masked token embeddings do not belong to a particular sentence on which the SLSA is performed.

Abstract: A computer-implemented method includes: accessing a plurality of datasets, where each dataset of the plurality of datasets includes training examples; selecting datasets that include the training examples in a source language and a target language; and sampling, based on a sampling weight that is determined for each of the selected datasets, the training examples from the selected datasets to generate the training batches; training an ML model for performing at least a first task using the training examples of the training batches, by interleavingly inputting the training batches to the ML model; and outputting the trained ML model configured to perform the at least the first task on input utterances provided in at least one among the source language and the target language. The sampling weight is determined for each of the selected datasets based on one or more attributes common to the training examples of the selected dataset.

Abstract: Techniques are provided for using context tags in named-entity recognition (NER) models. In one particular aspect, a method is provided that includes receiving an utterance, generating embeddings for words of the utterance, generating a regular expression and gazetteer feature vector for the utterance, generating a context tag distribution feature vector for the utterance, concatenating or interpolating the embeddings with the regular expression and gazetteer feature vector and the context tag distribution feature vector to generate a set of feature vectors, generating an encoded form of the utterance based on the set of feature vectors, generating log-probabilities based on the encoded form of the utterance, and identifying one or more constraints for the utterance.

Abstract: A system for generating an advanced surround view for a vehicle includes a plurality of cameras to generate image data. Sensors are configured to determine kinematic data of the vehicle. Kinematic data are also obtained by a visual odometry module and/or a deep-vision mechanism. A processor is configured to process image data captured by the plurality of cameras to generate a 360-degree surround view layer representative of a surrounding area. The processor is further configured to construct an improved bicycle kinematic model for processing kinematic data to generate an under-chassis layer representative of an area under the vehicle. The processor is further configured to generate a 3D vehicle model layer and overlay objects layer. Display screens display a GUI of combined scene of the 360-degree surround view layer, under-chassis layer, 3D vehicle model layer and overlay objects layer as would be viewed from a virtual camera viewpoint.

Abstract: A method to detect, identify and determine the position of laser points from infrared laser emitters in a firearm training simulation system. The method consists of two steps. This first step combines some common tools such as a camera with a filter switcher and edge and corner detecting algorithms in image processing to provide a fast automatic screen calibrating method. The second step is to detect laser point objects in each frame continuously captured by the camera, and then further process these objects to identify laser shots, filter out noisy objects, and eventually detect simulated shots from different laser emitters and determine the final mapped position of the shot.

Abstract: Systems and methods for prompt tuning can utilize previously-learned prompts for the initialization of tuning for prompts on different tasks that may differ from the task associated with the previously-learned prompt. The prompt being utilized for initialization can be a generic prompt and/or may be a prompt selected based on a determined similarity between two or more task embeddings.

Abstract: Implementations are described herein for performing joint optimization of multi-task learning of dense predictions (MT-DP) and hardware-aware neural architecture search (NAS). In various implementations, a set of tasks to be performed using a resource-constrained edge computing system may be determined. Based on a base multi-task dense-prediction (MT-DP) architecture template, the set of tasks, and a plurality of hardware-based constraints of a target edge computing system, a network architecture search (NAS) may be used to sample candidate MT-DP architecture(s) from a search space of neural network architecture components. Each sampled candidate MT-DP architecture may include a distinct assembly of sampled neural network architecture components applied to the base MT-DP architecture template. Image data may be processed using the candidate MT-DP architecture(s) to determine performance metrics. These performance metrics may be used to jointly train the MT-DP architecture(s) and/or the NAS.

Abstract: Methods of capturing carbon dioxide in a wellbore can include installing a sealing element in the wellbore. The sealing element swells in the presence of carbon dioxide and can be used for capturing the carbon. The sealing element can include a carbon-swelling material, such as a carbon-swelling polymers, metal-based materials, or combinations of elastomeric polymers and metal-based materials. The sealing element can also include combinations of different carbon-swelling materials, fillers or other compounds, and materials that are not carbon swellable. The sealing element can create a seal, form an anchor, or create a seal and form an anchor in the wellbore after swelling.

Abstract: Methods of capturing carbon dioxide in a wellbore can include installing a sealing element in the wellbore. The sealing element swells in the presence of carbon dioxide and can be used for capturing the carbon. The sealing element can include a carbon-swelling material, such as a carbon-swelling polymers, metal-based materials, or combinations of elastomeric polymers and metal-based materials. The sealing element can also include combinations of different carbon-swelling materials, fillers or other compounds, and materials that are not carbon swellable. The sealing element can create a seal, form an anchor, or create a seal and form an anchor in the wellbore after swelling.

Abstract: Methods of capturing carbon dioxide in a wellbore can include installing a sealing element in the wellbore. The sealing element swells in the presence of carbon dioxide and can be used for capturing the carbon. The sealing element can include a carbon-swelling material, such as a carbon-swelling polymers, metal-based materials, or combinations of elastomeric polymers and metal-based materials. The sealing element can also include combinations of different carbon-swelling materials, fillers or other compounds, and materials that are not carbon swellable. The sealing element can create a seal, form an anchor, or create a seal and form an anchor in the wellbore after swelling.

Abstract: Methods of capturing carbon dioxide in a wellbore can include installing a sealing element in the wellbore. The sealing element swells in the presence of carbon dioxide and can be used for capturing the carbon. The sealing element can include a carbon-swelling material, such as a carbon-swelling polymers, metal-based materials, or combinations of elastomeric polymers and metal-based materials. The sealing element can also include combinations of different carbon-swelling materials, fillers or other compounds, and materials that are not carbon swellable. The sealing element can create a seal, form an anchor, or create a seal and form an anchor in the wellbore after swelling.

Abstract: A mechanical cut-off apparatus of a high-voltage electric circuit includes: in a main electrical path a main mechanical switch; in a secondary electrical path, a secondary mechanical switch; a mechanical control configured such that, the secondary mechanical switch is brought to its mechanically open state after the main mechanical switch has been brought to its mechanically open state; the apparatus includes a transition dipole comprising a capacitance, the transition dipole arranged in series with the pair of secondary electrical contacts in the secondary electrical path, and in that the apparatus includes a controlled switch which, in an electrically closed state, creates inside the mechanical cut-off apparatus a bypass that short-circuits the capacitance of the transition dipole.

Abstract: Systems and methods can leverage task-specific unlabeled data to improve downstream performance in data-constrained scenarios. Given a target task, a first technique proposed herein, which can be referred to as task augmentation, uses unlabeled text from the target domain to synthesize a large amount of in-domain training data for an auxiliary task A second technique provides a self-training algorithm, where a model learns to improve itself using its predictions on unlabeled examples.

Abstract: A bottom rail for a covering of an architectural-structure covering is disclosed. In one example of an embodiment, the bottom rail includes a first or front member, and a second or rear member. The first and second members are arranged and configured to be coupled to each other with a bottom portion of a covering positioned in between the first and second members. Thus arranged, the first member is coupled to the second member, and the covering is securely coupled to the bottom rail. In addition, and/or alternatively, the covering may be received within the bottom rail along a tortuous pathway. In addition, and/or alternatively, the bottom rail may include a front profile or surface that substantially corresponds to a front profile of battens extending across a width of the covering. Moreover, the bottom rail may enclose the lowermost batten so that consistent spacing of the battens is maintained.

Abstract: A flow cassette has a housing with an inlet and an outlet and a passage therebetween. The flow cassette also has a temperature sensor disposed within the passage and configured to measure the temperature of a fluid flowing through the passage, a flow rate sensor disposed within the passage and configured to measure a flow rate of the fluid flowing through the passage, and a processor coupled to the temperature sensor and flow rate sensor. The processor is configured to accept measurements of temperature and flow rate from the temperature sensor and flow rate sensor, respectively, and provide a compensated flow rate.

Abstract: A flow sensor comprises a flow restriction disposed within a passage such that a fluid passing through the passage must pass through the flow restriction. The flow sensor also has an upstream pressure sensor coupled to the passage at a point upstream of the flow restriction and configured to measure and provide an upstream pressure of the fluid within the passage, a downstream pressure sensor coupled to the passage at a point downstream of the flow restriction and configured to measure and provide a downstream pressure of the fluid within the passage, and a temperature sensor coupled to the passage and configured to measure and provide a temperature of the fluid within the passage. The flow sensor also includes a flow sensor processor coupled to the upstream and downstream pressure sensors and the temperature sensor and configured to accept measurements therefrom and calculate a compensated flow rate based at least in part on the measured pressures and temperature.