Abstract: Embodiments of the present invention provide an approach for compressing data, and more particularly, to large-scale text data encoding and compression using absolute overfitting on pre-trained language models. Large-scale data is parsed into sentences. A unique token is generated for each sentence to form a token list. A generative (or compression) model is trained from the tokens in the token list to produce the corresponding sentence of each token through absolute overfitting of a pre-trained language model. The compressed text data is stored as the token list and generative model, resulting in a storage space savings.

Abstract: Mechanisms are provided for automatically predicting vehicle weight of a moving vehicle. Vehicle operation data is obtained from one or more on-board sensors/systems of the vehicle and features are extracted. The features are filtered based on a required working condition of the vehicle to identify intervals of features having valid feature data to thereby generate a filtered valid data. The filtered valid data is processed by a first artificial intelligence (AI) computer model based on time slices of the filtered valid data to generate a first prediction of vehicle weight, and by a second AI computer model based on buckets of key variables to generate a second prediction of vehicle weight. The first prediction is fused with the second prediction to generate a final prediction of vehicle weight which is output to downstream computing logic.

Abstract: From a set of point data, a set of scattered rays is constructed. From the set of scattered rays, a set of ray slopes is computed. The set of ray slopes is mapped to a corresponding set of trigonometric functions. Using an optimization method, a parameter of the set of trigonometric functions is selected. Using an inverse of the set trigonometric functions, a vehicle mass corresponding to the set of point data is computed. Based on the vehicle mass, a threshold braking distance of a collision avoidance system of the vehicle is adjusted, the threshold braking distance comprising a distance from an object predicted to collide with the vehicle. By braking the vehicle at least the threshold braking distance from the object, a predicted collision between the vehicle and the object is avoided.

Abstract: Systems and methods are provided for identifying and retrieving semantically similar images from a database. Semantic analysis is performed on an input query utilizing a vision language model to identify semantic concepts associated with the input query. A preliminary set of images is retrieved from the database for semantic concepts identified. Relevant concepts are extracted for images with a tokenizer by comparing images against a predefined label space to identify relevant concepts. A ranked list of relevant concepts is generated based on occurrence frequency within the set. The preliminary set of images is refined based on selecting specific relevant concepts from the ranked list by the user by combining the input query with the specific relevant concepts. Additional semantic analysis is iteratively performed to retrieve additional sets of images semantically similar to the combined input query and selection of the specific relevant concepts until a threshold condition is met.

Abstract: Systems and methods implemented via a broker in a cloud-based system include steps of, responsive to a user and associated user device executing a client connector being authenticated, receiving a notification from the client connector; determining private applications accessible by the user based on policy, wherein the private applications are located in one of a public cloud, a private cloud, and an enterprise network; and sending a Top-Level Domain+1 (TLD+1) list of the accessible private applications to the user device, wherein the TLD+1 includes a TLD and a domain name.

Abstract: Shutdown intervals between nodes of a dependency tree are determined. The dependency tree represents a system to be shut down and the nodes represent system components of the system. The determining the shutdown intervals includes calculating, for a pair of nodes of the dependency tree, a shutdown interval. The calculating is based on a relationship between the pair of nodes. The calculating is repeated for a plurality of pairs of nodes to obtain a plurality of shutdown intervals. The plurality of shutdown intervals is used in a shutdown of the system components. At least one system component represented by at least one node on a higher node level of the dependency tree is to be shut down prior to at least one other system component represented by at least one other node on a lower node level of the dependency tree.

Abstract: Information extraction and image restructuring includes generating semantic vectors to encode portions of text extracted from a document. For each semantic vector a semantic similarity between a schema key and other text encoded therein is determined based on their respective positions within the document. An enhanced NLP model is created using the semantic vectors, each labeled according to the semantic similarity. The text, including schema key, are re-encoded as a key and candidate vectors. Key-value pairs are generated by matching the key vector with a predetermined number of candidate vectors. The enhanced NLP model, using prompt learning, is repurposed to perform a next-sentence prediction that predicts which of the candidate vectors is logically related to the schema key. Based on the next-sentence prediction, the discrete portion of text identified as the schema key and portion of text determined to be logically related thereto are output.

Abstract: A rectifier bridge closed-loop control excitation method and an initial anode voltage establishment method are provided. In the method, an excitation function is periodically invoked to calculate a given value Ref, a rectifier bridge control angle is calculated, and a bridge rectifier is controlled; the whole excitation process is under closed-loop control, a terminal voltage of a generator rises quickly and smoothly, an excitation voltage is almost free of oscillation, overshoot is small, and adaptability to generator sets with different time constants is good; and when an initial anode voltage is established, excitation reliability is improved by combining an initial excitation power supply and a full conduction function of the bridge rectifier, and the full conduction control of the rectifier bridge is used as an emergency measure when the initial excitation power supply fails, so that the initial anode voltage is established with a relatively large probability.

Abstract: A light emitting diode (LED) pixel array and method of fabrication thereof. A semiconductor wafer template includes a dielectric layer formed over a lower n-type gallium nitride (n-GaN) layer. A first aperture and a second aperture are formed through the dielectric layer and extending to the lower n-GaN layer, the second aperture being narrower than the first aperture. A mesa is formed within the first aperture by successively forming a mesa n-GaN layer, a mesa MQW layer above the mesa n-GaN layer, and a mesa p-GaN layer above the mesa MQW layer. A pyramid having sidewalls is formed within the second aperture by successively forming a pyramidal n-GaN layer, a pyramidal MQW layer, and a pyramidal p-GaN layer. The mesa n-GaN layer, mesa MQW layer, and mesa p-GaN layer form a mesa LED. The pyramidal n-GaN layer, pyramidal MQW layer, and pyramidal p-GaN layer form a pyramid LED.

Abstract: A light emitting diode (LED) pixel array and method of fabrication thereof. A semiconductor wafer template includes a successively stacked first n-GaN layer, first MQW layer, p-GaN layer, and dielectric layer. A plurality of apertures is formed through the dielectric layer, extending to the p-GaN layer. A plurality of mesas is formed by forming, within each aperture, a second MQW layer and a second n-GaN layer above each second MQW layer. The second n-GaN layer and second MQW layer of each mesa form a respective mesa LED with the p-GaN layer. The first n-GaN layer and first MQW layer form a lower LED with the p-GaN layer.

Abstract: A light emitting diode (LED) pixel array and method of fabrication thereof. A semiconductor wafer template includes a successively stacked lower n-GaN layer, lower MQW layer, lower p-GaN layer, upper n-GaN layer, and dielectric layer. A plurality of apertures is formed through the dielectric layer, extending to the upper n-GaN layer. A plurality of mesas is formed by forming, within each aperture, a mesa n-GaN layer, a mesa MQW layer above each mesa n-GaN layer, and a mesa p-GaN layer above each mesa MQW layer. The mesa n-GaN layer, mesa MQW layer, and mesa p-GaN layer of each mesa form a respective mesa LED. The lower n-GaN layer, lower MQW layer, and lower p-GaN layer form a lower LED.

Abstract: An ion beam etching machine and a lower electrode structure thereof. The lower electrode structure comprises an electrode plate, a ring pressing mechanism, and a lifting mechanism, the ring pressing mechanism being configured to support a member to be etched, and the lifting mechanism being configured to drive the ring pressing mechanism to ascend and descend relative to the electrode plate; and same further comprises a position measurement mechanism for measuring a position to which the ring pressing mechanism descends relative to the electrode plate. The lifting mechanism is configured to stop driving the ring pressing mechanism to descend based on measurement information fed back by the position measurement mechanism. The arrangement of the lower electrode structure can avoid fragmenting of the member to be etched, and improve the product yield.

Abstract: Systems and processes for evaluating algorithms for aligning weakly-annotated data to recognized characters in a document are provided. In a method for evaluating an algorithm for aligning annotation data to recognized characters, strong annotations and weak-to-strong annotations, which are generated by applying a weak-to-strong annotation alignment algorithm, for a document are received and matched to generate respective pairs of matched annotations. For each pair of matched annotations, respective metrics are calculated including comparisons of aspects of the strong annotations to the weak-to-strong annotations. The respective metrics are aggregated, and an indication of the aggregated metrics are output to a graphical user interface or targeted application. Aggregated metrics determined for different weak-to-strong annotation alignment algorithms may be compared in order to select or adjust an algorithm to be used for Optical Character Recognition (OCR) operations.

Abstract: Methods of encoding and decoding for video data are described for encoding or decoding coefficients for a transform unit. In particular, the sign bits for the non-zero coefficients are encoded using sign bit hiding. Two or more sets of coefficients are defined for the transform unit and a sign bit may be hidden for each set, subject to satisfaction of a threshold test. The sets may correspond to coefficient groups that are otherwise used in multi-level significance map encoding and decoding.

Abstract: A Light Emitting Diode (LED) array precursor is provided. The LED array precursor comprises a substrate having a substrate surface, a first LED stack, a p++ layer, a n++ layer and a second LED stack. The first LED stack is provided on a first portion of the substrate surface. The first LED stack comprises a plurality of first Group III-nitride layers defining a first semiconductor junction configured to output light having a first wavelength wherein a n-type side of the first semiconductor junction is orientated towards the substrate surface. The p++ layer is provided on the first LED stack, the p++ layer comprising a Group III-nitride. The n++ layer has a first portion covering the p++ layer of the first LED stack and a second portion covering a second portion of the substrate surface, wherein a tunnel junction is formed at an interface between the n++ layer and the p++ layer, the n++ layer comprising a Group III-nitride.

Abstract: This invention relates to berberine alkaloids, formulations thereof, and their use in the prevention and/or treatment of infectious disease in animals, particularly Lawsonia infection. In particular, the invention relates to berberine alkaloids, formulations thereof, and their application as antimicrobial agents in the prevention and/or treatment of infectious disease including bacterial, viral, parasitic or fungal infections, and for improving feed conversion ratio in food-producing animals. Also described is the use of berberine alkaloids as feed preservatives, added to feed compositions, and used in combination with further agents suitable for treating infectious disease.

Abstract: A method for intelligently distributing messages across multiple devices and associated chat applications is disclosed. In one embodiment, such a method includes registering multiple devices and associated chat applications to receive intelligent distribution of messages. The method monitors, using an agent, a message copied from a first device of the multiple devices to a clipboard. The method automatically analyzes content of the message to determine which other devices and associated chat applications of the plurality are eligible to receive the message. The message is then distributed to other devices and associated chat applications that are deemed eligible to receive the message. In certain embodiments, the message originates from a first chat application of a first type and the message is distributed to a second chat application of a second type that is different from the first type. A corresponding system and computer program product are also disclosed.

Abstract: Systems and processes for aligning weakly-annotated data to recognized characters in a document are provided. In a method for aligning annotation data to recognized characters, annotation words and character recognition tokens are received, and a search algorithm is performed to align the annotation words to the tokens in a stepwise manner. At each step, an annotation word is aligned to one or more tokens, and a cost of each respective alignment is calculated. Once all annotation words are aligned, a full set of annotation word-token pairs corresponding to the annotation is selected based on a total cost of alignment for that set. A bounding box enclosing the tokens in the selected full set is generated and output to a target application.

Abstract: A given segment of computer code is obtained and modified to produce one or more inefficient versions of the given segment of computer code in comparison to the given segment of computer code. A code parse tree is generated for the given segment of computer code and each inefficient version of the given segment of computer code. Model embeddings are generated based on the generated code parse trees and a diffusion model is trained based on the generated model embeddings.

Abstract: A method, computer system, and program product facilitate identification of error image labels in training data. The method comprises: evenly dividing a training dataset into N subsets, where the training dataset includes M data items each comprising a pair of image and its original image label; training a prediction model to label images by respectively using each of the N subsets as training data to generate N respective trained prediction models; respectively using each of the N trained prediction models trained by using one of the N subsets as training data to label the images in other N?1 subsets of the N subsets to generate N?1 prediction labels for each of the M images in the training dataset. For each image in the M data items, whether the original image label of the image is a potential error image label is based on the N?1 prediction labels of the image.