Abstract: Sparsity-aware reconfiguration compute-in-memory (CIM) static random access memory (SRAM) systems are disclosed. In one aspect, a reconfigurable precision succession approximation register (SAR) analog-to-digital converter (ADC) that has the ability to form (n+m) bit precision using n-bit and m-bit sub-ADCs is provided. By controlling which sub-ADCs are used based on data sparsity, precision may be maintained as needed while providing a more energy efficient design.

Abstract: Disclosed are a multi-task training technique and resulting model for detecting distracted driving. In one embodiment, a method is disclosed comprising inputting a plurality of labeled examples into a multi-task network, the multi-task network comprising: a backbone network, the backbone network generating one or more feature vectors corresponding to each of the labeled examples, and a plurality of prediction heads coupled to the backbone network; minimizing a joint loss based on outputs of the plurality of prediction heads, the minimizing the joint loss causing a change in parameters of the backbone network; and storing a distraction classification model after minimizing the joint loss, the distraction classification model comprising the parameters of the backbone network and parameters of at least one of the prediction heads.

Abstract: Disclosed are methods, devices, and computer-readable media for detecting lanes and objects in image frames of a monocular camera. In one embodiment, a method is disclosed comprising receiving a plurality of images; identifying a horizon in the plurality of images by inputting the plurality of images into a deep learning (DL) model (either stored on a local device or via a network call); determining one or more camera parameters based on the horizon; and storing or using the camera parameters to initialize a camera.

Abstract: A system and method for training a neural network. In some embodiments the method includes training a full-sized network and a plurality of sub-networks, the training including performing a plurality of iterations of supervised co-training, the performing of each iteration including co-training the full-sized network and a subset of the sub-networks.

Abstract: A smart card including an upper metal layer and a lower metal layer disposed opposite each other; an internal cavity arranged between the upper and lower metal layers; and an RF antenna disposed in the internal cavity; in which the upper and lower metal layers each include a partially recessed area opposite the RF antenna, each partially recessed area including at least one opening through the thickness of the metal layer and each partially recessed area being partially covered by at least one metal protrusion extending from one edge of the partially recessed area so as to define the contour of the at least one opening.

Abstract: The modular intramedullary fixation device is a multi-piece bone fixation device for insertion into the medullary cavity of a bone. In one embodiment, the modular intramedullary fixation device is formed from a plurality of substantially Z-shaped modular elements, which are assembled together to a desired length and width. In additional embodiments, the modular intramedullary fixation device includes a proximal portion, a distal portion, and a connector for connecting the two portions to each other. The connector may be provided in a wide variety of configurations, including an adjustable connector for adjusting the length of the modular intramedullary fixation device. Each of the proximal and distal portions may further have an adjustable length.

Abstract: A system and method for handling processing with outliers. In some embodiments, the method includes: reading a first activation and a second activation, each including a least significant part and a most significant part, multiplying a first weight and a second weight by the respective activations, the multiplying of the first weight by the first activation including multiplying the first weight by the least significant part of the first activation in a first multiplier, the multiplying of the second weight by the second activation including: multiplying the second weight by the least significant part of the second activation in a second multiplier, and multiplying the second weight by the most significant part of the second activation in a shared multiplier, the shared multiplier being associated with a plurality of rows of an array of activations.

Abstract: A neural network accelerator is disclosed that includes a multiplication unit, an adder-tree unit and an accumulator unit. The multiplication unit and the adder tree unit are configured to perform lattice-multiplication operations. The accumulator unit is coupled to an output of the adder tree to form dot-product values from the lattice-multiplication operations performed by the multiplication unit and the adder tree unit. The multiplication unit includes n multiplier units that perform lattice-multiplication-based operations and output product values. Each multiplier unit includes a plurality of multipliers. Each multiplier unit receives first and second multiplicands that each include a most significant nibble (MSN) and a least significant nibble (LSN). The multipliers in each multiplier unit receive different combinations of the MSNs and the LSNs of the multiplicands. The multiplication unit and the adder can provide mixed-precision dot-product computations.

Abstract: Disclosed are a multi-task training technique and resulting model for detecting distracted driving. In one embodiment, a method is disclosed comprising inputting a plurality of labeled examples into a multi-task network, the multi-task network comprising: a backbone network, the backbone network generating one or more feature vectors corresponding to each of the labeled examples, and a plurality of prediction heads coupled to the backbone network; minimizing a joint loss based on outputs of the plurality of prediction heads, the minimizing the joint loss causing a change in parameters of the backbone network; and storing a distraction classification model after minimizing the joint loss, the distraction classification model comprising the parameters of the backbone network and parameters of at least one of the prediction heads.

Abstract: Disclosed are a multi-task training technique and resulting model for detecting distracted driving. In one embodiment, a method is disclosed comprising inputting a plurality of labeled examples into a multi-task network, the multi-task network comprising: a backbone network, the backbone network generating one or more feature vectors corresponding to each of the labeled examples, and a plurality of prediction heads coupled to the backbone network; minimizing a joint loss based on outputs of the plurality of prediction heads, the minimizing the joint loss causing a change in parameters of the backbone network; and storing a distraction classification model after minimizing the joint loss, the distraction classification model comprising the parameters of the backbone network and parameters of at least one of the prediction heads.

Abstract: Aspects of the disclosure provide a method of a LiDAR system to determine the distance of an object from the LiDAR system. Embodiments of the LiDAR system can use a coarse estimate of the distance of an object from the LiDAR system which is then used by a fast search to determine an estimate of the distance of an object from the LiDAR system within a precision threshold. In some embodiments the LiDAR system can use adaptive precision when determining the distance of an object from the LiDAR system.

Abstract: A device, method, and non-transitory computer readable medium for using a smartphone clustering application.

Abstract: A system includes an igniter rod having a ground rod with a first tip bent at a 90-degree angle, an internal rod with a second tip, a rod cover enclosing a portion of the internal rod, and a base tube, fixed to an external surface of the rod cover, enclosing a portion of the ground rod. The base tube has a plurality of adjustment screws that hold in place and position the ground rod. The igniter system further includes an igniter body configured to enclose a portion of the igniter rod and a fuel supply. The fuel supply is configured to be brought into contact with a gap between the first tip and the second tip.

Abstract: An information handling system for servicing search requests for information stored in data assets includes storage for storing a graph database based on at least one data catalog for the data assets. The information handling system also includes a processor that obtains, from a requestor, a search request of the search requests, the search request specifies a keyword search term; obtains a first identifier of a data asset of the data assets using the graph database and the keyword search term; and provides the first identifier as a supplemental result for the search request.

Abstract: Embodiments of zeolite composite catalysts and methods of producing the zeolite composite catalysts are provided, where the methods comprise dissolving in an alkaline solution a catalyst precursor comprising at least one mesoporous zeolite while heating, stirring, or both to yield a dissolved zeolite solution, where the mesoporous zeolite has a molar ratio of SiO2/Al2O3 of at least 30, where the mesoporous zeolite comprises zeolite beta, adjusting the pH of the dissolved zeolite solution, aging the pH adjusted dissolved zeolite solution to yield solid zeolite composite from the dissolved zeolite solution, and calcining the solid zeolite composite to produce the zeolite composite catalyst, where the zeolite composite catalyst has a mesostructure comprising at least one disordered mesophase and at least one ordered mesophase, and where the zeolite composite catalyst has a surface area defined by the Brunauer-Emmett-Teller (BET) analysis of at least 600 m2/g.

Abstract: A cleaning system and method use an ultrasound probe, a coupling mechanism, and a controller to clean equipment of a vehicle system. The ultrasound probe enters into an engine. The ultrasound probe emits ultrasound pulses and the coupling mechanism provides an ultrasound coupling medium between the ultrasound probe and one or more components of the engine. The controller drives the ultrasound probe to deliver the ultrasound pulse through the coupling medium to a surface of the one or more components of the engine. The ultrasound probe delivers the ultrasound pulse to remove deposits from the one or more components of the engine.

Abstract: It describes an intelligent environment management process and system based on Internet of Things (IoT) technology, Cloud Computing, and the application of Artificial Intelligence (AI) and Machine Learning (ML) techniques aimed at providing an environment adapted to its real use in a customized way, which will bring countless benefits to users with a more intelligent use of the environment's features based on actions and suggestions from data collection and pre-processing; Processing and Application of Artificial Intelligence (AI) Strategies in the Cloud and User Interface. The invention thus provides intelligent, adaptive management for greater comfort, in addition to automatically measuring, analyzing, and acting on the environments' comfort and efficiency based on the preferences of their users and managers.

Abstract: A method for the spatially resolved detection of an object outside a transportation vehicle with the aid of a sensor installed in a transportation vehicle wherein the viewing direction of a transportation vehicle occupant is captured. The captured viewing direction is analyzed over a period and a region of interest outside the transportation vehicle is determined. The sensor is then spatially sensitized and/or oriented in the direction of the determined region of interest.

Abstract: The disclosed systems, structures, and methods are directed to a LiDAR system comprising a radiation source configured to emit light pulses towards a region of interest (ROI), a detector configured to detect light pulses reflected from the ROI, a processor, communicatively coupled to the radiation source and the detector, configured to cause the radiation source to emit a preamble light pulse having an energy EP and a pulse width W1 towards the ROI, determine, if the preamble light pulse is detected by the detector and whether there is an object in the ROI, responsive to a determination that there is an object in the ROI, cause the radiation source to emit a scanning light pulse having an energy EL and a pulse width W2 towards the ROI, else the radiation source to emit a scanning light pulse having an energy EH and the pulse width W2 towards the ROI.

Abstract: There is provided a method and apparatus for detecting an object, especially an object remotely placed in a field of view (FOV) using optical pulses with non-uniform pulse power, without exceeding the Accessible Emitted Limit (AEL) A plurality of optical pulses including two or more levels of pulse power may be emitted to detect an object in the FOV. Information related to the object may be generated from the returning optical pulses. The optical pulses with non-uniform pulse power may result in increased probability of detection associated with the pulses returning from the remotely located object, and therefore the density of point clouds associated with the remotely located object may be increased.