Abstract: Various embodiments relate to a method for detecting anomalies in a system by an anomaly detector, including: receiving a trained machine learning model that detects anomalies; receiving a set of new inputs from the to the anomaly detector from the system; setting a label for each of the set of new inputs to a value indicating normal operation of the system; training a new anomaly detection model using incremental learning to update the trained machine learning model using the labeled set of new inputs; receiving a set of past model inputs with an associated label; producing a verification set by inputting the set of past model inputs into the new anomaly detection model; and comparing the verification set with the labelled past model inputs to determine if an anomaly is present.

Abstract: A method and data processing system for making a machine learning model more resistant to adversarial examples are provided. In the method, an input for a machine learning model is provided. A randomly generated mask is added to the input to produce a modified input. The modified input is provided to the machine learning model. The randomly generated mask negates the effect of a perturbation added to the input for causing the input to be an adversarial example. The method may be implemented using the data processing system.

Abstract: A method is provided for analyzing a classification in a machine learning model (ML). In the method, the ML model is trained using a training dataset to produce a trained ML model. One or more samples are provided to the trained ML model to produce one or more prediction classifications. A gradient is determined for the one of more samples at a predetermined layer of the trained ML model. The one or more gradients and the one or more prediction classifications for each sample are stored. Also, an intermediate value of the ML model may be stored. Then, a sample is chosen to analyze. A gradient of the sample is determined if the gradient was not already determined when the at least one gradient is determined. Using the at least one gradient, and one or more of a data structure, a predetermined metric, and an intermediate value, the k nearest neighbors to the sample are determined. A report comprising the sample and the k nearest neighbors may be provided for analysis.

Abstract: A method is provided for protecting a software program from copying. The method includes providing a first implementation of the software program. A second implementation of the software program is then provided. The second implementation provides a same functionality as the first implementation, and wherein the second implementation includes a plurality of dummy operations to increase a number of operations and an execution time of the second implementation compared to the first implementation. The dummy operations are encoded. The second implementation may then be compared to another software program to determine if the another software program is a copy of the first implementation of the software program. This allows a copy of the first implementation to be detected without disclosing the first implementation.

Abstract: A device (100) includes a substrate (101-106) with an upper semiconductor layer, buried semiconductor layer, and a DTI structure (107-108) defining an active device region; a dummy LDMOS device (121) in the active device region which includes a grounded drain (D1) in a drift region (105), a source (S1, S2) in a body region (109) which extends to contact the buried semiconductor layer, a gate electrode (G1-G4) formed so that the source and at least part of the gate electrode are connected with the body implant region, and a buffering semiconductor layer portion (104) adjacent the DTI structure; and one or more active LDMOS devices (122) positioned in the active device region to be separated from the DTI structure by the dummy LDMOS device (121) which reduces an electric field across the sidewall insulator layer (107) in the DTI structure.

Abstract: A communication device and method include a reconfigurable receiver that is reconfigurable between communication, ranging and radar modes. The reconfigurable receiver includes a mixer configured to mix digital samples with a carrier phase estimate signal and configured to generate in-phase digital samples based on the carrier phase estimate. The reconfigurable receiver further includes a symbol correlator configured to correlate against the in-phase digital samples and generate correlated data, and a symbol binning unit configured to bin the correlated data and generate a first order channel impulse response estimate. The reconfigurable receiver yet further includes a multiplexer configured to switch the digital samples to the symbol binning unit when the reconfigurable receiver is configured in radar mode and to switch the correlated data to the symbol binning unit when the reconfigurable receiver is configured in a ranging mode.

Abstract: A driver circuit of a voltage translator includes a bias voltage generator, a drive voltage generator, an output voltage generator, and a filter circuit. The bias voltage generator is configured to receive a supply voltage, a first input voltage, and a feedback voltage, and generate a bias voltage. The feedback voltage controls an amplitude of the bias voltage. The drive voltage generator is configured to receive the supply voltage, the first input voltage, and the bias voltage, and generate a drive voltage. The output voltage generator is configured to receive the supply voltage, a second input voltage, and the drive voltage, and generate an output voltage. The drive voltage controls a slew rate of the output voltage. The filter circuit is configured to receive the output voltage, and generates and provides the feedback voltage to the bias voltage generator.

Abstract: A method of manufacturing a semiconductor device packaging panel is provided. The method includes forming a panel by placing a plurality of semiconductor die on a major side of a carrier substrate and encapsulating with an encapsulant the plurality semiconductor die and the major side of the carrier substrate. A plurality of warpage control features are formed with the encapsulant while encapsulating. The method further includes placing the panel onto a warpage control fixture to substantially flatten the panel. The plurality of warpage control features interlock with mating features of the warpage control fixture.

Abstract: A first access point (AP) for enabling distributed multiple-input-multiple-output (DMIMO) communication in a wireless network is disclosed. The first AP is configured to select a wireless client from a plurality of wireless clients or a second AP from a plurality of APs such that a difference between a path loss value associated with the first AP and the wireless client, and a path loss value associated with the second AP and the wireless client is greater than or equal to a threshold value. The wireless client is selected when the first and second APs are available for DMIMO communication, and the second AP is selected when the wireless client and the first AP are available for DMIMO communication. The first AP is further configured to associate the first and second APs with the wireless client for enabling DMIMO communication between the first and second APs and the wireless client.

Abstract: A beamformee executes a singular value decomposition operation on a channel coefficient matrix to generate a diagonal matrix and a unitary matrix. The beamformee extracts a submatrix from each of the diagonal matrix and the unitary matrix. Each of a number of rows and a number of columns of the submatrix extracted from the diagonal matrix is equal to a number of spatial streams associated with a beamformer. Further, the submatrix extracted from the unitary matrix has a number of rows and a number of columns equal to a number of beamformer antennas of the beamformer and the number of spatial streams, respectively. The beamformee generates a subspace matrix that is a product of the submatrix extracted from the diagonal matrix and a conjugate transpose of the submatrix extracted from the unitary matrix. Further, the beamformee generates a steering matrix based on the subspace matrix.

Abstract: A method for controlling communication of information includes selecting a group of stations within range of an access point, transmitting a first signal from the access point to a first station in the group, and receiving at the access point a second signal from a second station in the group. The first signal is transmitted to the first station through a downlink channel. The second signal is received from the second station through an uplink channel. Transmission of the first signal takes place during a first period and reception of the second signal takes place during a second period overlapping the first period, in order to perform full-duplex different-frequency communications based on an 802.11 standard between the access point and the first station and the second station.

Abstract: A memory includes a pair of sense amplifiers where the pair of sense amplifiers perform a multiphase memory operation to read data from two memory cells. Each sense amplifier includes two current paths. During a first phase of the memory read operation, one of the two sense amplifiers provides current through both a first memory cell and a first reference cell and the other sense amplifier of the two provides current through both a second memory cell and a second reference cell. The reference cells each have different resistance values. During a second phase of the memory read operation, one of the sense amplifiers provides current through both of the first memory cell and the second reference cell and the second sense amplifier provides current through the second memory cell and the first reference cell.

Abstract: An integrated circuit having a fingered capacitor with multiple metal fingers formed in inverted-trapezoid-shaped trenches in a multi-layer structure having a polish stop layer over an ultra-low-K dielectric layer over a low-K dielectric layer over a dielectric cap layer. The ultra-low-K dielectric layer reduces capacitance variations between the fingers, while the polish stop layer prevents metal height variations that would otherwise result from performing CMP directly on the ultra-low-K dielectric layer. The layered structure may include another low-K dielectric layer over the polish stop layer that provides a soft landing for the CMP. The polish stop layer may be removed after the CMP polishing and another ultra-low-K dielectric layer may be formed to encapsulate the tops of the metal fingers in the ultra-low-K dielectric material.

Abstract: An ultra wide band, UWB, ranging device is disclosed comprising: an input for receiving data directly from a sensor hub, and a radio configured for UWB communication; wherein the UWB device is configured to: commence a ranging process comprising at least one of a repeated range determination, and a repeated angle of arrival determination; receive sensor data from the sensor hub; process the data, and modify the ranging process in dependence on the data. Personal communication devices disclosing such UWB ranging devices are also disclosed as are associated methods.

Abstract: A system for convolving an image includes a processing circuitry that retrieves the image including a set of rows, and a set of kernels, and merges serially all columns of each kernel, to generate a merged kernel. The processing circuitry executes parallelly multiple times, a multiply-accumulate (MAC) instruction on a row loaded in a corresponding vector register and a corresponding coefficient of the merged kernel and a load instruction on a subsequent row in one clock cycle. In the same clock cycle based on the MAC instruction, a logical shift operation is executed on the merged kernel to shift a current coefficient of the merged kernel with a subsequent coefficient such that the MAC instruction is executed on the subsequent row and the subsequent coefficient in the next clock cycle. Thus, each clock cycle is utilized by the system for executing both the MAC and load instructions.

Abstract: The present application relates to a system for generating a surround view and a method of operating the system. A synthesizer module synthesizes an output frame from input frames in accordance with predefined calibration data. The input frames have an overlapping region imaging an overlapping field of view captured by two adjacent cameras. An adjustment module receives height level information representative of a height level in the overlapping region; selects a data record out of a set of predefined calibration data records in accordance with the height level information; and updates the predefined a part of the calibration data with the selected data record.

Abstract: Embodiments of a device and method are disclosed. In an embodiment, a method of wireless communications involves at a receiver, receiving a first packet, subsequently, at the receiver, receiving a second packet, and determining whether the second packet is a repetition of the first packet based on wireless communications channel estimation information associated with the first and second packets.

Abstract: Various embodiments relate to a power converter including a resonant converter with an controller, the controller configured to control the converter to operate in a normal mode when output power is above a burst mode threshold level, start a timer when the output power falls below the burst mode threshold level, continue operating in the normal mode until the timer reaches a predetermined time and operate in burst mode when the timer reaches the predetermined time.

Abstract: An embedded information system includes a load control circuit coupleable to an external memory that contains instructions and constant data (organized by variable sized load units, LUs, and where at least one property of a LU is specified within metadata) associated with application code of a software application, at least one processor configured to execute the at least one application code; an internal memory configured as main system memory in a first part and as a cache for storing the instructions and constant data for an execution of the at least one application code from the external memory in a second part. The load control circuit is configured to load the LUs associated with the at least one application code from the external memory with a granularity of a single LU into the internal memory.

Abstract: A multi-channel digital to analog converter, DAC, comprising: a DAC configured to provide an analog signal comprising a plurality of time division multiplexed sub-signals, each provided for one of a plurality of output channels; wherein each of the output channels include: a sampling capacitor; a selector switch configured to couple the sampling capacitor of the respective output channel to the output terminal of the DAC such that the sampling capacitor samples the analog signal over a plurality of discrete sampling periods; a comparator configured to provide a comparator output signal, wherein the sampling capacitor is coupled to an input terminal of the comparator; and an output control gate configured to control whether or not the comparator output signal is output from the respective output channel at a predetermined time later than a first of the respective plurality of discrete sampling periods.