Abstract: Described is a system for continual learning using experience replay. In operation, the system receives a plurality of tasks sequentially, from which a current task is fed to an encoder. The current task has data points associated with the current task. The encoder then maps the data points into an embedding space, which reflects the data points as discriminative features. A decoder then generates pseudo-data points from the discriminative features, which are provided back to the encoder. The discriminative features are updated in the embedding space based on the pseudo-data points. The encoder then learns (updates) a classification of a new task by matching the new task with the discriminative features in the embedding space.

Abstract: A method for generating human-machine hybrid predictions of answers to forecasting problems includes: parsing text of an individual forecasting problem to identify keywords; generating machine models based on the keywords; scraping data sources based on the keywords to collect scraped data relevant to the individual forecasting problem; providing the scraped data to the machine models; receiving machine predictions of answers to the individual forecasting problem from the machine models based on the scraped data; providing, by the computer system via a user interface, the scraped data to human participants; receiving, by the computer system via the user interface, human predictions of answers to the individual forecasting problem from the human participants; aggregating the machine predictions with the human predictions to generate aggregated predictions; and generating and outputting a hybrid prediction based on the aggregated predictions.

Abstract: Described is a system for learning object labels for control of an autonomous platform. Pseudo-task optimization is performed to identify an optimal pseudo-task for each source model of one or more source models. An initial target network is trained using the optimal pseudo-task. Source image components are extracted from source models, and an attribute dictionary of attributes is generated from the source image components. Using zero-shot attribution distillation, the unlabeled target data is aligned with the source models similar to the unlabeled target data. The unlabeled target data are mapped onto attributes in the attribute dictionary. A new target network is generated from the mapping, and the new target network is used to assign an object label to an object in the unlabeled target data. The autonomous platform is controlled based on the object label.

Abstract: Described is a system for adapting a deep convolutional neural network (CNN). A deep CNN is first trained on an annotated source image domain. The deep CNN is adapted to a new target image domain without requiring new annotations by determining domain agnostic features that map from the annotated source image domain and a target image domain to a joint latent space, and using the domain agnostic features to map the joint latent space to annotations for the target image domain.

Abstract: A comb enhanced oscillator in which a drive signal from a first oscillator is split into two signals. The first signal is applied to a nonlinear resonator producing a phononic frequency comb of equally spaced resonances. The second signal is passed through an amplitude detector and a phase shifter. In one embodiment, the comb is applied to the phase shifter to correct for AM-PM cross-correlation noise and then applied to a phase lock loop (PLL) for locking to a second oscillator. The output of the second oscillator is used as the output of the comb enhanced oscillator.

Abstract: A hyperspectral imager (HSI) includes a first thin film filter, the first thin film filter including a first quarter wave mirror, a second quarter wave mirror, and a low-refractive-index wedge between the first quarter wave mirror and the second quarter wave mirror. The low-refractive-index wedge has a height dimension such that a distance between the first quarter wave mirror and the second quarter wave mirror increases linearly along a length of the low-refractive-index wedge.

Abstract: Some variations provide a metal matrix nanocomposite composition comprising metal-containing microparticles and nanoparticles, wherein the nanoparticles are chemically and/or physically disposed on surfaces of the microparticles, and wherein the nanoparticles are consolidated in a three-dimensional architecture throughout the composition. The composition may serve as an ingot for producing a metal matrix nanocomposite. Other variations provide a functionally graded metal matrix nanocomposite comprising a metal-matrix phase and a reinforcement phase containing nanoparticles, wherein the nanocomposite contains a gradient in concentration of the nanoparticles. This nanocomposite may be or be converted into a master alloy. Other variations provide methods of making a metal matrix nanocomposite, methods of making a functionally graded metal matrix nanocomposite, and methods of making a master alloy metal matrix nanocomposite. The metal matrix nanocomposite may have a cast microstructure.

Abstract: Some variations provide a composition for additive manufacturing (3D printing) of metals, comprising: from 10 vol % to 70 vol % of a photocurable liquid resin; from 10 vol % to 70 vol % of metal or metal alloy particles, optionally configured with a photoreflective surface; and from 0.01 vol % to 10 vol % of a photoinitiator. Other variations provide a composition for additive manufacturing of metals, comprising: from 1 vol % to 70 vol % of a photocurable liquid resin; from 0.1 vol % to 98 vol % of an organometallic compound containing a first metal; from 1 vol % to 70 vol % of metal or metal alloy particles containing a second metal (which may be the same as or different than the first metal); and from 0.01 vol % to 10 vol % of a photoinitiator. Many examples of metals, photocurable resins, organometallic compounds, photoinitiators, and optional additives are disclosed, and methods of making and using the composition are described.

Abstract: A method includes receiving, by machine-learning logic, observations indicative of a states associated with a first and second group of vehicles arranged within an engagement zone during a first interval of an engagement between the first and the second group of vehicles. The machine-learning logic determines actions based on the observations that, when taken simultaneously by the first group of vehicles during the first interval, are predicted by the machine-learning logic to result in removal of one or more vehicles of the second group of vehicles from the engagement zone during the engagement. The machine-learning logic is trained using a reinforcement learning technique and on simulated engagements between the first and second group of vehicles to determine sequences of actions that are predicted to result in one or more vehicles of the second group being removed from the engagement zone. The machine-learning logic communicates the plurality of actions to the first group of vehicles.

Abstract: A phononic composite material providing structural strength and blocking the propagation of elastic waves over a frequency range referred to as the bandgap. In one embodiment, the phononic composite material consists of a plurality of periodic units, each of which includes a central fiber, a relatively soft interface layer surrounding the fiber, and a matrix layer surrounding the interface layer. The properties of the interface layer may be adjusted, e.g., by adjusting the temperature of the phononic composite material, to transition from a state with a bandgap to a state lacking a bandgap.

Abstract: A signal processing circuit. In some embodiments, the signal processing circuit includes a first sample and hold circuit and a second sample and hold circuit. The first sample and hold circuit may include: a hold capacitor; an input switch connected between a common input node and the hold capacitor; a signal path amplifier having an input connected to the hold capacitor; and an output switch connected between an output of the signal path amplifier and a common output node. An input of a voltage feedback amplifier may be connected to the hold capacitor, and an output of the voltage feedback amplifier may be operatively coupled to an internal node of the input switch.

Abstract: Some variations provide an aluminum alloy feedstock for additive manufacturing, the aluminum alloy feedstock comprising from 79.8 wt % to 88.3 wt % aluminum; from 1.1 wt % to 2.1 wt % copper; from 3.0 wt % to 4.6 wt % magnesium; from 7.1 wt % to 9.0 wt % zinc; and from 0.5 wt % to 2.8 wt % zirconium as a grain-refiner element. The aluminum alloy feedstock may be in the form of an ingot powder. In some variations, the aluminum alloy feedstock comprises from 81.3 wt % to about 87.8 wt % aluminum; from 1.2 wt % to 2.0 wt % copper; from 3.2 wt % to 4.4 wt % magnesium; from 7.3 wt % to 8.7 wt % zinc; and from 0.5 wt % to 2.8 wt % zirconium.

Abstract: An array of preferably electrically small scatterers is spaced at more or less regular intervals from a central transmitter. Each scatterer element includes a tunable or static reactive load which allows the propagation and fields generated by the central transmitter to be precisely controlled. Each scatterer element in the array also includes a resistive element whose value may change as a function of a distance between each scatterer and the central transmitter and which typically increases as a function of that distance. The central transmitter in the array nominally comprises an antenna, matching network, RF driver, and a vehicle, which may be a maritime vehicle or platform. The antenna for this transmitter may be comprised of an electrically small monopole oriented normal to the surface of the ocean or an electrically small loop antenna oriented with its magnetic moment parallel to the surface of the ocean.

Abstract: A memory circuit configured to perform multiply-accumulate (MAC) operations for performance of an artificial neural network includes a series of synapse cells arranged in a cross-bar array. Each cell includes a memory transistor connected in series with a memristor. The memory circuit also includes input lines connected to the source terminal of the memory transistor in each cell, output lines connected to an output terminal of the memristor in each cell, and programming lines coupled to a gate terminal of the memory transistor in each cell. The memristor of each cell is configured to store a conductance value representative of a synaptic weight of a synapse connected to a neuron in the artificial neural network, and the memory transistor of each cell is configured to store a threshold voltage representative of a synaptic importance value of the synapse connected to the neuron in the artificial neural network.

Abstract: A method and apparatus for enhancing the sensitivity of an oscillator circuit functioning, in use, to sense changes in the inductance of inductive elements and/or the capacitance of capacitive elements coupled to said oscillator circuit. The oscillator circuit is coupled to a nonlinear resonator for generating a comb of frequencies in response to a drive frequency generated by the oscillator circuit, the comb of frequencies having at least a portion of at least one tooth for which an absolute value of the first derivative of the drive frequency with respect to said comb frequency is less than 1.0, comparing an output of the nonlinear resonator with an output of a reference oscillator for detecting changes in the drive signal of the oscillator circuit as enhanced by the slope of the at least a portion of at least one tooth for which the absolute value of the first derivative of the drive frequency with respect to said comb frequency is less than 1.0.

Abstract: A field effect transistor includes a substrate, a passivation layer on the substrate forming a passivated substrate, wherein the passivation layer is inert to XeF2, and a graphene lateral heterostructure field effect transistor (LHFET) on the passivated substrate.

Abstract: An energy conversion device disposed in series with an RF driver circuit and an RF antenna, the energy conversion device being arranged to convert a portion of available RF power from the RF driver circuit into a different form of energy (direct current, thermal, or higher frequency electromagnetic waves such as light) which is converted, if needed, to DC and stored in an energy storage device coupled with the RF driver circuit for supplying recycled electrical energy thereto. The RF antenna may be an electrically small antenna and thus a antenna matching network may be provided between the RF driver circuit and the RF antenna. The energy conversion device may comprise, for example, (i) a transformer in combination with a rectifying circuit, (ii) a full wave rectifier, (iii) a half wave rectifier, (iv) a heat and/or light producing device, an energy converter (such as a generator) or a combination of the foregoing.

Abstract: A vertical field-effect transistor (FET), comprising a first doped region of a first material, said first doped region having a first doping and being formed on a surface of a substrate, a second doped region of said first material, said second doped region having a second doping and being formed on the first doped region, and a third doped region of said first material, said third doped region having a third doping and being formed on the second doped region, wherein the first doped region has a first width along a first direction parallel to said surface of the substrate, the second doped region has a second width along said first direction, the third doped region has a third width along said first direction, the second width being smaller than the first and third widths.

Abstract: A differential gradiometer comprising a substrate with at least a pair of resonators disposed thereon, wherein each of the at least a pair of resonators is sensitive to environmental factors which produces differential strains between the resonators, a first one of said pair of resonators being connected with a circuit for forming a first oscillator, the second one of said pair of resonators being connected with another circuit for forming a non-linear oscillator, an output of the first oscillator being applied to the non-linear oscillator for generating a comb of frequencies, wherein an addition oscillator is locked to the nth tooth of the comb thereby increasing the sensitivity of the gradiometer by a factor of n.

Abstract: Described is a system for detection of network activities using transitive tensor analysis. The system divides a tensor into multiple subtensors, where the tensor represents communications on a communications network of streaming network data. Each subtensor is decomposed, separately and independently, into subtensor mode factors. Using transitive mode factor matching, orderings of the subtensor mode factors are determined. A set of subtensor factor coefficients is determined for the subtensor mode factors, and the subtensor factor coefficients are used to determine the relative weighting of the subtensor mode factors, and activity patterns represented by the subtensor mode factors are detected. Based on the detection, an alert of an anomaly is generated, indicating a in the communications network and a time of occurrence.