Abstract: Systems and methods for augmenting a data set are provided. An example method may include locating a foreground object disposed within a seed image, identifying an object class corresponding to the foreground object, and, based on the identified object class, determining a target value for an object property of the foreground object. The example method may also include applying a transformation function to transform the foreground object into a transformed object, where the transformation function modifies the object property of the foreground object from having an initial value to having the target value. The example method may further include transplanting the transformed object into a background image so as to produce an augmented image and augmenting an initial set of images with the augmented image so as to produce an augmented set of images for training a predictive model.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for analyzing ultrasonic signals to recognize mouthed articulations. One of the methods includes generating an ultrasonic carrier signal and coupling the ultrasonic carrier signal to a person's vocal tract. The method includes detecting a modulated ultrasonic signal, the modulated ultrasonic signal corresponding to the ultrasonic carrier signal modulated by the person's vocal tract to include information about articulations mouthed by the person; analyzing, using a data processing apparatus, the modulated ultrasonic signal to recognize the articulations mouthed by the person from the information in the modulated ultrasonic signal; and generating, using the data processing apparatus, an output in response to the recognized articulations.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for training a generator neural network to adapt input images.

Abstract: An apparatus includes an acoustic source for directing an acoustic wave towards a workspace, a variable acoustic diffractive device positioned in a path of the acoustic wave between the acoustic source and the workspace, the variable acoustic diffractive device including an array of elements each having independently variable acoustic properties, and an electronic controller in communication with the acoustic source and the variable acoustic diffractive device, the electronic controller programmed to supply signals to cause the acoustic source to produce the acoustic wave and the variable acoustic diffractive device to diffract the acoustic wave from the acoustic source to provide non-uniform acoustic forces in a medium located in the workspace.

Abstract: A system for collection of rainwater in the open ocean may include: (a) one or more ocean-going vessels, wherein each ocean-going vessel is configured for collection and storage of rainwater, wherein each ocean-going vessel is configured to drift with surface ocean currents in order to navigate to one or more delivery locations, wherein each delivery location is on or near to a land mass; and (b) one or more delivery stations located at the one or more delivery locations, wherein each delivery station is configured to receive stored rainwater from one or more of the ocean-going vessels.

Abstract: Methods, systems, and apparatus for quantum analog-digital conversion. In one aspect, a method includes obtaining a quantum analog signal; applying a hybrid analog-digital encoding operation to the quantum analog signal and a qudit in an initial state to obtain an evolved state of the qudit, wherein the hybrid analog-digital encoding operation is based on a swap operation comprising multiple adder operations; and providing the qudit in the evolved state as a quantum digital encoding of the quantum analog signal.

Abstract: Techniques are described for determining the strain on a cell wall using two models: 1) a short timescale model, describing the relationship between physical properties assumed to be fixed and 2) a long timescale model, describing the dynamic chemical composition of a cell wall. Short term modeling of the physical properties in a cell wall is used to properly understand how physical factors such as osmotic pressure affects the strain on the cell wall, which is in turn used to identify conditions under which a cell wall will cease to function properly or lyse entirely. Although temporally the physical properties which cause cell walls to underperform/lyse can be evaluated under a short time frame, the chemical properties that lead to the physical properties which cause that behavior themselves change over much longer timescales, in a relative sense.

Abstract: A method of calibrating a collimating lens system includes transmitting, using an optical transmitter, a beam out of an optical fiber and through a collimating lens of the collimating lens system. The beam is reflected off a perfect flat mirror positioned at an output of the collimating lens and back towards the collimating lens, and received, via the collimating lens, at a power meter connected to the optical fiber. The method also includes adjusting a position of a tip of the optical fiber proximal to the collimating lens while tracking a power reading using the power meter, selecting a calibration position of the optical fiber corresponding to a highest power reading, and securing the optical fiber relative to the collimating lens using the calibration position.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer-storage media, for a lighting controller for sea lice detection. In some implementations, a pulse of red light and a pulse of blue light can be timed with the exposure of a camera to capture multiple images of a fish or group of fishes in both red and blue light. By using the captured images with different color light, computers can detect features on the body of a fish including sea lice, skin lesions, shortened operculum or other physical deformities and skin features. Detection results can aid in mitigation techniques or be stored for analytics. For example, sea lice detection results can inform targeted treatments comprised of lasers, fluids, or mechanical devices such as a brush or suction.

Abstract: A system for physically simulating a neural network is described herein. The system includes a plurality of physical voxels configurable to represent nodes of the neural network operating in response to electromagnetic radiation. Each of the physical voxels includes an impedance adjuster, a field detector, and a signal adjuster. The impedance adjuster adjusts impedance to the electromagnetic radiation within a corresponding one of the physical voxels. Weights between nodes of the neural network are based on the adjusted impedance. The field detector measures local field response within the corresponding one of the physical voxels. The local field response is representative of the electromagnetic radiation with the adjusted impedance. The signal adjuster is coupled to receive the local field response and apply an adjustment to the received local field response. The adjustment corresponds to an activation function of the neural network at the corresponding one of the physical voxels.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer-readable storage media, for automated camera positioning for feeding behavior monitoring. In some implementations, a system obtains an image of a scene from one or more cameras and a spatial model that corresponds to a subfeeder. The system determines, based on data from the spatial model and the image of the scene, a current position of the one or more cameras relative to the subfeeder. The system selects an updated position of the one or more cameras relative to the subfeeder. The system moves the one or more cameras to the updated position. In some implementations, the system dynamically selects the updated position of the one or more cameras relative to the subfeeder.

Abstract: Methods, systems, and apparatus, including computer programs stored on a computer-readable storage medium, for performing hybrid additive manufacturing. In some implementations, a hybrid additive manufacturing system causes a three-dimensional printer to print a sheet of material that includes printed features on one side of the sheet, each printed feature having a structural characteristic that is different from structural characteristics of a majority of the sheet. The system causes a molding machine to form the sheet using a mold, where at least one printed feature corresponds with an area of the mold at which the sheet deforms while being formed.

Abstract: Implementations are described herein for using machine learning to determine whether candidate crop fields are suitable for management by particular agricultural entities. In various implementations, a machine learning model may be applied to input data to generate output data. The input data may include a first plurality of data points corresponding to field-level agricultural management practices of an agricultural entity. The output data may be indicative of one or more predicted outcomes of the agricultural entity implementing the field-level agricultural management practices on one or more candidate crop fields not currently managed by the agricultural entity. Based on one or more of the predicted outcomes, one or more computing devices may be caused to provide a user associated with the agricultural entity with information about one or more of the candidate crop fields, and/or one or more parameter inputs of a graphical user interface may be prepopulated.

Abstract: Implementations are described herein for coordinating semi-autonomous robots to perform agricultural tasks on a plurality of plants with minimal human intervention. In various implementations, a plurality of robots may be deployed to perform a respective plurality of agricultural tasks. Each agricultural task may be associated with a respective plant of a plurality of plants, and each plant may have been previously designated as a target for one of the agricultural tasks. It may be determined that a given robot has reached an individual plant associated with the respective agricultural task that was assigned to the given robot. Based at least in part on that determination, a manual control interface may be provided at output component(s) of a computing device in network communication with the given robot. The manual control interface may be operable to manually control the given robot to perform the respective agricultural task.

Abstract: Cascading variational autoencoder (“VAE”) models can be used to detect robot collisions while a robot is performing a task. For a current state of the robot, various implementations include a VAE used to generate a latent space of the current state, and a predictor network used to generate a predicted latent space for the current state. A collision can be determined based on a difference between the latent space for the current state and the predicted latent space for the current state.

Abstract: An example system includes a patterned light projector operable to direct first and second portions of patterned light toward first and second surfaces, respectively, in an environment. The first and second surfaces may be at first and second distances, respectively, from the structured light projector. A graduated optical filter may be situated along an optical path of the patterned light. The graduated optical filter includes first and second regions to attenuate an intensity of the first and second portions of the patterned light, respectively, by first and second amounts, respectively. The first amount is greater than the second amount. The system additionally includes an image sensor operable to generate image data based on at least the first and second portions of the patterned light and a processor configured to determine first and second values indicative of an estimate of the first and second distances, respectively, based on the image data.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for processing confidential data and confidential programs while providing mutual security to the data and programs. A method includes receiving, from a first system, customer energy data, including data representing energy consumption by a customer; receiving, from a second system, program data representing one or more programs for processing the customer energy data; executing the programs with the customer energy data as input to produce output that includes estimated energy consumption data, while providing security for the program data from access by the first system and any third party and while providing security for the customer energy data and the estimated energy consumption data from access by the second system and any third party; and providing the estimated energy consumption data as output (i) to the first system or (ii) to the customer or (iii) both.