Abstract: Dataset distillation compresses large datasets into smaller synthetic coresets that retain performance with the aim of reducing storage and computational burdens of processing an original, entire dataset. The present disclosure provides an improved algorithm that uses a non-deterministic feature approximation of neural network Gaussian process (NNGP) kernels, or other trained kernels, that reduces a kernel matrix computation to O(|S|). When combined with a modified Platt scaling loss, the disclosed algorithm can provide at least a 100-fold speedup over a Kernel-Inducing Points (KIP) algorithm and can run on a single graphics processing unit. The disclosed Random Feature Approximation Distillation (RFAD) algorithm can perform competitively with other dataset condensation algorithms in accuracy over a range of large-scale datasets, both in kernel regression and finite-width network training. The disclosed techniques can be effective on tasks such as model interpretability and data privacy preservation.

Abstract: Disclosed techniques include receiving a user model as a computational graph and a list of risk metrics indicating types of risk factors to assess, and selecting at least one composable wrapper for wrapping the user model based on a risk metric(s) in the list. The techniques further include determining a set of graph-level transformations to apply to the computational graph representation that implement a risk estimate for the user model, applying the set of graph-level transformations to modify the operations performed by the user model, and generating, based on the transformed computational graph, a modified executable variant of the user model that, when executed within a machine learning framework, produces: outputs that preserve the structure and interpretation of the user model outputs; and corresponding uncertainty estimates.

Abstract: A controller for an autonomous vehicle is trained using simulated paths on a roadway and simulated observations that are formed by transforming images previously acquired on similar paths on that roadway. Essentially an unlimited number of paths may be simulated, enabling optimization approaches including reinforcement learning to be applied to optimize the controller.

Abstract: Systems and methods described herein relate to using multimodal foundation models. In one embodiment, a method includes receiving images and a foundation multi-model, selecting a mask set, modifying the foundation multi-model to include query, key, and value matrices, and applying the mask set to the foundation multi-model to obtain patch-aligned features.

Abstract: Systems and methods are provided for generating a statistical parameter representing a state of a surgical procedure from sensor data. Sensor data representing a time period. is received from a sensor. Numerical features representing the time period are generated from the sensor data. Each of a plurality of long short term memory units are updated according to the plurality of numerical features via a message passing process. The long short term memory units are connected to form a graph, with a first set of the long short term memory units representing a plurality of nodes of the graph and a second set of the long short term memory units representing a plurality of hyperedges of the graph. A statistical parameter representing a state of the surgical procedure for the time period is derived from an output of one of the long short term memory units and provided to a user.

Abstract: A system and method for patient positioning during radiotherapy. The system can include a patient support structure configured to receive a patient during a radiotherapy process using a radiotherapy source to deliver a therapy to the patient when positioned on the patient support structure, a patient positioning system configured to adjust a position of the patient support structure relative to the radiotherapy source, a flexible actuator configured to secure the patient to the patient support and adjust a position of the patient relative to the patient support, and an imaging system configured to acquire imaging data of the patient, the patient support, and the flexible actuator during the radiotherapy process.

Abstract: Described herein are systems and methods for improving accuracy of model output generation. A method can include obtaining a risk-aware model and a user input, applying the risk-aware model to the user input, receiving, based on the applying, model output and corresponding risk values, comparing the corresponding risk values to a threshold risk value, and regenerating the user input based on the comparing. The method can also include iteratively performing the applying, receiving, comparing, and regenerating using the regenerated user input until one or more processing conditions is met. The user input can be regenerated in response to determining that the corresponding risk values are greater than the threshold risk value. The model output can include one or more sequences in a train-of-thought (TOT) of the risk-aware model.

Abstract: Methods for optimizing a user model are disclosed. In at least some instances, the methods include receiving a user model and then identifying one or more sub-optimal combinations of operations in the user model. The methods further include generating a modification (e.g., an optimal subgraph) for at least one identified sub-optimal combination of operations in the user model and replacing that identified sub-optimal combination of operations with the generated modification for that identified sub-optimal combination of operations. Still further, the method includes returning the user model with the generated modification as a transformed model. Other methods, and systems for performing any methods disclosed, are also provided.

Abstract: Methods of modifying an input, such as a prompt, for a machine learning model are disclosed. In at least some instances, the method includes transforming a received user model into a risk-aware model, applying that risk-aware model to a received input, and receiving, based on the application of the risk-aware model, output and corresponding risk values. In turn, the method includes determining whether the corresponding risk values, or an aggregate of such values, are less than a predetermined threshold level. If the values or aggregate are greater than the predetermined threshold level, then a process is performed to modify the input to minimize the corresponding risk values. The risk-aware model is iteratively applied to the modified input and the modification process continues to be performed until the corresponding risk values are less than the predetermined threshold level. Other methods, and systems for performing any methods disclosed, are also provided.

Abstract: An autonomous, mobile robotic device (AMR) is configured with one or more UVC radiation sources, and operates to traverse a path while disinfecting an interior space. Each UVC radiation source is connected to the AMR by an articulating arm that is controlled to orient each source towards a feature or surface that is selected for disinfection during the time that the AMR is moving through the space. The location of each feature selected for disinfection can be mapped, and this map information, a current AMR location and pose can be used to generate signals that are used to control the articulating arm to orient each UVC lamp towards a feature that is selected for disinfection.

Abstract: In certain aspects, a computer-implemented method includes generating a candidate structure based on constituent elements and a target behavior. The method includes simulating, in a prespecified environment, behavior of the candidate structure under prespecified constraints. The method includes evaluating the behavior of the candidate structure, during the simulating, under the prespecified constraints. The method includes calculating a gradient of the behavior of the candidate structure simulated under the prespecified constraints with respect to parameters associated with the constituent elements. The method includes optimally adjusting, based on the calculated gradient, the parameters for improving the behavior of the candidate structure under the prespecified constraints. The method includes applying the adjusted parameters to the candidate structure for generating an adjusted candidate structure.

Abstract: A method of generating an output trajectory of an ego vehicle is described. The method includes extracting high-level features from a bird-view image of a traffic environment of the ego vehicle. The method also includes generating, using an automaton generative network, an automaton including an automaton state distribution describing a behavior of the ego vehicle in the traffic environment according to the high-level features. The method further includes generating the output trajectory of the ego vehicle according to extracted bird-view features of the bird-view image and the automaton state distribution describing the behavior of the ego vehicle in the traffic environment.

Abstract: Dataset distillation compresses large datasets into smaller synthetic coresets that retain performance with the aim of reducing storage and computational burdens of processing an original, entire dataset. The present disclosure provides an improved algorithm that uses a non-deterministic feature approximation of neural network Gaussian process (NNGP) kernels, or other trained kernels, that reduces a kernel matrix computation to O(|S|). When combined with a modified Platt scaling loss, the disclosed algorithm can provide at least a 100-fold speedup over a Kernel-Inducing Points (KIP) algorithm and can run on a single graphics processing unit. The disclosed Random Feature Approximation Distillation (RFAD) algorithm can perform competitively with other dataset condensation algorithms in accuracy over a range of large-scale datasets, both in kernel regression and finite-width network training. The disclosed techniques can be effective on tasks such as model interpretability and data privacy preservation.

Abstract: Disclosed are techniques, and related systems, for developing and implementing a unified framework for quantifying risk in machine learning models, such as deep neural networks. The framework can be easy-to-use and flexible to apply to different types of machine learning models. The framework can be used to automatically assess different forms of risk in parallel, including but not limited to aleatoric uncertainty, epistemic uncertainty, and/or vacuitic uncertainty. To that end, the disclosed framework provides wrappers that compose different automatic risk assessment algorithms. The obtained risk estimates can be used, for example, for: providing a deeper insight into decision boundaries of neural networks; performing downstream tasks by integrating the risk estimates back into a learning lifecycle for the model to improve robustness and generalization; and/or improving safety by identifying potential model failures based on the risk values.

Abstract: System, methods, and other embodiments described herein relate to training a scene simulator for rendering 2D scenes using data from real and simulated agents. In one embodiment, a method includes acquiring trajectories and three-dimensional (3D) views for multiple agents from observations of real vehicles. The method also includes generating a 3D scene having the multiple agents using the 3D views and information from simulated agents. The method also includes training a scene simulator to render scene projections using the 3D scene. The method also includes outputting a 2D scene having simulated observations for a driving scene using the scene simulator.

Abstract: Systems and methods are provided for generating a statistical parameter representing a state of a surgical procedure from sensor data. Sensor data representing a time period. is received from a sensor. Numerical features representing the time period are generated from the sensor data. Each of a plurality of long short term memory units are updated according to the plurality of numerical features via a message passing process. The long short term memory units are connected to form a graph, with a first set of the long short term memory units representing a plurality of nodes of the graph and a second set of the long short term memory units representing a plurality of hyperedges of the graph. A statistical parameter representing a state of the surgical procedure for the time period is derived from an output of one of the long short term memory units and provided to a user.

Abstract: An autonomous, mobile robotic device (AMR) is configured with one or more UVC radiation sources, and operates to traverse a path while disinfecting an interior space. Each UVC radiation source is connected to the AMR by an articulating arm that is controlled to orient each source towards a feature or surface that is selected for disinfection during the time that the AMR is moving through the space. The location of each feature selected for disinfection can be mapped, and this map information, a current AMR location and pose can be used to generate signals that are used to control the articulating arm to orient each UVC lamp towards a feature that is selected for disinfection.

Abstract: In some aspects, a system comprises a computer hardware processor and a non-transitory computer-readable storage medium storing processor-executable instructions for receiving, from one or more sensors, sensor data relating to a robot; generating, using a statistical model, based on the sensor data, first control information for the robot to accomplish a task; transmitting, to the robot, the first control information for execution of the task; and receiving, from the robot, a result of execution of the task.

Abstract: Understanding the intent of human drivers and adapting to their driving styles is used to increased efficiency and safety of autonomous vehicles (AVs) by enabling them to behave in safe and predictable ways without requiring explicit inter-vehicle communication. A Social Value Orientation (SVO), which quantifies the degree of an agent's selfishness or altruism, is estimated by the AV for other vehicles to better predict how they will interact and cooperate with others. Interactions between agents are modeled as a best response game wherein each agent negotiates to maximize their own utility. A dynamic game solution uses the Nash equilibrium, yielding an online method of predicting multi-agent interactions given their SVOs. This approach allows autonomous vehicles to observe human drivers, estimate their SVOs, and generate an autonomous control policy in real time.

Abstract: An approach to autonomous navigation of a vehicle augments a static map of an environment with a clutter map characterizing a risk of encountering an object that is not represented in the static map of the environment. For example, the clutter map may be based on locations and velocities of those objects, and route planning may avoid planning a path through locations that have a high risk of occupancy, and therefore potential delay or collision.