Message-Passing Monte Carlo: Generating low-discrepancy point sets via graph neural networks.

Discrepancy is a well-known measure for the irregularity of the distribution of a point set. Point sets with small discrepancy are called low discrepancy and are known to efficiently fill the space in a uniform manner. Low-discrepancy points play a central role in many problems in science and engineering, including numerical integration, computer vision, machine perception, computer graphics, machine learning, and simulation. In this work, we present a machine learning approach to generate a new class of low-discrepancy point sets named Message-Passing Monte Carlo (MPMC) points. Motivated by the geometric nature of generating low-discrepancy point sets, we leverage tools from Geometric Deep Learning and base our model on graph neural networks. We further provide an extension of our framework to higher dimensions, which flexibly allows the generation of custom-made points that emphasize the uniformity in specific dimensions that are primarily important for the particular problem at hand. Finally, we demonstrate that our proposed model achieves state-of-the-art performance superior to previous methods by a significant margin. In fact, MPMC points are empirically shown to be either optimal or near-optimal with respect to the discrepancy for low dimension and small number of points, i.e., for which the optimal discrepancy can be determined.

Contextual and combinatorial structure in sperm whale vocalisations.

Sperm whales (Physeter macrocephalus) are highly social mammals that communicate using sequences of clicks called codas. While a subset of codas have been shown to encode information about caller identity, almost everything else about the sperm whale communication system, including its structure and information-carrying capacity, remains unknown. We show that codas exhibit contextual and combinatorial structure. First, we report previously undescribed features of codas that are sensitive to the conversational context in which they occur, and systematically controlled and imitated across whales. We call these rubato and ornamentation. Second, we show that codas form a combinatorial coding system in which rubato and ornamentation combine with two context-independent features we call rhythm and tempo to produce a large inventory of distinguishable codas. Sperm whale vocalisations are more expressive and structured than previously believed, and built from a repertoire comprising nearly an order of magnitude more distinguishable codas. These results show context-sensitive and combinatorial vocalisation can appear in organisms with divergent evolutionary lineage and vocal apparatus.

MultiSenseBadminton: Wearable Sensor-Based Biomechanical Dataset for Evaluation of Badminton Performance.

The sports industry is witnessing an increasing trend of utilizing multiple synchronized sensors for player data collection, enabling personalized training systems with multi-perspective real-time feedback. Badminton could benefit from these various sensors, but there is a scarcity of comprehensive badminton action datasets for analysis and training feedback. Addressing this gap, this paper introduces a multi-sensor badminton dataset for forehand clear and backhand drive strokes, based on interviews with coaches for optimal usability. The dataset covers various skill levels, including beginners, intermediates, and experts, providing resources for understanding biomechanics across skill levels. It encompasses 7,763 badminton swing data from 25 players, featuring sensor data on eye tracking, body tracking, muscle signals, and foot pressure. The dataset also includes video recordings, detailed annotations on stroke type, skill level, sound, ball landing, and hitting location, as well as survey and interview data. We validated our dataset by applying a proof-of-concept machine learning model to all annotation data, demonstrating its comprehensive applicability in advanced badminton training and research.

Adaptive tactile interaction transfer via digitally embroidered smart gloves.

Human-machine interfaces for capturing, conveying, and sharing tactile information across time and space hold immense potential for healthcare, augmented and virtual reality, human-robot collaboration, and skill development. To realize this potential, such interfaces should be wearable, unobtrusive, and scalable regarding both resolution and body coverage. Taking a step towards this vision, we present a textile-based wearable human-machine interface with integrated tactile sensors and vibrotactile haptic actuators that are digitally designed and rapidly fabricated. We leverage a digital embroidery machine to seamlessly embed piezoresistive force sensors and arrays of vibrotactile actuators into textiles in a customizable, scalable, and modular manner. We use this process to create gloves that can record, reproduce, and transfer tactile interactions. User studies investigate how people perceive the sensations reproduced by our gloves with integrated vibrotactile haptic actuators. To improve the effectiveness of tactile interaction transfer, we develop a machine-learning pipeline that adaptively models how each individual user reacts to haptic sensations and then optimizes haptic feedback parameters. Our interface showcases adaptive tactile interaction transfer through the implementation of three end-to-end systems: alleviating tactile occlusion, guiding people to perform physical skills, and enabling responsive robot teleoperation.

Concept Graph Neural Networks for Surgical Video Understanding.

Analysis of relations between objects and comprehension of abstract concepts in the surgical video is important in AI-augmented surgery. However, building models that integrate our knowledge and understanding of surgery remains a challenging endeavor. In this paper, we propose a novel way to integrate conceptual knowledge into temporal analysis tasks using temporal concept graph networks. In the proposed networks, a knowledge graph is incorporated into the temporal video analysis of surgical notions, learning the meaning of concepts and relations as they apply to the data. We demonstrate results in surgical video data for tasks such as verification of the critical view of safety, estimation of the Parkland grading scale as well as recognizing instrument-action-tissue triplets. The results show that our method improves the recognition and detection of complex benchmarks as well as enables other analytic applications of interest.

Efficient automatic design of robots.

Robots are notoriously difficult to design because of complex interdependencies between their physical structure, sensory and motor layouts, and behavior. Despite this, almost every detail of every robot built to date has been manually determined by a human designer after several months or years of iterative ideation, prototyping, and testing. Inspired by evolutionary design in nature, the automated design of robots using evolutionary algorithms has been attempted for two decades, but it too remains inefficient: days of supercomputing are required to design robots in simulation that, when manufactured, exhibit desired behavior. Here we show de novo optimization of a robot's structure to exhibit a desired behavior, within seconds on a single consumer-grade computer, and the manufactured robot's retention of that behavior. Unlike other gradient-based robot design methods, this algorithm does not presuppose any particular anatomical form; starting instead from a randomly-generated apodous body plan, it consistently discovers legged locomotion, the most efficient known form of terrestrial movement. If combined with automated fabrication and scaled up to more challenging tasks, this advance promises near-instantaneous design, manufacture, and deployment of unique and useful machines for medical, environmental, vehicular, and space-based tasks.

Prescribing Cartesian Stiffness of Soft Robots by Co-Optimization of Shape and Segment-Level Stiffness.

Soft robots aim to revolutionize how robotic systems interact with the environment thanks to their inherent compliance. Some of these systems are even able to modulate their physical softness. However, simply equipping a robot with softness will not generate intelligent behaviors. Indeed, most interaction tasks require careful specification of the compliance at the interaction point; some directions must be soft and others firm (e.g., while drawing, entering a hole, tracing a surface, assembling components). On the contrary, without careful planning, the preferential directions of deformation of a soft robot are not aligned with the task. With this work, we propose a strategy to prescribe variations of the physical stiffness and the robot's posture so to implement a desired Cartesian stiffness and location of the contact point. We validate the algorithm in simulation and with experiments. To perform the latter, we also present a new tendon-driven soft manipulator, equipped with variable-stiffness segments and proprioceptive sensing and capable to move in three dimensional. We show that, combining the intelligent hardware with the proposed algorithm, we can obtain the desired stiffness at the end-effector over the workspace.

Origami-based integration of robots that sense, decide, and respond.

Origami-inspired engineering has enabled intelligent materials and structures to process and react to environmental stimuli. However, it is challenging to achieve complete sense-decide-act loops in origami materials for autonomous interaction with environments, mainly due to the lack of information processing units that can interface with sensing and actuation. Here, we introduce an integrated origami-based process to create autonomous robots by embedding sensing, computing, and actuating in compliant, conductive materials. By combining flexible bistable mechanisms and conductive thermal artificial muscles, we realize origami multiplexed switches and configure them to generate digital logic gates, memory bits, and thus integrated autonomous origami robots. We demonstrate with a flytrap-inspired robot that captures 'living prey', an untethered crawler that avoids obstacles, and a wheeled vehicle that locomotes with reprogrammable trajectories. Our method provides routes to achieve autonomy for origami robots through tight functional integration in compliant, conductive materials.

Robust flight navigation out of distribution with liquid neural networks.

Autonomous robots can learn to perform visual navigation tasks from offline human demonstrations and generalize well to online and unseen scenarios within the same environment they have been trained on. It is challenging for these agents to take a step further and robustly generalize to new environments with drastic scenery changes that they have never encountered. Here, we present a method to create robust flight navigation agents that successfully perform vision-based fly-to-target tasks beyond their training environment under drastic distribution shifts. To this end, we designed an imitation learning framework using liquid neural networks, a brain-inspired class of continuous-time neural models that are causal and adapt to changing conditions. We observed that liquid agents learn to distill the task they are given from visual inputs and drop irrelevant features. Thus, their learned navigation skills transferred to new environments. When compared with several other state-of-the-art deep agents, experiments showed that this level of robustness in decision-making is exclusive to liquid networks, both in their differential equation and closed-form representations.

Managing United States-China university relations and risks.

US universities must proactively address potential concerns.

TransCenter: Transformers With Dense Representations for Multiple-Object Tracking.

Transformers have proven superior performance for a wide variety of tasks since they were introduced. In recent years, they have drawn attention from the vision community in tasks such as image classification and object detection. Despite this wave, an accurate and efficient multiple-object tracking (MOT) method based on transformers is yet to be designed. We argue that the direct application of a transformer architecture with quadratic complexity and insufficient noise-initialized sparse queries - is not optimal for MOT. We propose TransCenter, a transformer-based MOT architecture with dense representations for accurately tracking all the objects while keeping a reasonable runtime. Methodologically, we propose the use of image-related dense detection queries and efficient sparse tracking queries produced by our carefully designed query learning networks (QLN). On one hand, the dense image-related detection queries allow us to infer targets' locations globally and robustly through dense heatmap outputs. On the other hand, the set of sparse tracking queries efficiently interacts with image features in our TransCenter Decoder to associate object positions through time. As a result, TransCenterexhibits remarkable performance improvements and outperforms by a large margin the current state-of-the-art methods in two standard MOT benchmarks with two tracking settings (public/private). TransCenter is also proven efficient and accurate by an extensive ablation study and, comparisons to more naive alternatives and concurrent works. The code is made publicly available at https://github.com/yihongxu/transcenter.

Evaluating Autonomous Urban Perception and Planning in a 1/10th Scale MiniCity.

We present the MiniCity, a multi-vehicle evaluation platform for testing perception hardware and software for autonomous vehicles. The MiniCity is a 1/10th scale city consisting of realistic urban scenery, intersections, and multiple fully autonomous 1/10th scale vehicles with state-of-the-art sensors and algorithms. The MiniCity is used to evaluate and test perception algorithms both upstream and downstream in the autonomy stack, in urban driving scenarios such as occluded intersections and avoiding multiple vehicles. We demonstrate the MiniCity's ability to evaluate different sensor and algorithm configurations for perception tasks such as object detection and localization. For both tasks, the MiniCity platform is used to evaluate the task itself (accuracy in estimating obstacle pose and ego pose in the map) as well as the downstream performance in collision avoidance and lane following, respectively.

Fluidic innervation sensorizes structures from a single build material.

Multifunctional materials with distributed sensing and programmed mechanical properties are required for myriad emerging technologies. However, current fabrication techniques constrain these materials' design and sensing capabilities. We address these needs with a method for sensorizing architected materials through fluidic innervation, where distributed networks of empty, air-filled channels are directly embedded within an architected material's sparse geometry. By measuring pressure changes within these channels, we receive feedback regarding material deformation. Thus, this technique allows for three-dimensional printing of sensorized structures from a single material. With this strategy, we fabricate sensorized soft robotic actuators on the basis of handed shearing auxetics and accurately predict their kinematics from the sensors' proprioceptive feedback using supervised learning. Our strategy for facilitating structural, sensing, and actuation capabilities through control of form alone simplifies sensorized material design for applications spanning wearables, smart structures, and robotics.

Toward understanding the communication in sperm whales.

Machine learning has been advancing dramatically over the past decade. Most strides are human-based applications due to the availability of large-scale datasets; however, opportunities are ripe to apply this technology to more deeply understand non-human communication. We detail a scientific roadmap for advancing the understanding of communication of whales that can be built further upon as a template to decipher other forms of animal and non-human communication. Sperm whales, with their highly developed neuroanatomical features, cognitive abilities, social structures, and discrete click-based encoding make for an excellent model for advanced tools that can be applied to other animals in the future. We outline the key elements required for the collection and processing of massive datasets, detecting basic communication units and language-like higher-level structures, and validating models through interactive playback experiments. The technological capabilities developed by such an undertaking hold potential for cross-applications in broader communities investigating non-human communication and behavioral research.

Identifying and Mitigating Potential Biases in Predicting Drug Approvals.

INTRODUCTION: Machine learning models are increasingly applied to predict the drug development outcomes based on intermediary clinical trial results. A key challenge to this task is to address various forms of bias in the historical drug approval data. OBJECTIVE: We aimed to identify and mitigate the bias in drug approval predictions and quantify the impacts of debiasing in terms of financial value and drug safety. METHODS: We instantiated the Debiasing Variational Autoencoder, the state-of-the-art model for automated debiasing. We trained and evaluated the model on the Citeline dataset provided by Informa Pharma Intelligence to predict the final drug development outcome from phase II trial results. RESULTS: The debiased Debiasing Variational Autoencoder model achieved better performance (measured by the [Formula: see text] score 0.48) in predicting the drug development outcomes than its un-debiased baseline ([Formula: see text] score 0.25). It had a much higher true-positive rate than baseline (60% vs 15%), while its true-negative rate was slightly lower (88% vs 99%). The Debiasing Variational Autoencoder distinguished between drugs developed by large pharmaceutical firms and those by small biotech companies. The model prediction is strongly influenced by multiple factors such as prior approval of the drug for another indication, whether the trial meets the positive/negative endpoints, and the year when the trial is completed. We estimate that the debiased model generates financial value for the drug developer in six major therapeutic areas, with a range of US$763-1,365 million. CONCLUSIONS: Our analysis shows that debiasing improves the financial efficiency of late-stage drug development. From the pharmacovigilance perspective, the debiased model is more likely to identify drugs that are both safe and effective. Meanwhile, it may predict a higher probability of success for drugs with potential adverse effects (because of its lower true-negative rate), thus it must be used with caution to predict the development outcomes of drug candidates currently in the pipeline.

Estimating the state of epidemics spreading with graph neural networks.

When an epidemic spreads into a population, it is often impractical or impossible to continuously monitor all subjects involved. As an alternative, we propose using algorithmic solutions that can infer the state of the whole population from a limited number of measures. We analyze the capability of deep neural networks to solve this challenging task. We base our proposed architecture on Graph Convolutional Neural Networks. As such, it can reason on the effect of the underlying social network structure, which is recognized as the main component in spreading an epidemic. The proposed architecture can reconstruct the entire state with accuracy above 70%, as proven by two scenarios modeled on the CoVid-19 pandemic. The first is a generic homogeneous population, and the second is a toy model of the Boston metropolitan area. Note that no retraining of the architecture is necessary when changing the model.

Scaling Up Soft Robotics: A Meter-Scale, Modular, and Reconfigurable Soft Robotic System.

Today's use of large-scale industrial robots is enabling extraordinary achievement on the assembly line, but these robots remain isolated from the humans on the factory floor because they are very powerful, and thus dangerous to be around. In contrast, the soft robotics research community has proposed soft robots that are safe for human environments. The current state of the art enables the creation of small-scale soft robotic devices. In this article we address the gap between small-scale soft robots and the need for human-sized safe robots by introducing a new soft robotic module and multiple human-scale robot configurations based on this module. We tackle large-scale soft robots by presenting a modular and reconfigurable soft robotic platform that can be used to build fully functional and untethered meter-scale soft robots. These findings indicate that a new wave of human-scale soft robots can be an alternative to classic rigid-bodied robots in tasks and environments where humans and machines can work side by side with capabilities that include, but are not limited to, autonomous legged locomotion and grasping.

MIT Emergency-Vent: An Automated Resuscitator Bag for the COVID-19 Crisis.

MIT's Emergency-Vent Project was launched in March 2020 to develop safe guidance and a reference design for a bridge ventilator that could be rapidly produced in a distributed manner worldwide. The system uses a novel servo-based robotic gripper to automate the squeezing of a manual resuscitator bag evenly from both sides to provide ventilation according to clinically specified parameters. In just one month, the team designed and built prototype ventilators, tested them in a series of porcine trials, and collaborated with industry partners to enable mass production. We released the design, including mechanical drawings, design spreadsheets, circuit diagrams, and control code into an open source format and assisted production efforts worldwide.Clinical relevance- This work demonstrated the viability of automating the compression of a manual resuscitator bag, with pressure feedback, to provide bridge ventilation support.

Evidential Deep Learning for Guided Molecular Property Prediction and Discovery.

While neural networks achieve state-of-the-art performance for many molecular modeling and structure-property prediction tasks, these models can struggle with generalization to out-of-domain examples, exhibit poor sample efficiency, and produce uncalibrated predictions. In this paper, we leverage advances in evidential deep learning to demonstrate a new approach to uncertainty quantification for neural network-based molecular structure-property prediction at no additional computational cost. We develop both evidential 2D message passing neural networks and evidential 3D atomistic neural networks and apply these networks across a range of different tasks. We demonstrate that evidential uncertainties enable (1) calibrated predictions where uncertainty correlates with error, (2) sample-efficient training through uncertainty-guided active learning, and (3) improved experimental validation rates in a retrospective virtual screening campaign. Our results suggest that evidential deep learning can provide an efficient means of uncertainty quantification useful for molecular property prediction, discovery, and design tasks in the chemical and physical sciences.

Covid-19 and Flattening the Curve: A Feedback Control Perspective.

Many of the policies that were put into place during the Covid-19 pandemic had a common goal: to flatten the curve of the number of infected people so that its peak remains under a critical threshold. This letter considers the challenge of engineering a strategy that enforces such a goal using control theory. We introduce a simple formulation of the optimal flattening problem, and provide a closed form solution. This is augmented through nonlinear closed loop tracking of the nominal solution, with the aim of ensuring close-to-optimal performance under uncertain conditions. A key contribution of this letter is to provide validation of the method with extensive and realistic simulations in a Covid-19 scenario, with particular focus on the case of Codogno - a small city in Northern Italy that has been among the most harshly hit by the pandemic.