Article
|
Open Access
Featured
-
-
Article
| Open AccessA Deep Gravity model for mobility flows generation
The movements of individuals within and among cities influence critical aspects of our society, such as well-being, the spreading of epidemics, and the quality of the environment. Here, the authors use deep neural networks to discover non-linear relationships between geographical variables and mobility flows.
- Filippo Simini
- , Gianni Barlacchi
- & Luca Pappalardo
-
Article
| Open AccessSpectral analysis of climate dynamics with operator-theoretic approaches
The Earth’s climate system is highly complex, however it exhibits certain persistent cyclic patterns like the El Niño Southern Oscillation. The authors apply the spectral theory of dynamical systems and data science techniques to extract such coherent modes of climate variability from high-dimensional observational data.
- Gary Froyland
- , Dimitrios Giannakis
- & Joanna Slawinska
-
Article
| Open AccessAutonomous extraction of millimeter-scale deformation in InSAR time series using deep learning
A deep neural network is developed to automatically extract ground deformation from Interferometric Synthetic Aperture Radar time series. Applied to data over the North Anatolian Fault, the method can detect 2 mm deformation transients and reveals a slow earthquake twice as extensive as previously recognized.
- Bertrand Rouet-Leduc
- , Romain Jolivet
- & Claudia Hulbert
-
Article
| Open AccessSpatiotemporal prediction of COVID-19 cases using inter- and intra-county proxies of human interactions
Measurements of human interaction through proxies such as social connectedness or movement patterns have proved useful for predictive modeling of COVID-19. In this study, the authors develop a spatiotemporal machine learning model to predict county level new cases in the US using a variety of predictive features.
- Behzad Vahedi
- , Morteza Karimzadeh
- & Hamidreza Zoraghein
-
Article
| Open AccessComputational design and optimization of electro-physiological sensors
Though skin-conformable electro-physiological sensors are attractive for epidermal electronics, their optimal design remains a challenge. Here, the authors report a computational design approach for realizing multi-modal electro-physiological sensors that optimizes electrode layout design.
- Aditya Shekhar Nittala
- , Andreas Karrenbauer
- & Jürgen Steimle
-
Article
| Open AccessThe Tharsis mantle source of depleted shergottites revealed by 90 million impact craters
The ejection sites of the martian meteorites are still unknown. Here, the authors build a database of 90 million craters and show that Tharsis region is the most likely source of depleted shergottites ejected 1.1 Ma ago, thus confirming that some portions of the mantle were recently anomalously hot.
- A. Lagain
- , G. K. Benedix
- & K. Miljković
-
Article
| Open AccessCorrespondence between neuroevolution and gradient descent
Gradient-based and non-gradient-based methods for training neural networks are usually considered to be fundamentally different. The authors derive, and illustrate numerically, an analytic equivalence between the dynamics of neural network training under conditioned stochastic mutations, and under gradient descent.
- Stephen Whitelam
- , Viktor Selin
- & Isaac Tamblyn
-
Article
| Open AccessUnderstanding how Victoria, Australia gained control of its second COVID-19 wave
The state of Victoria, Australia experienced a substantial second wave of COVID-19 but brought it under control with strict non-pharmaceutical interventions. Here, the authors model the second wave in Victoria to estimate the impacts of the different interventions.
- James M. Trauer
- , Michael J. Lydeamore
- & Romain Ragonnet
-
Article
| Open AccessObjective comparison of methods to decode anomalous diffusion
Deviations from Brownian motion leading to anomalous diffusion are ubiquitously found in transport dynamics but often difficult to characterize. Here the authors compare approaches for single trajectory analysis through an open competition, showing that machine learning methods outperform classical approaches.
- Gorka Muñoz-Gil
- , Giovanni Volpe
- & Carlo Manzo
-
Article
| Open AccessPhysics-informed learning of governing equations from scarce data
Recovery of underlying governing laws or equations describing the evolution of complex systems from data can be challenging if dataset is damaged or incomplete. The authors propose a learning approach which allows to discover governing partial differential equations from scarce and noisy data.
- Zhao Chen
- , Yang Liu
- & Hao Sun
-
Article
| Open AccessIndividualised and non-contact post-mortem interval determination of human bodies using visible and thermal 3D imaging
Establishing the time since death (TSD) is vital in many forensic investigations. By combining thermometry, photogrammetry and numerical thermodynamic modelling, the TSD can be determined non-invasively for bodies of arbitrary shape and posture with an unprecedented accuracy of 0.26 h ± 1.38 h.
- Leah S. Wilk
- , Gerda J. Edelman
- & Maurice C. G. Aalders
-
Article
| Open AccessNationwide rollout reveals efficacy of epidemic control through digital contact tracing
The effectiveness of digital contact tracing for COVID-19 control remains uncertain. Here, the authors use data from the Smittestopp app, used in Norway in spring 2020, and estimate that 80% of nearby devices were detected by the app, and at least 11% of close contacts were not visible to manual contact tracing.
- Ahmed Elmokashfi
- , Joakim Sundnes
- & Olav Lysne
-
Article
| Open AccessA frequency-amplitude coordinator and its optimal energy consumption for biological oscillators
Calibrating both anomalous frequency and amplitude of biorhythm prevents physiological dysfunctions or diseases. Here, the authors propose a universal approach to design a frequency-amplitude coordinator rigorously via dynamical systems tools.
- Bo-Wei Qin
- , Lei Zhao
- & Wei Lin
-
Article
| Open AccessEmbodied intelligence via learning and evolution
The authors propose a new framework, deep evolutionary reinforcement learning, evolves agents with diverse morphologies to learn hard locomotion and manipulation tasks in complex environments, and reveals insights into relations between environmental physics, embodied intelligence, and the evolution of rapid learning.
- Agrim Gupta
- , Silvio Savarese
- & Li Fei-Fei
-
Article
| Open AccessHousehold cooking fuel estimates at global and country level for 1990 to 2030
Household air pollution derived from cooking fuels is a major source of health and environmental problems. Here, the authors provide detailed global, regional and country estimates of cooking fuel usage from 1990 to 2030 and project that 31% of people will still be mainly using polluting fuels in 2030.
- Oliver Stoner
- , Jessica Lewis
- & Heather Adair-Rohani
-
Article
| Open AccessConfronting false discoveries in single-cell differential expression
Differential expression analysis of single-cell transcriptomics allows scientists to dissect cell-type-specific responses to biological perturbations. Here, the authors show that many commonly used methods are biased and can produce false discoveries.
- Jordan W. Squair
- , Matthieu Gautier
- & Grégoire Courtine
-
Article
| Open AccessArtificial intelligence system reduces false-positive findings in the interpretation of breast ultrasound exams
Ultrasound is an important imaging modality for the detection and characterization of breast cancer, but it has been noted to have high false-positive rates. Here, the authors present an artificial intelligence system that achieves radiologist-level accuracy in identifying breast cancer in ultrasound imaging.
- Yiqiu Shen
- , Farah E. Shamout
- & Krzysztof J. Geras
-
Article
| Open AccessPeering into lunar permanently shadowed regions with deep learning
Some regions on the Moon are permanently covered in shadow and are therefore extremely difficult to see into. We develop a deep learning driven algorithm which enhances images of these regions, allowing us to see inside them with high resolution for the first time.
- V. T. Bickel
- , B. Moseley
- & M. Shirley
-
Article
| Open AccessTurn-key constrained parameter space exploration for particle accelerators using Bayesian active learning
Characterizing an unknown, complex system, like an accelerator, in multi-dimensional space is a challenging task. Here the authors report a Bayesian active learning method - Constrained Proximal Bayesian Exploration - for the characterization of a complex, constrained measurement as a function of multiple free parameters.
- Ryan Roussel
- , Juan Pablo Gonzalez-Aguilera
- & Auralee Edelen
-
Article
| Open AccessModelling the persistence and control of Rift Valley fever virus in a spatially heterogeneous landscape
Rift Valley fever is a zoonotic haemorrhagic fever with complex transmission dynamics influenced by environmental variables and animal movements. Here, the authors develop a metapopulation model incorporating these factors and use it to identify the main drivers of transmission in the Comoros archipelago.
- Warren S. D. Tennant
- , Eric Cardinale
- & Raphaëlle Métras
-
Article
| Open AccessNeutral bots probe political bias on social media
Social media platforms moderating misinformation have been accused of political bias. Here, the authors use neutral social bots to show that, while there is no strong evidence for such a bias, the content to which Twitter users are exposed depends strongly on the political leaning of early Twitter connections.
- Wen Chen
- , Diogo Pacheco
- & Filippo Menczer
-
Article
| Open AccessNext generation reservoir computing
Reservoir computers are artificial neural networks that can be trained on small data sets, but require large random matrices and numerous metaparameters. The authors propose an improved reservoir computer that overcomes these limitations and shows advantageous performance for complex forecasting tasks
- Daniel J. Gauthier
- , Erik Bollt
- & Wendson A. S. Barbosa
-
Article
| Open AccessQuantifying the unknown impact of segmentation uncertainty on image-based simulations
Image-based simulation for obtaining physical quantities is limited by the uncertainty in the underlying image segmentation. Here, the authors introduce a workflow for efficiently quantifying segmentation uncertainty and creating uncertainty distributions of the resulting physics quantities.
- Michael C. Krygier
- , Tyler LaBonte
- & Scott A. Roberts
-
Article
| Open AccessEpistatic Net allows the sparse spectral regularization of deep neural networks for inferring fitness functions
Finding a biologically-relevant inductive bias for training DNNs on large fitness landscapes is challenging. Here, the authors propose a method called Epistatic Net that improves DNN prediction accuracy and interpretation speed by integrating the knowledge that higher-order epistatic interactions are usually sparse.
- Amirali Aghazadeh
- , Hunter Nisonoff
- & Kannan Ramchandran
-
Article
| Open AccessMachine learning dismantling and early-warning signals of disintegration in complex systems
Network dismantling allows to find minimum set of units attacking which leads to system’s break down. Grassia et al. propose a deep-learning framework for dismantling of large networks which can be used to quantify the vulnerability of networks and detect early-warning signals of their collapse.
- Marco Grassia
- , Manlio De Domenico
- & Giuseppe Mangioni
-
Article
| Open AccessA pre-registered short-term forecasting study of COVID-19 in Germany and Poland during the second wave
Forecasting models have been used extensively to inform decision making during the COVID-19 pandemic. In this preregistered and prospective study, the authors evaluated 14 short-term models for Germany and Poland, finding considerable heterogeneity in predictions and highlighting the benefits of combined forecasts.
- J. Bracher
- , D. Wolffram
- & Frost Tianjian Xu
-
Article
| Open AccessDeep neural networks using a single neuron: folded-in-time architecture using feedback-modulated delay loops
Development of deep neural networks benefits from new approaches and perspectives. Stelzer et al. propose to fold a deep neural network of arbitrary size into a single neuron with multiple time-delayed feedback loops which is also of relevance for new hardware implementations and applications.
- Florian Stelzer
- , André Röhm
- & Serhiy Yanchuk
-
Article
| Open AccessSeasonal Arctic sea ice forecasting with probabilistic deep learning
Accurate seasonal forecasts of sea ice are highly valuable, particularly in the context of sea ice loss due to global warming. A new machine learning tool for sea ice forecasting offers a substantial increase in accuracy over current physics-based dynamical model predictions.
- Tom R. Andersson
- , J. Scott Hosking
- & Emily Shuckburgh
-
Article
| Open AccessDifferentiable sampling of molecular geometries with uncertainty-based adversarial attacks
Neural Networks are known to perform poorly outside of their training domain. Here the authors propose an inverse sampling strategy to train neural network potentials enabling to drive atomistic systems towards high-likelihood and high-uncertainty configurations without the need for molecular dynamics simulations.
- Daniel Schwalbe-Koda
- , Aik Rui Tan
- & Rafael Gómez-Bombarelli
-
Article
| Open AccessDynamics of moisture diffusion and adsorption in plant cuticles including the role of cellulose
The plant cuticle provides a barrier between internal leaf tissues and the environment. Here the authors develop a mathematical model of water movement through the cuticle and describe a prominent role for cellulose in controlling the dynamics of moisture diffusion and adsorption.
- E. C. Tredenick
- & G. D. Farquhar
-
Article
| Open AccessLineageOT is a unified framework for lineage tracing and trajectory inference
Lineage tracing and snapshots of transcriptional state at the single-cell level are powerful, complementary tools for studying development. Here, the authors propose a mathematical method combining lineage tracing with trajectory inference to improve our understanding of development.
- Aden Forrow
- & Geoffrey Schiebinger
-
Review Article
| Open AccessPrinciples of seed banks and the emergence of complexity from dormancy
Seed banks are generated when individuals enter a dormant state, a phenomenon that has evolved among diverse taxa, but that is also found in stem cells, brains, and tumors. Here, Lennon et al. synthesize the fundamentals of seed-bank theory and the emergence of complex patterns and dynamics in mathematics and the life sciences.
- Jay T. Lennon
- , Frank den Hollander
- & Jochen Blath
-
Article
| Open AccessMolecular-level similarity search brings computing to DNA data storage
Storage technology based on DNA is emerging as an information dense and durable medium. Here the authors use machine learning-based encoding and hybridization probes to execute similarity searches in a DNA database.
- Callista Bee
- , Yuan-Jyue Chen
- & Luis Ceze
-
Article
| Open AccessDeep learning of contagion dynamics on complex networks
Prediction of contagion dynamics is of relevance for epidemic and social complex networks. Murphy et al. propose a data-driven approach based on deep learning which allows to learn mechanisms governing network dynamics and make predictions beyond the training data for arbitrary network structures.
- Charles Murphy
- , Edward Laurence
- & Antoine Allard
-
Article
| Open AccessMobility patterns are associated with experienced income segregation in large US cities
Urban income segregation is often discussed in terms of where people live. Here, the authors show that the way people experience income segregation is also associated with their mobility patterns and the places they visit.
- Esteban Moro
- , Dan Calacci
- & Alex Pentland
-
Article
| Open AccessUnfolding the multiscale structure of networks with dynamical Ollivier-Ricci curvature
The analysis of networks and network processes can require low-dimensional representations, possible for specific structures only. The authors propose a geometric formalism which allows to unfold the mechanisms of dynamical processes propagation in various networks, relevant for control and community detection.
- Adam Gosztolai
- & Alexis Arnaudon
-
Article
| Open AccessDevelopment of a quantitative prediction algorithm for target organ-specific similarity of human pluripotent stem cell-derived organoids and cells
Quantitative methods to assess the quality of hPSC-derived organoids have not been developed. Here they present a prediction algorithm to assess the transcriptomic similarity between hPSC-derived organoids and the corresponding human target organs and perform validation on lung bud organoids, antral gastric organoids, and cardiomyocytes.
- Mi-Ok Lee
- , Su-gi Lee
- & Hyun-Soo Cho
-
Article
| Open AccessA generalizable and accessible approach to machine learning with global satellite imagery
This paper presents MOSAIKS, a system for planet-scale prediction of multiple outcomes using satellite imagery and machine learning (SIML). MOSAIKS generalizes across prediction domains and has the potential to enhance accessibility of SIML across research disciplines.
- Esther Rolf
- , Jonathan Proctor
- & Solomon Hsiang
-
Article
| Open AccessEmpirical assessment and comparison of neuro-evolutionary methods for the automatic off-line design of robot swarms
Off-line neuro-evolution produces robot swarms whose good performance in simulation does not often transfer to the real word. With an extensive empirical study, Hasselmann et al. substantiate overfitting as the dominant cause.
- Ken Hasselmann
- , Antoine Ligot
- & Mauro Birattari
-
Article
| Open AccessFast holographic scattering compensation for deep tissue biological imaging
Wavefront shaping is used to overcome scattering in biological tissues during imaging, but determining the compensation is slow. Here, the authors use holographic phase stepping interferometry, where new phase information is updated after each measurement, enabling fast improvement of the wavefront correction.
- Molly A. May
- , Nicolas Barré
- & Alexander Jesacher
-
Article
| Open AccessPredicting trends in the quality of state-of-the-art neural networks without access to training or testing data
In many machine learning applications, one uses pre-trained neural networks, having limited access to training and test data. Martin et al. show how to predict trends in the quality of such neural networks without access to this information, relevant for reproducibility, diagnostics, and validation.
- Charles H. Martin
- , Tongsu (Serena) Peng
- & Michael W. Mahoney
-
Article
| Open AccessOne-way dependent clusters and stability of cluster synchronization in directed networks
Mechanisms of cluster formation in networks with directed links differ from those in undirected networks. Lodi et al. propose a method to compute interdependencies among clusters of nodes in directed networks. They show that clusters can be one-way dependent, as found in social and neural networks.
- Matteo Lodi
- , Francesco Sorrentino
- & Marco Storace
-
Article
| Open AccessFast and strong amplifiers of natural selection
Population structure can influence the probability of and time to fixation of new mutants. Here, Tkadlec et al. demonstrate mathematically that structures that increase fixation probability necessarily slow fixation, but also identify amplifying structures with minimal reductions in fixation time.
- Josef Tkadlec
- , Andreas Pavlogiannis
- & Martin A. Nowak
-
Article
| Open AccessContinuous-capture microwave imaging
The authors present a microwave imaging system that can operate in continuous transmit-receive mode. Using an array of transmitters, a single receiver and a reconstruction matrix that correlate random time patterns with the captured signal, they demonstrate real-time imaging and tracking through a wall.
- Fabio C. S. da Silva
- , Anthony B. Kos
- & Archita Hati
-
Article
| Open AccessCorrelator convolutional neural networks as an interpretable architecture for image-like quantum matter data
Physical principles underlying machine learning analysis of quantum gas microscopy data are not well understood. Here the authors develop a neural network based approach to classify image data in terms of multi-site correlation functions and reveal the role of fourth-order correlations in the Fermi-Hubbard model.
- Cole Miles
- , Annabelle Bohrdt
- & Eun-Ah Kim
-
Article
| Open AccessReconstruction of plant–pollinator networks from observational data
Networks describe the intricate patterns of interaction occurring within ecological systems, but they are unfortunately difficult to construct from data. Here, the authors show how Bayesian statistical techniques can separate structure from noise in networks gathered in observational studies of plant-pollinator systems.
- Jean-Gabriel Young
- , Fernanda S. Valdovinos
- & M. E. J. Newman
-
Article
| Open AccessIn-silico trial of intracranial flow diverters replicates and expands insights from conventional clinical trials
In-silico trials rely on virtual populations and interventions simulated using patient-specific models and may offer a solution to lower costs. Here, the authors present the flow diverter performance assessment in-silico trial, which models the treatment of intracranial aneurysms with a flow-diverting stent.
- Ali Sarrami-Foroushani
- , Toni Lassila
- & Alejandro F. Frangi
-
Matters Arising
| Open AccessInadequate methods undermine a study of malaria, deforestation and trade
- Nikolas Kuschnig
-
Article
| Open AccessPrincipled approach to the selection of the embedding dimension of networks
Network embedding is a machine learning technique for construction of low-dimensional representations of large networks. Gu et al. propose a method for the identification of an optimal embedding dimension for the encoding of network structural information inspired by natural language processing.
- Weiwei Gu
- , Aditya Tandon
- & Filippo Radicchi