Article
|
Open Access
Featured
-
-
Article
| Open AccessComputationally efficient design of directionally compliant metamaterials
Designing mechanical metamaterials is challenging because of the large number of non-periodic constituent elements. Here, the authors develop an approach to design arbitrarily shaped metamaterials that is more computationally efficient by six orders of magnitude compared to other approaches.
- Lucas A. Shaw
- , Frederick Sun
- & Jonathan B. Hopkins
-
Article
| Open AccessMachine learning plastic deformation of crystals
Predicting plastic deformation in crystals remains challenging owing to the nonlinear nature of stochastic avalanches involved, which resemble the critical phenomena. Salmenjoki et al. use machine learning to predict plastic deformation and show that it works better for those under large strains.
- Henri Salmenjoki
- , Mikko J. Alava
- & Lasse Laurson
-
Article
| Open AccessThe preeminence of ethnic diversity in scientific collaboration
Diversity is believed to raise effectiveness and performance but it contains many aspects. Here the authors studied the relationship between research impact and five classes of diversity and found that ethnic diversity had the strongest correlation with scientific impact.
- Bedoor K. AlShebli
- , Talal Rahwan
- & Wei Lee Woon
-
Article
| Open AccessPredicting natural language descriptions of mono-molecular odorants
It is now possible to predict what a chemical smells like based on its chemical structure, however to date, this has only been done for a small number of odor descriptors. Here, using natural-language semantic representations, the authors demonstrate prediction of a much wider range of descriptors.
- E. Darío Gutiérrez
- , Amit Dhurandhar
- & Guillermo A. Cecchi
-
Article
| Open AccessA continuous-time MaxSAT solver with high analog performance
Continuous-time computation paradigm could represent a viable alternative to the standard digital one when dealing with certain classes of problems. Here, the authors propose a generalised version of a continuous-time solver and simulate its performances in solving MaxSAT and two-colour Ramsey problems.
- Botond Molnár
- , Ferenc Molnár
- & Mária Ercsey-Ravasz
-
Article
| Open AccessPathway-based subnetworks enable cross-disease biomarker discovery
Accurate and actionable biomarkers that integrate diverse molecular, functional and clinical information hold great promise in precision medicine. Here, the authors develop SIMMS, a method for pathway-based cross-disease biomarker discovery.
- Syed Haider
- , Cindy Q. Yao
- & Paul C. Boutros
-
Article
| Open AccessAn interpretable approach for social network formation among heterogeneous agents
Complex networks can be a useful tool to investigate problems in social science. Here the authors use game theory to establish a network model and then use a machine learning approach to characterize the role of nodes within a social network.
- Yuan Yuan
- , Ahmad Alabdulkareem
- & Alex ‘Sandy’ Pentland
-
Article
| Open AccessChemical shifts in molecular solids by machine learning
Solid-state nuclear magnetic resonance combined with quantum chemical shift predictions is limited by high computational cost. Here, the authors use machine learning based on local atomic environments to predict experimental chemical shifts in molecular solids with accuracy similar to density functional theory.
- Federico M. Paruzzo
- , Albert Hofstetter
- & Lyndon Emsley
-
Article
| Open AccessMaster clinical medical knowledge at certificated-doctor-level with deep learning model
AI is used increasingly in medical diagnostics. Here, the authors present a deep learning model that masters medical knowledge, demonstrated by it having passed the written test of the 2017 National Medical Licensing Examination in China, and can provide help with clinical diagnosis based on electronic health care records.
- Ji Wu
- , Xien Liu
- & Ping Lv
-
Article
| Open AccessMultiparameter optimisation of a magneto-optical trap using deep learning
Dynamics in cold atomic ensembles involve complex many-body interactions that are hard to treat analytically. Here, the authors use machine learning to optimise the cooling and trapping of neutral atoms, showing an improvement in the resulting resonant optical depth compared to more traditional solutions.
- A. D. Tranter
- , H. J. Slatyer
- & G. T. Campbell
-
Article
| Open AccessQuantum machine learning for electronic structure calculations
With the rapid development of quantum computers, quantum machine learning approaches are emerging as powerful tools to perform electronic structure calculations. Here, the authors develop a quantum machine learning algorithm, which demonstrates significant improvements in solving quantum many-body problems.
- Rongxin Xia
- & Sabre Kais
-
Article
| Open AccessEvidence of a turbulent ExB mixing avalanche mechanism of gas breakdown in strongly magnetized systems
Gas breakdown mechanism in plasma under the influence of complex electromagnetic field topology is still debatable. Here the authors present the evidence of the E×B mixing avalanche for gas breakdown in magnetized plasmas in fusion devices as tokamak.
- Min-Gu Yoo
- , Jeongwon Lee
- & Yong-Su Na
-
Article
| Open AccessSimulations tackle abrupt massive migrations of energetic beam ions in a tokamak plasma
Understanding the occurrence of sudden changes in plasma parameters is important for the operation of magnetically confined fusion devices. Here the authors use simulation to shed light on the formation of abrupt large-amplitude events and the associated redistribution of energetic ions in a tokamak.
- Andreas Bierwage
- , Kouji Shinohara
- & Masatoshi Yagi
-
Article
| Open AccessA model for super El Niños
Despite advances in ENSO modeling, super El Niño events remain largely unpredictable. Hameed et al. postulate that ENSO-IOD interaction is crucial for super El Niño development and identify a self-limiting factor that constrains ENSO dynamics from generating these extreme events on their own.
- Saji N. Hameed
- , Dachao Jin
- & Vishnu Thilakan
-
Article
| Open AccessNeuromorphic computing with multi-memristive synapses
Memristive technology is a promising avenue towards realizing efficient non-von Neumann neuromorphic hardware. Boybat et al. proposes a multi-memristive synaptic architecture with a counter-based global arbitration scheme to address challenges associated with the non-ideal memristive device behavior.
- Irem Boybat
- , Manuel Le Gallo
- & Evangelos Eleftheriou
-
Article
| Open AccessConcurrence of form and function in developing networks and its role in synaptic pruning
How structure and function coevolve in developing brains is little understood. Here, the authors study a coupled model of network development and memory, and find that due to the feedback networks with some initial memory capacity evolve into heterogeneous structures with high memory performance.
- Ana P. Millán
- , J. J. Torres
- & J Marro
-
Article
| Open AccessToward a universal decoder of linguistic meaning from brain activation
Previous work decoding linguistic meaning from imaging data has generally been limited to a small number of semantic categories. Here, authors show that a decoder trained on neuroimaging data of single concepts sampling the semantic space can robustly decode meanings of semantically diverse new sentences with topics not encountered during training.
- Francisco Pereira
- , Bin Lou
- & Evelina Fedorenko
-
Article
| Open AccessGeneralized leaky integrate-and-fire models classify multiple neuron types
Simplified neuron models, such as generalized leaky integrate-and-fire (GLIF) models, are extensively used in network modeling. Here the authors systematically generate and compare GLIF models of varying complexity for their ability to classify cell types in the Allen Cell Types Database and faithfully reproduce spike trains.
- Corinne Teeter
- , Ramakrishnan Iyer
- & Stefan Mihalas
-
Article
| Open AccessStatistical modeling of RNA structure profiling experiments enables parsimonious reconstruction of structure landscapes
Different experimental and computational approaches can be used to study RNA structures. Here, the authors present a computational method for data-directed reconstruction of complex RNA structure landscapes, which predicts a parsimonious set of co-existing structures and estimates their abundances from structure profiling data.
- Hua Li
- & Sharon Aviran
-
Article
| Open AccessTo infinity and some glimpses of beyond
Certain physical problems such as the rupture of a thin sheet can be difficult to solve as computations breakdown at the point of rupture. Here the authors propose a regularization approach to overcome this breakdown which could help dealing with mathematical models that have finite time singularities.
- Panayotis G. Kevrekidis
- , Constantinos I. Siettos
- & Yannis G. Kevrekidis
-
Article
| Open AccessTemporal correlation detection using computational phase-change memory
New computing paradigms, such as in-memory computing, are expected to overcome the limitations of conventional computing approaches. Sebastian et al. report a large-scale demonstration of computational phase change memory (PCM) by performing high-level computational primitives using one million PCM devices.
- Abu Sebastian
- , Tomas Tuma
- & Evangelos Eleftheriou
-
Article
| Open AccessAn autonomous organic reaction search engine for chemical reactivity
While automated reaction systems typically work for the synthesis of pre-defined molecules, automated systems to discover reactivity are more challenging. Here the authors report an autonomous organic reaction search engine that allows discovery of the most reactive pathways in a multi-reagent, multistep reaction system.
- Vincenza Dragone
- , Victor Sans
- & Leroy Cronin
-
Article
| Open AccessChaos as an intermittently forced linear system
The huge amount of data generated in fields like neuroscience or finance calls for effective strategies that mine data to reveal underlying dynamics. Here Brunton et al.develop a data-driven technique to analyze chaotic systems and predict their dynamics in terms of a forced linear model.
- Steven L. Brunton
- , Bingni W. Brunton
- & J. Nathan Kutz
-
Article
| Open AccessRegulation of persistent sodium currents by glycogen synthase kinase 3 encodes daily rhythms of neuronal excitability
It is not clear how circadian biochemical cascades are encoded into neural electrical signals. Here, using a combination of electrophysiology and modelling approaches in mice, the authors show activation of glycogen synthase kinase 3 modulates neural activity in the suprachiasmatic nuclei via regulation of the persistent sodium current, INaP.
- Jodi R. Paul
- , Daniel DeWoskin
- & Karen L. Gamble
-
Article
| Open AccessThe backtracking survey propagation algorithm for solving random K-SAT problems
The K-satisfability problem is a combinatorial discrete optimization problem, which for K=3 is NP-complete, and whose random formulation is of interest for understanding computational complexity. Here, the authors introduce the backtracking survey propagation algorithm for studying it for K=3 and K=4.
- Raffaele Marino
- , Giorgio Parisi
- & Federico Ricci-Tersenghi
-
Article
| Open AccessSpatio-temporal propagation of cascading overload failures in spatially embedded networks
Overload failures propagate through hidden functional dependencies across networked systems. Here, the authors study the spatio-temporal propagation behaviour of cascading overload failures, and find that they spread radially from their origin with an approximately constant velocity.
- Jichang Zhao
- , Daqing Li
- & Shlomo Havlin
-
Article
| Open AccessChemical reaction mechanisms in solution from brute force computational Arrhenius plots
Obtaining activation entropies and enthalpies of a reaction is important for distinguishing between alternative reaction mechanisms. Here the authors use computational methods to accurately obtain the enthalpic/entropic components of the activation free energy for hydrolytic deamination reactions.
- Masoud Kazemi
- & Johan Åqvist
-
Article |
Ranking in interconnected multilayer networks reveals versatile nodes
A challenging problem is to identify the most central agents in interconnected multilayer networks. Here, De Domenico et al. present a mathematical framework to calculate centrality in such networks—versatility—and rank nodes accordingly.
- Manlio De Domenico
- , Albert Solé-Ribalta
- & Alex Arenas
-
Article
| Open AccessTrainable hardware for dynamical computing using error backpropagation through physical media
Machine learning systems use algorithms that can interpret data to make improved decisions. Hermans et al. develop a physical scheme for a computing system based on recurrent neural networks that physically implements the error backpropagation algorithm, thus performing its own training process.
- Michiel Hermans
- , Michaël Burm
- & Peter Bienstman
-
Article |
The theory of pattern formation on directed networks
The study of pattern formation in reaction–diffusion systems has been mainly restricted to symmetric (undirected) networks. Here, Asllani et al.identify a different pattern formation mechanism in a larger class of networks incorporating the possibility of unequal weights for transport along edges.
- Malbor Asllani
- , Joseph D. Challenger
- & Duccio Fanelli
-
Article |
Cluster synchronization and isolated desynchronization in complex networks with symmetries
Many networks exhibit patterns of synchronized clusters, but the conditions under which this occurs are poorly understood. Pecora et al. develop an analytical approach based on computational group theory to predict the emergence and disappearance of synchrony among local clusters in complex networks.
- Louis M. Pecora
- , Francesco Sorrentino
- & Rajarshi Roy
-
Article
| Open AccessExperimental demonstration of reservoir computing on a silicon photonics chip
Reservoir computing uses computational techniques related to neural networks to perform certain computing tasks. Here, the authors implement a passive optical reservoir computing scheme integrated on a silicon chip, operating at speeds up to 12.5 Gbit s−1.
- Kristof Vandoorne
- , Pauline Mechet
- & Peter Bienstman
-
Article |
Quantum computing on encrypted data
Practical quantum computers will require protocols to carry out computation on encrypted data, just like their classical counterparts. Here, the authors present such a protocol that allows an untrusted server to implement universal quantum gates on encrypted qubits without learning about the inputs.
- K. A. G. Fisher
- , A. Broadbent
- & K. J. Resch
-
Article
| Open AccessA chiral-based magnetic memory device without a permanent magnet
Most new device concepts for random-access memory are based on inorganic spin filters, which need a permanent magnet to operate. Here, the authors exploit the chiral-induced spin selectivity effect in an organic spin filter to construct a novel type of memory device, which works without a permanent magnet.
- Oren Ben Dor
- , Shira Yochelis
- & Yossi Paltiel