Suchergebnisse

AbstractEarly warning indicators often suffer from the shortness and coarse-graining of real-world time series. Furthermore, the typically strong and correlated noise contributions in real applications are severe drawbacks for statistical measures. Even under favourable simulation conditions the measures are of limited capacity due to their qualitative nature and sometimes ambiguous trend-to-noise ratio. In order to solve these shortcomings, we analyze the stability of the system via the slope of the deterministic term of a Langevin equation, which is hypothesized to underlie the system dynamics close to the fixed point. The open-source available method is applied to a previously studied seasonal ecological model under noise levels and correlation scenarios commonly observed in real world data. We compare the results to autocorrelation, standard deviation, skewness, and kurtosis as leading indicator candidates by a Bayesian model comparison with a linear and a constant model. We show that the slope of the deterministic term is a promising alternative due to its quantitative nature and high robustness against noise levels and types. The commonly computed indicators apart from the autocorrelation with deseasonalization fail to provide reliable insights into the stability of the system in contrast to a previously performed study in which the standard deviation was found to perform best. In addition, we discuss the significant influence of the seasonal nature of the data to the robust computation of the various indicators, before we determine approximately the minimal amount of data per time window that leads to significant trends for the drift slope estimations.

Abstract
Network structures underlie the dynamics of many complex phenomena, from gene regulation and foodwebs to power grids and social media. Yet, as they often cannot be observed directly, their connectivities must be inferred from observations of the dynamics to which they give rise. In this work, we present a powerful computational method to infer large network adjacency matrices from time series data using a neural network, in order to provide uncertainty quantification on the prediction in a manner that reflects both the degree to which the inference problem is underdetermined as well as the noise on the data. This is a feature that other approaches have hitherto been lacking. We demonstrate our method's capabilities by inferring line failure locations in the British power grid from its response to a power cut, providing probability densities on each edge and allowing the use of hypothesis testing to make meaningful probabilistic statements about the location of the cut. Our method is significantly more accurate than both Markov-chain Monte Carlo sampling and least squares regression on noisy data and when the problem is underdetermined, while naturally extending to the case of nonlinear dynamics, which we demonstrate by learning an entire cost matrix for a nonlinear model of economic activity in Greater London. Not having been specifically engineered for network inference, this method in fact represents a general parameter estimation scheme that is applicable to any high-dimensional parameter space.

Abstract
In recent years, advances in computing hardware and computational methods have prompted a wealth of activities for solving inverse problems in physics. These problems are often described by systems of partial differential equations (PDEs). The advent of machine learning has reinvigorated the interest in solving inverse problems using neural networks (NNs). In these efforts, the solution of the PDEs is expressed as NNs trained through the minimization of a loss function involving the PDE. Here, we show how to accelerate this approach by five orders of magnitude by deploying, instead of NNs, conventional PDE approximations. The framework of optimizing a discrete loss (ODIL) minimizes a cost function for discrete approximations of the PDEs using gradient-based and Newton's methods. The framework relies on grid-based discretizations of PDEs and inherits their accuracy, convergence, and conservation properties. The implementation of the method is facilitated by adopting machine-learning tools for automatic differentiation. We also propose a multigrid technique to accelerate the convergence of gradient-based optimizers. We present applications to PDE-constrained optimization, optical flow, system identification, and data assimilation. We compare ODIL with the popular method of physics-informed neural networks and show that it outperforms it by several orders of magnitude in computational speed while having better accuracy and convergence rates. We evaluate ODIL on inverse problems involving linear and nonlinear PDEs including the Navier–Stokes equations for flow reconstruction problems. ODIL bridges numerical methods and machine learning and presents a powerful tool for solving challenging, inverse problems across scientific domains.

AbstractHuman vision, thought, and planning involve parsing and representing objects and scenes using structured representations based on part-whole hierarchies. Computer vision and machine learning researchers have recently sought to emulate this capability using neural networks, but a generative model formulation has been lacking. Generative models that leverage compositionality, recursion, and part-whole hierarchies are thought to underlie human concept learning and the ability to construct and represent flexible mental concepts. We introduce Recursive Neural Programs (RNPs), a neural generative model that addresses the part-whole hierarchy learning problem by modeling images as hierarchical trees of probabilistic sensory-motor programs. These programs recursively reuse learned sensory-motor primitives to model an image within different spatial reference frames, enabling hierarchical composition of objects from parts and implementing a grammar for images. We show that RNPs can learn part-whole hierarchies for a variety of image datasets, allowing rich compositionality and intuitive parts-based explanations of objects. Our model also suggests a cognitive framework for understanding how human brains can potentially learn and represent concepts in terms of recursively defined primitives and their relations with each other.

Abstract
Networks offer a powerful approach to modeling complex systems by representing the underlying set of pairwise interactions. Link prediction is the task that predicts links of a network that are not directly visible, with profound applications in biological, social, and other complex systems. Despite intensive utilization of the topological feature in this task, it is unclear to what extent a feature can be leveraged to infer missing links. Here, we aim to unveil the capability of a topological feature in link prediction by identifying its prediction performance upper bound. We introduce a theoretical framework that is compatible with different indexes to gauge the feature, different prediction approaches to utilize the feature, and different metrics to quantify the prediction performance. The maximum capability of a topological feature follows a simple yet theoretically validated expression, which only depends on the extent to which the feature is held in missing and nonexistent links. Because a family of indexes based on the same feature shares the same upper bound, the potential of all others can be estimated from one single index. Furthermore, a feature's capability is lifted in the supervised prediction, which can be mathematically quantified, allowing us to estimate the benefit of applying machine learning algorithms. The universality of the pattern uncovered is empirically verified by 550 structurally diverse networks. The findings have applications in feature and method selection, and shed light on network characteristics that make a topological feature effective in link prediction.

Abstract
We consider the looming threat of bad actors using artificial intelligence (AI)/Generative Pretrained Transformer to generate harms across social media globally. Guided by our detailed mapping of the online multiplatform battlefield, we offer answers to the key questions of what bad-actor-AI activity will likely dominate, where, when—and what might be done to control it at scale. Applying a dynamical Red Queen analysis from prior studies of cyber and automated algorithm attacks, predicts an escalation to daily bad-actor-AI activity by mid-2024—just ahead of United States and other global elections. We then use an exactly solvable mathematical model of the observed bad-actor community clustering dynamics, to build a Policy Matrix which quantifies the outcomes and trade-offs between two potentially desirable outcomes: containment of future bad-actor-AI activity vs. its complete removal. We also give explicit plug-and-play formulae for associated risk measures.

Abstract
Avoiding many of the most severe consequences of anthropogenic climate change in the coming century will very likely require the development of "negative emissions technologies"—practices that lead to net carbon dioxide removal (CDR) from Earth's atmosphere. However, feedbacks within the carbon cycle place intrinsic limits on the long-term impact of CDR on atmospheric CO2 that are likely to vary across CDR technologies in ways that are poorly constrained. Here, we use an ensemble of Earth system models to provide new insights into the efficiency of CDR through enhanced rock weathering (ERW) by explicitly quantifying long-term storage of carbon in the ocean during ERW relative to an equivalent modulated emissions scenario. We find that although the backflux of CO2 to the atmosphere in the face of CDR is in all cases significant and time-varying, even for direct removal and underground storage, the leakage of initially captured carbon associated with ERW is well below that currently assumed. In addition, net alkalinity addition to the surface ocean from ERW leads to significant increases in seawater carbonate mineral saturation state relative to an equivalent emissions trajectory, a co-benefit for calcifying marine organisms. These results suggest that potential carbon leakage from the oceans during ERW is a small component of the overall ERW life cycle and that it can be rigorously quantified and incorporated into technoeconomic assessments of ERW at scale.

Abstract
Spontaneous synchronization is ubiquitous in natural and man-made systems. It underlies emergent behaviors such as neuronal response modulation and is fundamental to the coordination of robot swarms and autonomous vehicle fleets. Due to its simplicity and physical interpretability, pulse-coupled oscillators has emerged as one of the standard models for synchronization. However, existing analytical results for this model assume ideal conditions, including homogeneous oscillator frequencies and negligible coupling delays, as well as strict requirements on the initial phase distribution and the network topology. Using reinforcement learning, we obtain an optimal pulse-interaction mechanism (encoded in phase response function) that optimizes the probability of synchronization even in the presence of nonideal conditions. For small oscillator heterogeneities and propagation delays, we propose a heuristic formula for highly effective phase response functions that can be applied to general networks and unrestricted initial phase distributions. This allows us to bypass the need to relearn the phase response function for every new network.

Abstract
Solar cells are a promising optoelectronic device for the simultaneous solution of energy resource and environmental problems. However, their high cost and slow, laborious production process so far severely hinder a sufficient widespread of clean, renewable photovoltaic energy as a major alternative electricity generator. This undesirable situation is mainly attributed to the fact that photovoltaic devices have been manufactured through a series of vacuum and high-temperature processes. Here we realize a poly(3,4-ethylenedioxythiophene)–poly(styrenesulfonate) (PEDOT:PSS)/Si heterojunction solar cell fabricated only in ambient and room temperature conditions from a plain Si wafer, with an over 10% energy conversion efficiency. Our production scheme is based on our finding that PEDOT:PSS photovoltaic layers actively operate even on highly doped Si substrates, which substantially mitigates the condition requirements for electrode implementation. Our approach may pave the way for facile, low-cost, high-throughput solar cell fabrication, useful in various fields even including developing countries and educational sites.

Abstract
The detection of elementary carriers in transport phenomena is one of the most important keys to understand nontrivial properties of strongly correlated quantum matter. Here, we propose a method to identify the tunneling current carrier in strongly interacting fermions from nonequilibrium noise in the Bardeen–Cooper–Schrieffer to Bose–Einstein condensate crossover. The noise-to-current ratio, the Fano factor, can be a crucial probe for the current carrier. Bringing strongly correlated fermions into contact with a dilute reservoir produces a tunneling current in between. The associated Fano factor increases from one to two as the interaction becomes stronger, reflecting the fact that the dominant conduction channel changes from the quasiparticle tunneling to the pair tunneling.

Abstract
Decomposing a graph into groups of nodes that share similar connectivity properties is essential to understand the organization and function of complex networks. Previous works have focused on groups with specific relationships between group members, such as assortative communities or core–periphery structures, developing computational methods to find these mesoscale structures within a network. Here, we go beyond these two traditional cases and introduce a methodology that is able to identify and systematically classify all possible community types in directed multi graphs, based on the pairwise relationship between groups. We apply our approach to 53 different networks and find that assortative communities are the most common structures, but that previously unexplored types appear in almost every network. A particularly prevalent new type of relationship, which we call a source–basin structure, has information flowing from a sparsely connected group of nodes (source) to a densely connected group (basin). We look in detail at two online social networks—a new network of Twitter users and a well-studied network of political blogs—and find that source–basin structures play an important role in both of them. This confirms not only the widespread appearance of nonassortative structures but also the potential of hitherto unidentified relationships to explain the organization of complex networks.

Abstract
The honey bee waggle dance is one of the most prominent examples of abstract communication among animals: successful foragers convey new resource locations to interested followers via characteristic "dance" movements in the nest, where dances advertise different locations on different overlapping subregions of the "dance floor." To this day, this spatial separation has not been described in detail, and it remains unknown how it affects the dance communication. Here, we evaluate long-term recordings of Apis mellifera foraging at natural and artificial food sites. Using machine learning, we detect and decode waggle dances, and we individually identify and track dancers and dance followers in the hive and at artificial feeders. We record more than a hundred thousand waggle phases, and thousands of dances and dance-following interactions to quantitatively describe the spatial separation of dances on the dance floor. We find that the separation of dancers increases throughout a dance and present a motion model based on a positional drift of the dancer between subsequent waggle phases that fits our observations. We show that this separation affects follower bees as well and results in them more likely following subsequent dances to similar food source locations, constituting a positive feedback loop. Our work provides evidence that the positional drift between subsequent waggle phases modulates the information that is available to dance followers, leading to an emergent optimization of the waggle dance communication system.

Abstract
Brain–computer interfaces (BCI) using electroencephalography provide a noninvasive method for users to interact with external devices without the need for muscle activation. While noninvasive BCIs have the potential to improve the quality of lives of healthy and motor-impaired individuals, they currently have limited applications due to inconsistent performance and low degrees of freedom. In this study, we use deep learning (DL)-based decoders for online continuous pursuit (CP), a complex BCI task requiring the user to track an object in 2D space. We developed a labeling system to use CP data for supervised learning, trained DL-based decoders based on two architectures, including a newly proposed adaptation of the PointNet architecture, and evaluated the performance over several online sessions. We rigorously evaluated the DL-based decoders in a total of 28 human participants, and found that the DL-based models improved throughout the sessions as more training data became available and significantly outperformed a traditional BCI decoder by the last session. We also performed additional experiments to test an implementation of transfer learning by pretraining models on data from other subjects, and midsession training to reduce intersession variability. The results from these experiments showed that pretraining did not significantly improve performance, but updating the models' midsession may have some benefit. Overall, these findings support the use of DL-based decoders for improving BCI performance in complex tasks like CP, which can expand the potential applications of BCI devices and help to improve the quality of lives of healthy and motor-impaired individuals.