State Space Of Network Research Articles

Realized Random Graphs, with an Application to the Interbank Network

Abstract We introduce a new inferential methodology for dynamic network models driven by latent state variables. The main idea is to obtain a noisy representation of the state variables dynamics by computing a sequence of cross-sectional estimates of the network model at each point in time. The dynamic modeling of these cross-sectional estimates, that we name realized random graphs, transforms the original nonlinear state-space network model into a linear time-series model that can be easily estimated. Under the assumption of a mixed-membership blockmodel structure, the model parameters and the unobservable state variables can be consistently estimated when both the size of the network and the time-series length are large. By allowing for an extremely rich parameterization of the model in high dimensions, the proposed methodology describes the heterogeneous topology of real-world networks. We corroborate our findings by using this novel framework to estimate and forecast the dynamic common factors driving the evolution of the Italian electronic market of interbank deposits, and we show that the interbank lending rate is a key factor determining the network topology.

Опыт внедрения информационной системы с технологиями искусственного интеллекта в работу поликлиники для оптимизации организации периодических медицинских осмотров.

Organizing medical examinations for patients with minimal time losses is the most difficult task for an outpatient clinic when creating a New model of a medical organization providing primary health care. Since during medical examinations it is necessary to perform a whole range of works associated with certain sets of actions, the authors developed an information system with technologies of artificial intelligence, which was tested during periodic medical examinations. The purpose of the study is to develop and test an information system with technologies of artificial intelligence that allows balancing the capabilities of a medical organization and the needs of the patient when undergoing periodic medical examinations. Materials and methods. The tool for solving this problem was an artificial intelligence platform based on the principle of decision making by controlling the positions of attractors in the state space of the Hopfield neural network, which was built into the medical information system using HL7 FHIR. Results. The result of the work was the “Smart Clinic” project, which provides for the creation of an information system with technologies of artificial intelligence for routing patients during periodic medical examinations. The information system provided dynamic routing of patients in real time with the construction of the shortest trajectory and minimal movement across the floors of the building, based on current information about the workload of specialists and the duration of each stage of the examination. Conclusions. The use of an artificial intelligence information system in the outpatient clinic made it possible to reduce the average duration of a patient’s trajectory during a periodic medical examination by 45%, the average waiting time for an appointment – from 7 minutes to 3 minutes, and the total time for undergoing a preventive examination – from 200 to 150 minutes. This made it possible to increase the number of periodic medical examinations per year by 1.5 times, and the level of patient satisfaction reached 95%.

A new dynamic state estimation method for distribution networks based on modified SVSF considering photovoltaic power prediction

The fluctuations brought by the renewable energy access to the distribution network make it difficult to accurately describe the state space model of the distribution network’s dynamic process, which is the basis of the existing dynamic state estimation methods such as the Kalman filter. The inaccurate state space model directly causes an error of dynamic state estimation results. This paper proposed a new dynamic state estimation method which can mitigates the impact of renewable energy fluctuation by considering PV power prediction in establishing distribution network state space model. Firstly, the proposed method mitigates the impact of renewable energy fluctuation by considering PV power prediction in establishing distribution network state space model. Secondly, SVSF filter is introduced to achieve more accurate estimation under noise. The case study and evaluations are carried out based on MATLAB simulation. The results prove that the smooth variable structure filter with photovoltaic power prediction has a better dynamic state estimation effect under the fluctuation of the distribution network compared with the existing Kalman filter.

GatekeepR: an R Shiny application for the identification of nodes with high dynamic impact in Boolean networks.

Boolean networks can serve as straightforward models for understanding processes such as gene regulation, and employing logical rules. These rules can either be derived from existing literature or by data-driven approaches. However, in the context of large networks, the exhaustive search for intervention targets becomes challenging due to the exponential expansion of a Boolean network's state space and the multitude of potential target candidates, along with their various combinations. Instead, we can employ the logical rules and resultant interaction graph as a means to identify targets of specific interest within larger-scale models. This approach not only facilitates the screening process but also serves as a preliminary filtering step, enabling the focused investigation of candidates that hold promise for more profound dynamic analysis. However, applying this method requires a working knowledge of R, thus restricting the range of potential users. We, therefore, aim to provide an application that makes this method accessible to a broader scientific community. Here, we introduce GatekeepR, a graphical, web-based R Shiny application that enables scientists to screen Boolean network models for possible intervention targets whose perturbation is likely to have a large impact on the system's dynamics. This application does not require a local installation or knowledge of R and provides the suggested targets along with additional network information and visualizations in an intuitive, easy-to-use manner. The Supplementary Material describes the underlying method for identifying these nodes along with an example application in a network modeling pancreatic cancer. https://www.github.com/sysbio-bioinf/GatekeepR https://abel.informatik.uni-ulm.de/shiny/GatekeepR/.

Non-linear dynamic state-space network modeling for decoding neurodegeneration.

Non-linear dynamic state-space network modeling for decoding neurodegeneration.

Self-Localization Using Trajectory Attractors in Outdoor Environments

Self-localization in probabilistic robotics requires detailed, geographically consistent environmental maps, which increases the computational cost. In this study, we propose a simple self-localization method that does not require such maps. In the proposed method, the order structure, such as the mobile robot’s navigation route, is embedded as trajectory attractors in the state space of a nonmonotone neural network, and self-position estimation is performed by processing based on the autonomous dynamics of the network. From experiments, we demonstrated the basic performance of the proposed method, including robust self-localization in complex outdoor environments. Furthermore, self-localization is possible on multiple courses with overlapping paths by suitably varying the network dynamics based on environmental information. While issues remain, this study points to the great potential of neurodynamics-based robotic self-localization.

Exploring strategy differences between humans and monkeys with recurrent neural networks.

Animal models are used to understand principles of human biology. Within cognitive neuroscience, non-human primates are considered the premier model for studying decision-making behaviors in which direct manipulation experiments are still possible. Some prominent studies have brought to light major discrepancies between monkey and human cognition, highlighting problems with unverified extrapolation from monkey to human. Here, we use a parallel model system-artificial neural networks (ANNs)-to investigate a well-established discrepancy identified between monkeys and humans with a working memory task, in which monkeys appear to use a recency-based strategy while humans use a target-selective strategy. We find that ANNs trained on the same task exhibit a progression of behavior from random behavior (untrained) to recency-like behavior (partially trained) and finally to selective behavior (further trained), suggesting monkeys and humans may occupy different points in the same overall learning progression. Surprisingly, what appears to be recency-like behavior in the ANN, is in fact an emergent non-recency-based property of the organization of the neural network's state space during its development through training. We find that explicit encouragement of recency behavior during training has a dual effect, not only causing an accentuated recency-like behavior, but also speeding up the learning process altogether, resulting in an efficient shaping mechanism to achieve the optimal strategy. Our results suggest a new explanation for the discrepency observed between monkeys and humans and reveal that what can appear to be a recency-based strategy in some cases may not be recency at all.

Seeing double with a multifunctional reservoir computer.

Multifunctional biological neural networks exploit multistability in order to perform multiple tasks without changing any network properties. Enabling artificial neural networks (ANNs) to obtain certain multistabilities in order to perform several tasks, where each task is related to a particular attractor in the network's state space, naturally has many benefits from a machine learning perspective. Given the association to multistability, in this paper, we explore how the relationship between different attractors influences the ability of a reservoir computer (RC), which is a dynamical system in the form of an ANN, to achieve multifunctionality. We construct the "seeing double" problem in order to systematically study how a RC reconstructs a coexistence of attractors when there is an overlap between them. As the amount of overlap increases, we discover that for multifunctionality to occur, there is a critical dependence on a suitable choice of the spectral radius for the RC's internal network connections. A bifurcation analysis reveals how multifunctionality emerges and is destroyed as the RC enters a chaotic regime that can lead to chaotic itinerancy.

Deep Reinforcement Learning for Shared Offloading Strategy in Vehicle Edge Computing

Vehicular edge computing (VEC) effectively reduces the computing load of vehicles by offloading computing tasks from vehicle terminals to edge servers. However, offloading of tasks increase in quantity the transmission time and energy of the network. In order to reduce the computing load of edge servers and improve the system response, a shared offloading strategy based on deep reinforcement learning is proposed for the complex computing environment of Internet of Vehicles (IoVs). The shared offloading strategy exploits the commonality of vehicles task requests, similar computing tasks coming from different vehicles can share the computing results of former task submitted. The shared offloading strategy can be adapted to the complex scenarios of the IoVs. Each vehicle can share the offloading conditions of the VEC servers, and then adaptively select three computing modes: local execution, task offloading, and shared offloading. In this article, the network state and offloading strategy space are the input of the deep reinforcement learning (DRL). Through the DRL, each task unit selects the offloading strategy with the optimal energy consumption at each time period in the dynamic IoVs transmission and computing environment. Compared with the existing proposals and DRL-based algorithms, it can effectively reduce the delay and energy consumption required for tasks offloading.

Persistent homology of coarse-grained state-space networks.

This work is dedicated to the topological analysis of complex transitional networks for dynamic state detection. Transitional networks are formed from time series data and they leverage graph theory tools to reveal information about the underlying dynamic system. However, traditional tools can fail to summarize the complex topology present in such graphs. In this work, we leverage persistent homology from topological data analysis to study the structure of these networks. We contrast dynamic state detection from time series using a coarse-grained state-space network (CGSSN) and topological data analysis (TDA) to two state of the art approaches: ordinal partition networks (OPNs) combined with TDA and the standard application of persistent homology to the time-delay embedding of the signal. We show that the CGSSN captures rich information about the dynamic state of the underlying dynamical system as evidenced by a significant improvement in dynamic state detection and noise robustness in comparison to OPNs. We also show that because the computational time of CGSSN is not linearly dependent on the signal's length, it is more computationally efficient than applying TDA to the time-delay embedding of the time series.

Machine learning with nonlinear state space models

Abstract In this dissertation, a novel class of model structures and associated training algorithms for building data-driven nonlinear state space models is developed. The new identification procedure with the resulting model is called local model state space network (LMSSN). Furthermore, recurrent neural networks (RNNs) and their similarities to nonlinear state space models are elaborated on. The overall outstanding performance of the LMSSN is demonstrated on various applications.

A systematic exploration of local network state space in neocortical mouse brain slices

The ex vivo cortical slice is an extremely versatile preparation, but its utility ultimately depends on understanding its limitations and functional constraints. A question for experimentalists new to the field of cortical slice electrophysiology might be — what are the different network dynamical states available to a cortical slice as a function of excitatory drive? The purpose of this study is to provide a coherent answer to this question, within the context of extracellularly recorded population field potentials.Cortical slices (400 µm) were prepared from adult male or female C57 mice. Evoked responses were recorded within cortical layer III/IV using extracellularly positioned metal electrodes. In the first part of the study, slice excitatory drive was increased by reducing the concentration of magnesium ions in the artificial cerebrospinal fluid — and the evoked responses categorized during the transition. In the second part, each of the identified functional states were explored in greater detail with tissue perfusion conditions and excitatory drive optimised for the requisite response state.As expected, rodent cortical slices did not generate spontaneous, persistent EEG-like field potential activity. However, distinct response states (spontaneous and evoked) characterized by intermittent population bursts could be differentiated as a function of excitatory drive. Each state reflected different modes of neocortical activation: “monosynaptic” responses were brief, non-propagating activations, reflecting an inhibited cortex with sensory processing blocked; polysynaptic and epileptiform activity propagated intra-cortically, the latter reflecting a hyperactivated, hypersynchronous “seizing” cortex. Polysynaptic activity most closely resembled sensory “up states” associated with intracortical sensory processing.Understanding the functional distinction between the different cortical slice response states is the starting point for designing experiments that maximise the utility of this ex vivo model. The results and descriptions in this study should help slice experimentalists less experienced in the nuances of cortical slice neurophysiology to make informed choices about how to tailor the parameters of the model to suit the specific aims of their research.

Optimization Approaches for Nonlinear State Space Models

The Local Model State Space Network (LMSSN) is a recently developed black box algorithm in nonlinear system identification. It has proven to be an appropriate tool on benchmark problems as well as for real-world processes. A severe shortcoming though is the long computation time that is necessary for model training. Therefore, a different optimization strategy, the adaptive moment estimation (ADAM) method with mini batches is used for the LMSSN and compared to the current Quasi-Newton (QN) optimization method. It is shown on a numerical Hammerstein example and on a well known Wiener-Hammerstein benchmark that the use of ADAM and mini batches does not limit the performance of the LMSSN algorithm and speeds up the nonlinear optimization per investigated split by more than 30 times. The price to be paid, however, is higher parameter variance (less interpretability) and more tedious hyperparameter tuning.

Slow manifolds within network dynamics encode working memory efficiently and robustly

Working memory is a cognitive function involving the storage and manipulation of latent information over brief intervals of time, thus making it crucial for context-dependent computation. Here, we use a top-down modeling approach to examine network-level mechanisms of working memory, an enigmatic issue and central topic of study in neuroscience. We optimize thousands of recurrent rate-based neural networks on a working memory task and then perform dynamical systems analysis on the ensuing optimized networks, wherein we find that four distinct dynamical mechanisms can emerge. In particular, we show the prevalence of a mechanism in which memories are encoded along slow stable manifolds in the network state space, leading to a phasic neuronal activation profile during memory periods. In contrast to mechanisms in which memories are directly encoded at stable attractors, these networks naturally forget stimuli over time. Despite this seeming functional disadvantage, they are more efficient in terms of how they leverage their attractor landscape and paradoxically, are considerably more robust to noise. Our results provide new hypotheses regarding how working memory function may be encoded within the dynamics of neural circuits.

Applications of optimal nonlinear control to a whole-brain network of FitzHugh-Nagumo oscillators.

We apply the framework of optimal nonlinear control to steer the dynamics of a whole-brain network of FitzHugh-Nagumo oscillators. Its nodes correspond to the cortical areas of an atlas-based segmentation of the human cerebral cortex, and the internode coupling strengths are derived from diffusion tensor imaging data of the connectome of the human brain. Nodes are coupled using an additive scheme without delays and are driven by background inputs with fixed mean and additive Gaussian noise. Optimal control inputs to nodes are determined by minimizing a cost functional that penalizes the deviations from a desired network dynamic, the control energy, and spatially nonsparse control inputs. Using the strength of the background input and the overall coupling strength as order parameters, the network's state-space decomposes into regions of low- and high-activity fixed points separated by a high-amplitude limit cycle, all of which qualitatively correspond to the states of an isolated network node. Along the borders, however, additional limit cycles, asynchronous states, and multistability can be observed. Optimal control is applied to several state-switching and network synchronization tasks, and the results are compared to controllability measures from linear control theory for the same connectome. We find that intuitions from the latter about the roles of nodes in steering the network dynamics, which are solely based on connectome features, do not generally carry over to nonlinear systems, as had been previously implied. Instead, the role of nodes under optimal nonlinear control critically depends on the specified task and the system's location in state space. Our results shed new light on the controllability of brain network states and may serve as an inspiration for the design of new paradigms for noninvasive brain stimulation.

DDQP: A Double Deep Q-Learning Approach to Online Fault-Tolerant SFC Placement

Since Network Function Virtualization (NFV) decouples network functions (NFs) from the underlying dedicated hardware and realizes them in the form of software called Virtual Network Functions (VNFs), they are enabled to run in any resource-sufficient virtual machines. A service function chain (SFC) is composed of a sequential set of VNFs. As VNFs are vulnerable to various faults such as software failures, we consider how to deploy both active and standby SFC instances. Given the complexity and unpredictability of the network state, we propose a double deep Q-networks based online SFC placement scheme DDQP. Specifically, DDQP uses deep neural networks to deal with large continuous network state space. In the case of stateful VNFs, we offer constant generated state updates from active instances to standby instances to guarantee seamless redirection after failures. With the goal of balancing the waste of resources and ensuring service reliability, we introduce five progressive schemes of resource reservations to meet different customer needs. Our experimental results demonstrate that DDQP responds rapidly to arriving requests and reaches near-optimal performance. Specifically, DDQP outweighs the state-of-the-art method by 16.30% and 38.51% higher acceptance ratio under different schemes with 82x speedup on average. In order to enhance the integrity of the SFC state transition, we further proposed DDQP+, which extends DDQP by adding the delayed placement mechanism. Compared with DDQP, the design of the DDQP+ algorithm is more reasonable and comprehensive. The experiment results also show that DDQP+ achieved further improvement in multiple performance indicators.

An Investigation of Gene Regulatory Network State Space Variability

Genes are segments of DNA that provide a blueprint for cells and organisms to effectively control processes and regulations within individuals. There have been many attempts to quantify these processes, as a greater understanding of how genes operate could have large impacts on both personalized and precision medicine. Current biological methods cannot easily reveal the details of gene interactions. Therefore, we use gene expression data to infer networks of interactions, which are called gene regulatory networks or GRNs. These methods are designed to bypass the need for large amounts of data and extensive knowledge about a network. In this work, we extend previous work by investigating additional ways to incorporate stochasticity into gene regulatory networks.

A Systematic Exploration of Local Network State Space in Neocortical Mouse Brain Slices

A Systematic Exploration of Local Network State Space in Neocortical Mouse Brain Slices

Extrapolation Behavior Comparison of Nonlinear State Space Models

Abstract The extrapolation behavior and its consequences of two different state space models is studied. More specifically, the extrapolation behavior of the local model state space network (LMSSN) is compared to the polynomial nonlinear state space model (PNLSS). Both approaches represent favorable alternatives in the light of their results on recent benchmark studies. The models are compared with regards to their static extrapolation behavior and their tendency to erratic behavior if test amplitudes grow larger than training amplitudes. It is shown on a Hammerstein and Wiener artificial test process that the PNLSS shows severe shortcomings concerning stability when operated close to or outside the boundaries in which it was trained. The LMSSN instead shows more favorable behavior in extrapolation and does only exhibit minor erratic behavior. If the process is inherently polynomial in some way, the PNLSS outperforms other models in extrapolation, as shown on the Silverbox benchmark test.

0439 Nonlinear Dynamics Forecasting for Personalize Prognosis of Obstructive Sleep Apnea Onsets

Abstract Introduction The emphasis on disease prevention, early detection, and preventive treatments will revolutionize the way sleep clinicians evaluate their patients. Obstructive Sleep Apnea (OSA) is one of the most prevalent sleep disorders with approximately 100 millions patients been diagnosed worldwide. The effectiveness of sleep disorder therapies can be enhanced by providing personalized and real-time prediction of OSA episode onsets. Previous attempts at OSA prediction are limited to capturing the nonlinear, nonstationary dynamics of the underlying physiological processes. Methods This paper reports an investigation into heart rate dynamics aiming to predict in real time the onsets of OSA episode before the clinical symptoms appear. The method includes (a) a representation of a transition state space network to characterize dynamic transition of apneic states (b) a Dirichlet-Process Mixture-Gaussian-Process prognostic method for estimating the distribution of the time estimate the remaining time until the onset of an impending OSA episode by considering the stochastic evolution of the normal states to an anomalous (apnea) Results The approach was tested using three datasets including (1) 20 records from 14 OSA subjects in benchmark ECG apnea databases (Physionet.org), (2) records of eight subjects from previous work. The average prediction accuracy (R2) is reported as 0.75%, with 87% of observations within the 95% confidence interval. Estimated risk indicators at 1 to 3 min till apnea onset are reported as 85.8 %, 80.2 %, and 75.5 %, respectively. Conclusion The present prognosis approach can be integrated with wearable devices to facilitate individualized treatments and timely prevention therapies. Support N/A