Input Nodes Research Articles

Binary Graph Convolutional Network With Capacity Exploration.

The current success of Graph Neural Networks (GNNs) usually relies on loading the entire attributed graph for processing, which may not be satisfied with limited memory resources, especially when the attributed graph is large. This paper pioneers to propose a Binary Graph Convolutional Network (Bi-GCN), which binarizes both the network parameters and input node attributes and exploits binary operations instead of floating-point matrix multiplications for network compression and acceleration. Meanwhile, we also propose a new gradient approximation based back-propagation method to properly train our Bi-GCN. According to the theoretical analysis, our Bi-GCN can reduce the memory consumption by an average of ∼31x for both the network parameters and input data, and accelerate the inference speed by an average of ∼51x, on three citation networks, i.e., Cora, PubMed, and CiteSeer. Besides, we introduce a general approach to generalize our binarization method to other variants of GNNs, and achieve similar efficiencies. Although the proposed Bi-GCN and Bi-GNNs are simple yet efficient, these compressed networks may also possess a potential capacity problem, i.e., they may not have enough storage capacity to learn adequate representations for specific tasks. To tackle this capacity problem, an Entropy Cover Hypothesis is proposed to predict the lower bound of the width of Bi-GNN hidden layers. Extensive experiments have demonstrated that our Bi-GCN and Bi-GNNs can give comparable performances to the corresponding full-precision baselines on seven node classification datasets and verified the effectiveness of our Entropy Cover Hypothesis for solving the capacity problem.

Improving Fault Diagnosis of the Drilling Permanent Magnet Synchronous Motor (DPMSM) in Harsh Environments: A Novel Approach Using Object-Oriented Bayesian Network (OOBN)

The drilling permanent magnet synchronous motor (DPMSM) contains multiple subsystems with identical structures and has a high probability of failure because the downhole working conditions are harsh. Therefore, the quick localization of faults is difficult to determine although the fault type may be identified in time. The system diagnostic model based on the Bayesian network (BN) can be used for fault diagnosis and localization for components in subsystems, but it is difficult to build and modify due to the complex system in practice. New methods are necessary to reduce the difficulty of building and modifying models. In this study, object-oriented ideas are introduced into the BN to establish a system diagnostic model based on an Object-oriented Bayesian network (OOBN) for the DPMSM. First, the fault diagnostic models for subsystems based on BN are established, respectively. Then, submodels of forward and backward based on BN are instantiated as instance nodes. Next, instance nodes are connected through input nodes and output nodes to establish the OOBN-based system diagnosis model. Finally, the system diagnosis model is validated by sensitivity analysis and the effectiveness is discussed in Cases. The system diagnosis model can effectively reduce the difficulties of modeling and modifying.

Frequency Propagation: Multimechanism Learning in Nonlinear Physical Networks.

We introduce frequency propagation, a learning algorithm for nonlinear physical networks. In a resistive electrical circuit with variable resistors, an activation current is applied at a set of input nodes at one frequency and an error current is applied at a set of output nodes at another frequency. The voltage response of the circuit to these boundary currents is the superposition of an activation signal and an error signal whose coefficients can be read in different frequencies of the frequency domain. Each conductance is updated proportionally to the product of the two coefficients. The learning rule is local and proved to perform gradient descent on a loss function. We argue that frequency propagation is an instance of a multimechanism learning strategy for physical networks, be it resistive, elastic, or flow networks. Multimechanism learning strategies incorporate at least two physical quantities, potentially governed by independent physical mechanisms, to act as activation and error signals in the training process. Locally available information about these two signals is then used to update the trainable parameters to perform gradient descent. We demonstrate how earlier work implementing learning via chemical signaling in flow networks (Anisetti, Scellier, etal., 2023) also falls under the rubric of multimechanism learning.

Application of Deep Belief Network in Weather Modeling: PM2.5 Concentration in Thailand

In Thailand, the number of particles matter with diameter of less than 2.5 microns or PM2.5 concentration exceed the standard in many areas, especially in Chiang Mai. This affects the image of the country in terms of economy, health, and environment. The objective of this research is to study the structure of model for PM2.5 concentration by using a Deep Belief Network (DBN) with the daily data set of PM2.5 concentration from the air quality monitoring station at Yupparaj Wittayalai School, Chiang Mai. The data was analyzed through an unsupervised path using the Minimizing Contrastive Divergence (MCD) algorithm, followed by a supervised path using Back-Propagation Neural Network (BPNN) algorithm to estimate the parameters of DBN. The result shows that the optimal DBN structure has 5 input nodes and 20 hidden neurons in the first hidden layer. This model has an 88.4 percent accuracy in forecasting PM2.5 concentration. In addition, this model can be applied for other weather forecasting such as rainfall or water level in a basin.

Disinfection scheduling in water distribution networks considering input time-delay uncertainty

Abstract A significant challenge when attempting to regulate the spatial-temporal concentration of a disinfectant in a water distribution network is the large and uncertain delay between the time that the chemical is injected at the input node and the time that the concentration is measured at the monitoring output nodes. Uncertain time delays are due to varying water flows, which depend mainly on consumer water demands. Existing approaches cannot guarantee that the concentration of the disinfectant will remain within a specified range at the output, even though bounds on time-delay uncertainty may be known. In this work, given bounded water-flow uncertainty, we use the input–output modeling approach to develop a disinfectant scheduling methodology that guarantees a bounded output disinfectant concentration. The proposed methodology creates an input–output model uncertainty characterization by utilizing estimated bounds on water-quality states using the backtracking approach. An optimization problem is formulated and solved to find an input schedule that keeps the disinfectant concentration within predefined bounds for a specified time horizon. Simulation results in two case studies where water demands varied between ±20% of their nominal value show that the proposed scheduler is able to avoid lower bound violations of disinfectant concentration.

Controlling target brain regions by optimal selection of input nodes.

The network control theory framework holds great potential to inform neurostimulation experiments aimed at inducing desired activity states in the brain. However, the current applicability of the framework is limited by inappropriate modeling of brain dynamics, and an overly ambitious focus on whole-brain activity control. In this work, we leverage recent progress in linear modeling of brain dynamics (effective connectivity) and we exploit the concept of target controllability to focus on the control of a single region or a small subnetwork of nodes. We discuss when control may be possible with a reasonably low energy cost and few stimulation loci, and give general predictions on where to stimulate depending on the subset of regions one wishes to control. Importantly, using the robustly asymmetric effective connectome instead of the symmetric structural connectome (as in previous research), we highlight the fundamentally different roles in- and out-hubs have in the control problem, and the relevance of inhibitory connections. The large degree of inter-individual variation in the effective connectome implies that the control problem is best formulated at the individual level, but we discuss to what extent group results may still prove useful.

Polyregular Functions on Unordered Trees of Bounded Height

We consider injective first-order interpretations that input and output trees of bounded height. The corresponding functions have polynomial output size, since a first-order interpretation can use a k -tuple of input nodes to represent a single output node. We prove that the equivalence problem for such functions is decidable, i.e. ‍given two such interpretations, one can decide whether, for every input tree, the two output trees are isomorphic. We also give a calculus of typed functions and combinators which derives exactly injective first-order interpretations for unordered trees of bounded height. The calculus is based on a type system, where the type constructors are products, coproducts and a monad of multisets. Thanks to our results about tree-to-tree interpretations, the equivalence problem is decidable for this calculus. As an application, we show that the equivalence problem is decidable for first-order interpretations between classes of graphs that have bounded tree-depth. In all cases studied in this paper, first-order logic and MSO have the same expressive power, and hence all results apply also to MSO interpretations.

Minimal Controllability Problem on Linear Structural Descriptor Systems With Forbidden Nodes

We consider a minimal controllability problem (MCP), which determines the minimum number of input nodes for a descriptor system to be structurally controllable. We investigate the “forbidden nodes” in descriptor systems, denoting nodes that are unable to establish connections with input components. The three main results of this work are as follows. First, we show a solvability condition for the MCP with forbidden nodes using graph theory such as a bipartite graph and its Dulmage–Mendelsohn decomposition. Next, we derive the optimal value of the MCP with forbidden nodes. The optimal value is determined by an optimal solution for constrained maximum matching, and this result includes that of the standard MCP in the previous work. Finally, we provide an efficient algorithm for solving the MCP with forbidden nodes based on an alternating path algorithm.

Biphasic amplitude oscillator characterized by distinct dynamics of trough and crest.

Biphasic amplitude dynamics (BAD) of oscillation have been observed in many biological systems. However, the specific topology structure and regulatory mechanisms underlying these biphasic amplitude dynamics remain elusive. Here, we searched all possible two-node circuit topologies and identified the core oscillator that enables robust oscillation. This core oscillator consists of a negative feedback loop between two nodes and a self-positive feedback loop of the input node, which result in the fast and slow dynamics of the two nodes, thereby achieving relaxation oscillation. Landscape theory was employed to study the stochastic dynamics and global stability of the system, allowing us to quantitatively describe the diverse positions and sizes of the Mexican hat. With increasing input strength, the size of the Mexican hat exhibits a gradual increase followed by a subsequent decrease. The self-activation of input node and the negative feedback on input node, which dominate the fast dynamics of the input node, were observed to regulate BAD in a bell-shaped manner. Both deterministic and statistical analysis results reveal that BAD is characterized by the linear and nonlinear dependence of the oscillation trough and crest on the input strength. In addition, combining with computational and theoretical analysis, we addressed that the linear response of trough to input is predominantly governed by the negative feedback, while the nonlinear response of crest is jointly regulated by the negative feedback loop and the self-positive feedback loop within the oscillator. Overall, this study provides a natural and physical basis for comprehending the occurrence of BAD in oscillatory systems, yielding guidance for the design of BAD in synthetic biology applications.

Breaking the curse of dimensional collapse in graph contrastive learning: A whitening perspective

Dimensional collapse in graph contrastive learning (GCL) confines node embeddings to their lower-dimensional subspace, diminishing their distinguishability. However, the causes and solutions of this curse remain relatively underexplored. In statistics, whitening presents a powerful tool to eliminate correlations among multiple variables. This motivates us to relieve the dimensional collapse of GCL from a whitening perspective. In this paper, we propose an intuitive analysis suggesting that high similarity scores of node embeddings may cause dimensional collapse, providing more evidence for its presence. Considering the success of whitening in statistics, we introduce a new plug-and-play module called the ▪hitening ▪raph ▪ontrastive ▪earning (WGCL) to address the dimensional collapse issue in existing GCL methods. WGCL plugin standardises the covariance matrices of dimensions, eliminating correlations among node embeddings' dimensions. Additionally, we enhance the conventional GCL training objective by introducing a mutual information maximisation loss between input features and node embeddings to maintain information capacity. Our experiments demonstrate that WGCL effectively addresses dimensional collapse, leading to an average improvement of 0.93% (up to 2.0%) in classification accuracy across three GCL backbones on nine widely-used datasets. The code to reproduce the experiments is available at https://github.com/acboyty/WGCL.

The Orienteering Problem with Drones

We extend the classical problem setting of the orienteering problem (OP) to incorporate multiple drones that cooperate with a truck to visit a subset of the input nodes. We call this problem the OP with multiple drones (OP-mD). Drones have a limited battery endurance, and thus, they can either move together with the truck at no energy cost for the battery or be launched by the truck onto short flights that must start and end at different customer locations. A drone serves exactly one customer per flight. Moreover, the truck and the drones must wait for each other at the landing locations. A customer prize can be collected at most once, either upon visiting it by the truck or upon serving it by a drone. Similarly to the OP, we maximize the total collected prize under the condition that the truck and the drones return to the depot within a given amount of time. We provide a mixed-integer linear programming formulation for the OP-mD and devise a tailored branch-and-cut algorithm based on a novel decomposition of the problem. We solve instances of the OP-mD with up to 50 nodes within one hour of CPU time with a standard computational setup. Finally, we adapt our framework to solve closely related problems in the literature and compare the resulting computational performance with that of previous studies.

Micromagnetic and quantitative prediction of yield strength and tensile strength in DP590 steels based on ReliefF + Clustering feature selection method

The correlation between multiple patterns of micromagnetic signatures and the mechanical properties (yield strength (Rp) and tensile strength (Rm) of high-strength steel (referred to as DP590 steel in Chinese standards) was investigated in this study. Feedforward neural network (FF-NN) models with carefully selected magnetic features as input nodes were established for quantitative prediction of yield strength and tensile strength. The accuracy FF-NN models highly relied on the quality of calibration specimens and the way of selecting magnetic features. The variations of the measured target properties were used to evaluate the quality of the calibration specimens. The specimens with similar yield strength (or tensile strength) were merged to share the same target properties in the model training process. The results demonstrated that merging proper target properties (label) could improve the performance of the models in quantitative prediction of yield strength and tensile strength in DP590 steels. In addition, the performances of FF-NN models combined with the algorithms of ReliefF and ReliefF + clustering were evaluated. The comparison results proved that the FF-NN models employing input nodes selection strategy of ReliefF + clustering realized the advantages of smaller dimensions of input nodes, less training time consumption at the cost of slight accuracy reduction.

GlioPredictor: A Deep Learning Model for Identification of High-Risk Low-Grade Glioma toward Adjuvant Treatment Planning

High-risk low-grade glioma (LGG) patients are recommended to undergo adjuvant radiotherapy whereas watchful waiting is recommended for low-risk LGG patients per the latest NCCN guidelines. Based on Radiation Therapy Oncology Group (RTOG) 9802, high-risk features include age >40 or subtotal resection (STR). However, in the era of molecular-based classification for tumors of central neural system, current risk classification criteria based on gross disease and patient demographics may be outdated. Here, we aim to develop a molecular-based glioma risk classification system (GlioPredictor) that could potentially facilitate identification of high-risk LGG patients. A total of 507 LGG cases from The Cancer Genome Atlas-low grade glioma (TCGA-LGG), and 1,309 cases from AACR GENIE v13.0 datasets were studied for genetic disparities between IDH1-wildtype and mutated cohorts, and varying age groups. Through a feature selection technique using genomic profiling and correlation analyses, features such as mutation status, copy number variations (CNVs), among other clinicopathologic features prognostic of IDH1 mutation status were selected as potential inputs to train an artificial neural networks (ANNs) that could predict IDH1 mutation status. Model performance was assessed using the area under the receiver operating characteristic curve (AUC). Memorial Sloan Kettering (MSK) dataset (n = 404) for LGG was used to cross-validate the trained ANN. The optimized ANN model has 6 layers with 6 input nodes, 20 hidden nodes, and a binary output layer. The weights and biases of the hidden layers of the best-performing model were retrieved and reconstructed to yield the GlioPredictor score-the predicted risk of progression for IDH1-wildtype LGG. Over 81% of glioma patients age less than 40 have IDH1 mutation, as compared with 31% in those age above 60. Using age > 40 as a cutoff failed to identify high-risk IDH1-mutant LGG with early progression. IDH1 mutation is associated with decreased CNVs of EGFR (21 % vs. 3%), CDKN2A (20% vs. 6%) and PTEN (14% vs. 1.7%), and increased percentage of mutations for TP53 (15% vs. 63%), and ATRX (10% vs. 54%) (p<0.001). Using these molecular features, along with the patient's age, an ANN model with 6 layers and 20 hidden nodes can predict IDH1 mutation status with over 90% accuracy and AUC score over 0.91. We have developed an ANN model that is capable of learning the prognostic features of LGG associated with an IDH1-mutated LGG cohort and using the features to predict high-risk patients from the IDH1-wildtype cohort. This ANN model facilitates the selection of LGG patients who could benefit from immediate adjuvant radiotherapy. Future work includes the integration of image features to improve the prediction performance of the GlioPredictor system.

A spatiotemporal inference model for hazard chains based on weighted dynamic Bayesian networks for ground subsidence in mining areas

Ground subsidence concerns the long-term development of mining areas, and if not addressed effectively, it could gradually evolve into a major issue limiting the future economic development and survival of mining firms and local populations. However, there is unpredictability and uncertainty in the analysis of ground subsidence in mining areas, which is a quantitative and qualitative problem coupled with multiple indicators. By creating a chain relationship between ground subsidence in mining areas, this research provides a spatiotemporal inference model that integrates remote sensing (RS), geographic information system (GIS), and probabilistic map theory. The model uses a dynamic Bayesian framework to integrate the ground subsidence hazard chain in mining areas, standardizes multi-source data using GIS, computes node probabilities, and applies the entropy weight approach to improve model parameters. The Pingshuo mining area in China served as the study area for the model, and the mean values of area under the curve (AUC) and Brier score (BS) of the inferred results were 0.85 and 0.18, respectively, demonstrating that the model had some accuracy and dependability. Further analysis was performed on the impact of weights on the outcomes and the sensitivity of the model to the input nodes. The findings indicated that the spatiotemporal distribution of the results inferred from the model essentially matched the actual circumstance and could offer data assistance for mine safety management. The matching of the subsidence areas was effectively improved by optimizing the model with weights. The accuracy would also grow as the number of input nodes increased. The model proposed in this study is not limited by data, and the structure can be adjusted with the change of disaster chains, which is applicable to the study of multiple uncertainty problems.

Analysis of the Use of Artificial Neural Network Models in Predicting Bitcoin Prices

Bitcoin is one of the fastest-growing digital currencies or cryptocurrencies in the world. However, the highly volatile Bitcoin price poses a very extreme risk for traders investing in cryptocurrencies, especially Bitcoin. To anticipate these risks, a prediction system is needed to predict the fluctuations in cryptocurrency prices. Artificial Neural Network (ANN) is a relatively new model discovered and can solve many complex problems because the way it works mimics human nerve cells. ANN has the advantage of being able to describe both linear and non-linear models with a fairly wide range. This research aims to determine the best performance and level of accuracy of the ANN model using the Back-Propagation Neural Network (BPNN) algorithm in predicting Bitcoin prices. This study uses Bitcoin price data for the period 2020 to 2023 taken from the CoinDesk market. The results of this study indicate that the ANN model produces the best performance in the form of four input nodes, 12 hidden nodes, and one output node (4-12-1) with an accuracy rate of around 3.0617175%.

Path Diversity and Survivability for the HyperX Datacenter Topology

Network survivability has been recognized as an issue of a major importance in terms of security, stability and prosperity. This paper studies fundamental properties of HyperX, an emerging topology for connecting supercomputing and datacenter networks. We focus on the establishment of paths with guaranteed survivability, allowing path availability even upon a restricted number of link failures. We first examine the availability of disjoint paths connecting a pair of input nodes. Disjoint paths guarantee path existence even upon a bounded number of link failures. We explore the inherent tradeoff between allowing slightly longer paths and the ability to extend available sets of mutually disjoint paths. Second, we study the availability of paths in a HyperX topology that already observed link failures. Such failures can increase the length of available paths or even eliminate connectivity between network parts. We also compare basic properties of shortest paths in HyperX to other datacenter topologies. Last, we provide an evaluation to illustrate path availability along with the potential impact of failures.

Graph neural network method for the intelligent selection of river system

The spatial features and generalisation rules for river network generalisation are difficult to directly quantify using indicators. To consider dimensional information hidden in river networks and improve river network selection accuracy, this study introduces a graph convolutional neural network-based method. First, we modelled the river network as a graph structure, where the nodes represent each river segment and edges represent the connections between river segments. The semantic, geometric, and morphological features of individual river segments and topological and constraint features between river segments were then calculated to characterise the relevant nodes. Second, under supervised classification, the input node attributes and labels were sampled and aggregated to obtain richer and more abstract high-level features. The graph convolutional neural network model then selected or deleted river segments. Finally, the selected individual river segments were connected to obtain the complete integrated river network. A 1:10,000 scale map of the Min River system in the Yangtze River Basin was tested, with a 1:50,000 scale map used as the control, and it yielded a correct classification rate >95%. Moreover, the correct classification rate was 7.35%–5.31% and 7.7%–3.3% higher than that of other graph neural network methods and traditional machine learning methods, respectively.

Detection of apple moldy cores using transmittance spectroscopy combined with supervised classifier comparison and parameter optimization

It is difficult to determine which apples have moldy cores just by looking at the outside of the apple. In the present study, we investigated identifying moldy cores using near-infrared transmittance spectra. First, input spectral features selected by noise adjusted principal component analysis (NAPCA) for back propagation artificial neural network (BP ANN) was used to reduce the dimensions of the original data. Then, four factors and five levels uniform design of the input nodes, training functions, transfer layer functions and output layer functions for NAPCA-BP ANN optimization is proposed. And the original data were input into NAPCA-BP ANN to obtain the recognition accuracy and NAPCA-support vector machine (SVM) was as a comparative recognition model. The results showed that through the uniform design-based NAPCA-BP ANN optimization, the NAPCA method had higher identification accuracy, precision, recall and F1 score, than either full spectrum or principal component analysis. Being assessed by different ratio of model test, functions in the hidden layer and output layer of NAPCA-BP ANN, the proposed method achieved the best accuracy to 98.03%. The accuracy, precision, recall and F1 score based on NAPCA-BP ANN were 3.92%, 2.86%, 2.78% and 2.82% higher than those based on NAPCA-SVM, respectively. This method provides a theoretical basis for the development of on-line monitoring of the internal quality of apples.

Sales rate forecasting of single-detached houses using artificial neural network technique

Although the predicted number of sold units per month is one of the most necessary information to study the feasibility of detached-housing estate projects, traditional forecasting methods are limited. The research objective was to apply the Artificial Neural Network (ANN) technique to develop a sales rate forecasting model by compiling factors from an appropriate literature review. Then, 100 housing project data were collected from market research reports and the projects' websites and analyzed using the ANN technique. The results showed the ANN network with 16 input nodes from 10 factors: Selling price, Number of bathrooms, Number of bedrooms, Distance from the main road, Distance from the bus stop, Distance from the expressway, Distance from the metro station, Distance from the gas station, Distance from the shopping mall, and Project location zone. The acquired model had a Root Mean Square Error (RMSE) of ±6.296, and the slope and R2 values of the linear regression analysis between the forecasted rate and the actual rate were 0.620 and 0.571, respectively. The findings guide real estate developers and academia on the factors affecting the sales rate and provide a decision support model for investment and design of projects and confirm the potential of ANN in solving the problem with limited numbers of data.

Benchmarking graph representation learning algorithms for detecting modules in molecular networks

Background: A common task in molecular network analysis is the detection of community structures or modules. Such modules are frequently associated with shared biological functions and are often disrupted in disease. Detection of community structure entails clustering nodes in the graph, and many algorithms apply a clustering algorithm on an input node embedding. Graph representation learning offers a powerful framework to learn node embeddings to perform various downstream tasks such as clustering. Deep embedding methods based on graph neural networks can have substantially better performance on machine learning tasks on graphs, including module detection; however, existing studies have focused on social and citation networks. It is currently unclear if deep embedding methods offer any advantage over shallow embedding methods for detecting modules in molecular networks. Methods: Here, we investigated deep and shallow graph representation learning algorithms on synthetic and real cell-type specific gene interaction networks to detect gene modules and identify pathways affected by sequence nucleotide polymorphisms. We used multiple criteria to assess the quality of the clusters based on connectivity as well as overrepresentation of biological processes. Results: On synthetic networks, deep embedding based on a variational graph autoencoder had superior performance as measured by modularity metrics, followed closely by shallow methods, node2vec and Graph Laplacian embedding. However, the performance of the deep methods worsens when the overall connectivity between clusters increases. On real molecular networks, deep embedding methods did not have a clear advantage and the performance depended upon the properties of the graph and the metrics. Conclusions: Deep graph representation learning algorithms for module detection-based tasks can be beneficial for some biological networks, but the performance depends upon the metrics and graph properties. Across different network types, Graph Laplacian embedding followed by node2vec are the best performing algorithms.