Hierarchical Network Architecture Research Articles

Heterogeneous Hierarchical Fusion Network for Multimodal Sentiment Analysis in Real-World Environments

Multimodal sentiment analysis models can determine users’ sentiments by utilizing rich information from various sources (e.g., textual, visual, and audio). However, there are two key challenges when deploying the model in real-world environments: (1) the limitations of relying on the performance of automatic speech recognition (ASR) models can lead to errors in recognizing sentiment words, which may mislead the sentiment analysis of the textual modality, and (2) variations in information density across modalities complicate the development of a high-quality fusion framework. To address these challenges, this paper proposes a novel Multimodal Sentiment Word Optimization Module and a heterogeneous hierarchical fusion (MSWOHHF) framework. Specifically, the proposed Multimodal Sentiment Word Optimization Module optimizes the sentiment words extracted from the textual modality by the ASR model, thereby reducing sentiment word recognition errors. In the multimodal fusion phase, a heterogeneous hierarchical fusion network architecture is introduced, which first utilizes a Transformer Aggregation Module to fuse the visual and audio modalities, enhancing the high-level semantic features of each modality. A Cross-Attention Fusion Module then integrates the textual modality with the audiovisual fusion. Next, a Feature-Based Attention Fusion Module is proposed that enables fusion by dynamically tuning the weights of both the combined and unimodal representations. It then predicts sentiment polarity using a nonlinear neural network. Finally, the experimental results on the MOSI-SpeechBrain, MOSI-IBM, and MOSI-iFlytek datasets show that the MSWOHHF outperforms several baselines, demonstrating better performance.

Energy-Efficient Clustering in Wireless Sensor Networks Using Grey Wolf Optimization and Enhanced CSMA/CA.

Survivability is a critical concern in WSNs, heavily influenced by energy efficiency. Addressing severe energy constraints in WSNs requires solutions that meet application goals while prolonging network life. This paper presents an Energy Optimization Approach (EOAMRCL) for WSNs, integrating the Grey Wolf Optimization (GWO) for enhanced performance. EOAMRCL aims to enhance energy efficiency by selecting the optimal duty-cycle schedule, transmission power, and routing paths. The proposed approach employs a centralized strategy using a hierarchical network architecture. During the cluster formation phase, an objective function, augmented with GWO, determines the ideal cluster heads (CHs). The routing protocol then selects routes with minimal energy consumption for data transmission to CHs, using transmission power as a metric. In the transmission phase, the MAC layer forms a duty-cycle schedule based on cross-layer routing information, enabling nodes to switch between active and sleep modes according to their network allocation vectors (NAVs). This process is further optimized by an enhanced CSMA/CA mechanism, which incorporates sleep/activate modes and pairing nodes to alternate between active and sleep states. This integration reduces collisions, improves channel assessment accuracy, and lowers energy consumption, thereby enhancing overall network performance. EOAMRCL was evaluated in a MATLAB environment, demonstrating superior performance compared with EEUC, DWEHC, and CGA-GWO protocols, particularly in terms of network lifetime and energy consumption. This highlights the effectiveness of integrating GWO and the updated CSMA/CA mechanism in achieving optimal energy efficiency and network performance.

Adapting differential molecular representation with hierarchical prompts for multi-label property prediction.

Accurate prediction of molecular properties is crucial in drug discovery. Traditional methods often overlook that real-world molecules typically exhibit multiple property labels with complex correlations. To this end, we propose a novel framework, HiPM, which stands for Hierarchical Prompted Molecular representation learning framework. HiPM leverages task-aware prompts to enhance the differential expression of tasks in molecular representations and mitigate negative transfer caused by conflicts in individual task information. Our framework comprises two core components: the Molecular Representation Encoder (MRE) and the Task-Aware Prompter (TAP). MRE employs a hierarchical message-passing network architecture to capture molecular features at both the atom and motif levels. Meanwhile, TAP utilizes agglomerative hierarchical clustering algorithm to construct a prompt tree that reflects task affinity and distinctiveness, enabling the model to consider multi-granular correlation information among tasks, thereby effectively handling the complexity of multi-label property prediction. Extensive experiments demonstrate that HiPM achieves state-of-the-art performance across various multi-label datasets, offering a novel perspective on multi-label molecular representation learning.

Resource Cluster-Based Resource Search and Allocation Scheme for Vehicular Clouds in Vehicular Ad Hoc Networks.

Vehicular clouds represent an appealing approach, leveraging vehicles' resources to generate value-added services. Thus, efficiently searching for and allocating resources is a challenge for the successful construction of vehicular clouds. Many recent schemes have relied on hierarchical network architectures using clusters to address this challenge. These clusters are typically constructed based on vehicle proximity, such as being on the same road or within the same region. However, this approach struggles to rapidly search for and consistently allocate resources, especially considering the diverse resource types and varying mobility of vehicles. To address these limitations, we propose the Resource Cluster-based Resource Search and Allocation (RCSA) scheme. RCSA constructs resource clusters based on resource types rather than vehicle proximity. This allows for more efficient resource searching and allocation. Within these resource clusters, RCSA supports both intra-resource cluster search for the same resource type and inter-resource cluster search for different resource types. In RCSA, vehicles with longer connection times and larger resource capacities are allocated in vehicular clouds to minimize cloud breakdowns and communication traffic. To handle the reconstruction of resource clusters due to vehicle mobility, RCSA implements mechanisms for replacing Resource Cluster Heads (RCHs) and managing Resource Cluster Members (RCMs). Simulation results validate the effectiveness of RCSA, demonstrating its superiority over existing schemes in terms of resource utilization, allocation efficiency, and overall performance.

The aging trajectories of brain functional hierarchy and its impact on cognition across the adult lifespan.

The hierarchical network architecture of the human brain, pivotal to cognition and behavior, can be explored via gradient analysis using restingstate functional MRI data. Although it has been employed to understand brain development and disorders, the impact of aging on this hierarchical architecture and its link to cognitive decline remains elusive. This study utilized resting-state functional MRI data from 350 healthy adults (aged 20-85) to investigate the functional hierarchical network using connectome gradient analysis with a cross-age sliding window approach. Gradient-related metrics were estimated and correlated with age to evaluate trajectory of gradient changes across lifespan. The principal gradient (unimodal-to-transmodal) demonstrated a significant non-linear relationship with age, whereas the secondary gradient (visual-to-somatomotor) showed a simple linear decreasing pattern. Among the principal gradient, significant age-related changes were observed in the somatomotor, dorsal attention, limbic and default mode networks. The changes in the gradient scores of both the somatomotor and frontal-parietal networks were associated with greater working memory and visuospatial ability. Gender differences were found in global gradient metrics and gradient scores of somatomotor and default mode networks in the principal gradient, with no interaction with age effect. Our study delves into the aging trajectories of functional connectome gradient and its cognitive impact across the adult lifespan, providing insights for future research into the biological underpinnings of brain function and pathological models of atypical aging processes.

A New Vision of Wireless Edge Caching Networks (WECNs): Issues, Technologies, and Open Research Trends

The future generation of wireless cellular system is expected to accommodate extensive latency-sensitive, tactus-aware, and data-intensive intelligent applications, thereby posing significant challenges on current network architectures. For meeting the challenges, wireless edge caching has been envisioned as a viable enabler. However, the underlying key issues therein, such as reliability, effectiveness, and the joint design, are significantly undervalued. The reliability issue refers to the data-corruption problem incurred by hardware impairments, software upgrades, and network congestion at edges. The effectiveness issue here expresses the unsatisfactory cache hit efficiency. Whilst the joint design aims to acquire the tradeoff among different metrics. In this article, we are dedicated to filling these research gaps. Therewith, we first brief the hierarchical wireless edge caching network architecture with proactive mechanisms. Then, we elaborate on the research roadways towards these critical issues and examine how fault-tolerant coding and bundle recommendation strategies can potentially enhance the reliability and the effectiveness of edge caching in future intelligent communication era. Moreover, the interplay between forgoing mentioned aspects is illuminated as well. A pair of case studies are exemplified to further demonstrate the validity of our devised versatile solutions. At last, the open research opportunities and future trends are identified.

HGNN: A Hierarchical Graph Neural Network Architecture for Joint Resource Management in Dynamic Wireless Sensor Networks

HGNN: A Hierarchical Graph Neural Network Architecture for Joint Resource Management in Dynamic Wireless Sensor Networks

Full Point Encoding for Local Feature Aggregation in 3-D Point Clouds.

Point cloud processing methods exploit local point features and global context through aggregation which does not explicitly model the internal correlations between local and global features. To address this problem, we propose full point encoding which is applicable to convolution and transformer architectures. Specifically, we propose full point convolution (FuPConv) and full point transformer (FPTransformer) architectures. The key idea is to adaptively learn the weights from local and global geometric connections, where the connections are established through local and global correlation functions, respectively. FuPConv and FPTransformer simultaneously model the local and global geometric relationships as well as their internal correlations, demonstrating strong generalization ability and high performance. FuPConv is incorporated in classical hierarchical network architectures to achieve local and global shape-aware learning. In FPTransformer, we introduce full point position encoding in self-attention, that hierarchically encodes each point position in the global and local receptive field. We also propose a shape-aware downsampling block that takes into account the local shape and the global context. Experimental comparison to existing methods on benchmark datasets shows the efficacy of FuPConv and FPTransformer for semantic segmentation, object detection, classification, and normal estimation tasks. In particular, we achieve state-of-the-art semantic segmentation results of 76.8% mIoU on S3DIS sixfold and 73.1% on S3DIS Area 5. Our code is available at https://github.com/hnuhyuwa/FullPointTransformer.

Brep2Seq: a dataset and hierarchical deep learning network for reconstruction and generation of computer-aided design models

Abstract Three-dimensional (3D) reconstruction is a significant research topic in the field of computer-aided design (CAD), which is used to recover editable CAD models from original shapes, including point clouds, voxels, meshes, and boundary representations (B-rep). Recently, there has been considerable research interest in deep model generation due to the increasing potential of deep learning methods. To address the challenges of 3D reconstruction and generation, we propose Brep2Seq, a novel deep neural network designed to transform the B-rep model into a sequence of editable parametrized feature-based modeling operations comprising principal primitives and detailed features. Brep2Seq employs an encoder-decoder architecture based on the transformer, leveraging geometry and topological information within B-rep models to extract the feature representation of the original 3D shape. Due to its hierarchical network architecture and training strategy, Brep2Seq achieved improved model reconstruction and controllable model generation by distinguishing between the primary shape and detailed features of CAD models. To train Brep2Seq, a large-scale dataset comprising 1 million CAD designs is established through an automatic geometry synthesis method. Extensive experiments on both DeepCAD and Fusion 360 datasets demonstrate the effectiveness of Brep2Seq, and show its applicability to simple mechanical components in real-world scenarios. We further apply Brep2Seq to various downstream applications, including point cloud reconstruction, model interpolation, shape constraint generation, and CAD feature recognition.

A Hybrid 3D–2D Feature Hierarchy CNN with Focal Loss for Hyperspectral Image Classification

Hyperspectral image (HSI) classification has been extensively applied for analyzing remotely sensed images. HSI data consist of multiple bands that provide abundant spatial information. Convolutional neural networks (CNNs) have emerged as powerful deep learning methods for processing visual data. In recent work, CNNs have shown impressive results in HSI classification. In this paper, we propose a hierarchical neural network architecture called feature extraction with hybrid spectral CNN (FE-HybridSN) to extract superior spectral–spatial features. FE-HybridSN effectively captures more spectral–spatial information while reducing computational complexity. Considering the prevalent issue of class imbalance in experimental datasets (IP, UP, SV) and real-world hyperspectral datasets, we apply the focal loss to mitigate these problems. The focal loss reconstructs the loss function and facilitates effective achievement of the aforementioned goals. We propose a framework (FEHN-FL) that combines FE-HybridSN and the focal loss for HSI classification and then conduct extensive HSI classification experiments using three remote sensing datasets: Indian Pines (IP), University of Pavia (UP), and Salinas Scene (SV). Using cross-entropy loss as a baseline, we assess the hyperspectral classification performance of various backbone networks and examine the influence of different spatial sizes on classification accuracy. After incorporating focal loss as our loss function, we not only compare the classification performance of the FE-HybridSN backbone network under different loss functions but also evaluate their convergence rates during training. The proposed classification framework demonstrates satisfactory performance compared to state-of-the-art end-to-end deep-learning-based methods, such as 2D-CNN, 3D-CNN, etc.

DANet: Density Adaptive Convolutional Network With Interactive Attention for 3D Point Clouds

Local features and contextual dependencies are crucial for 3D point cloud analysis. Many works have been devoted to designing better local convolutional kernels that exploit the contextual dependencies. However, current point convolutions lack robustness to varying point cloud density. Moreover, contextual modeling is dominated by non-local or self-attention models which are computationally expensive. To solve these problems, we propose density adaptive convolution, coined DAConv. The key idea is to adaptively learn the convolutional weights from geometric connections obtained from the point <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">density</i> and position. To extract precise context dependencies with fewer computations, we propose an interactive attention module (IAM) that embeds spatial information into channel attention along different spatial directions. DAConv and IAM are integrated in a hierarchical network architecture to achieve local density and contextual direction-aware learning for point cloud analysis. Experiments show that DAConv is significantly more robust to point density compared to existing methods and extensive comparisons on challenging 3D point cloud datasets show that our network achieves state-of-the-art classification results of 93.6% on ModelNet40, competitive semantic segmentation results of 68.71% mIoU on S3DIS and part segmentation results of 86.7% mIoU on ShapeNet.

Self-Supervised Learning for Multilevel Skeleton-Based Forgery Detection via Temporal-Causal Consistency of Actions

Skeleton-based human action recognition and analysis have become increasingly attainable in many areas, such as security surveillance and anomaly detection. Given the prevalence of skeleton-based applications, tampering attacks on human skeletal features have emerged very recently. In particular, checking the temporal inconsistency and/or incoherence (TII) in the skeletal sequence of human action is a principle of forgery detection. To this end, we propose an approach to self-supervised learning of the temporal causality behind human action, which can effectively check TII in skeletal sequences. Especially, we design a multilevel skeleton-based forgery detection framework to recognize the forgery on frame level, clip level, and action level in terms of learning the corresponding temporal-causal skeleton representations for each level. Specifically, a hierarchical graph convolution network architecture is designed to learn low-level skeleton representations based on physical skeleton connections and high-level action representations based on temporal-causal dependencies for specific actions. Extensive experiments consistently show state-of-the-art results on multilevel forgery detection tasks and superior performance of our framework compared to current competing methods.

Stool Image Analysis for Digital Health Monitoring By Smart Toilets

Gastrointestinal (GI) conditions are widespread and significantly impact the quality of life and healthcare. Stool appearance is a valuable GI diagnostic indicator, and with the advent of the Internet of Things (IoT), daily monitoring of excreta from a toilet is emerging as a promising digital health tool. This article describes a stool image analysis approach that classifies two physiologically relevant stool characteristics: 1) form and 2) color for an IoT-based smart toilet. We constructed a stool image data set with 3275 images, spanning all seven types of the Bristol stool-form scale (BSFS), a widely used metric for stool consistency and a variety of colors. We used ground-truth data obtained through the annotation of our data set by two gastroenterologists and developed a stool-color card to standardize the labeling of stool colors. We addressed two limitations associated with the application of computer-vision techniques to a smart-toilet system: 1) uneven separability between different stool-form categories and 2) class imbalance in the data set. We present results on hierarchical convolutional neural network (CNN) architectures for training a stool-form classifier and on perceptual color quantization coupled with machine-learning techniques to optimize the color-feature space for the classification of stool color. We utilized an edge–cloud approach to pursue an optimal balance between accuracy and latency and for the classification of stool form, we achieved a balanced accuracy of 84.4% and 84.2% reduction in latency compared to a cloud-model only. For color classification, the logistic-regression (LR) classifier provided 80.8% balanced accuracy and 47.3% reduction in communication latency.

$Q$-Learning Aided Intelligent Routing With Maximum Utility in Cognitive UAV Swarm for Emergency Communications

This article studies the routing problem in a cognitive unmanned aerial vehicle (UAV) swarm (CU-SWARM), which employs the cognitive radio into a swarm of UAVs within a three-layer hierarchical aerial-ground integrated network architecture for emergency communications. In particular, the flexibly converged architecture utilizes a UAV swarm and a high-altitude platform to support aerial sensing and access, respectively, over the disaster-affected areas. We develop a <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$Q$</tex-math></inline-formula> -learning framework to achieve the intelligent routing to maximize the utility for CU-SWARM. To characterize the reward function, we take into account both the routing metric design and the candidate UAV selection optimization. The routing metric jointly captures the achievable rate and the residual energy of UAV. Besides, under the location, arc, and direction constraints, the circular sector is modeled by properly choosing the central angle and the acceptable signal-to-noise ratio for UAV to optimize the candidate UAV selection. With this setup, we further propose a low-complexity iterative algorithm using the dynamic learning rate to update <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$Q$</tex-math></inline-formula> -values during the training process for achieving a fast convergence speed. Simulation results are provided to assess the potential of the <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$Q$</tex-math></inline-formula> -learning framework of intelligent routing as well as to verify our overall iterative algorithm via the dynamic learning rate for training procedure. Our findings reveal that the proposed algorithm converges in a few number of iterations. Furthermore, the proposed algorithm can increase the accumulated rewards, and achieve significant performance gains, as compared to the benchmark schemes.

Swelling-induced Mechanochromism in Multinetwork Polymers.

We report a novel and versatile approach to achieving swelling-induced mechanochemistry using a multinetwork (MN) strategy that enables polymer networks to repeatedly swell with monomers and solvents. The isotropic expansion of the first network (FN) provides sufficient force to drive the mechanochemical scission of a radical-based mechanophore, difluorenylsuccinonitrile (DFSN). Although prompt recombination generally occurs in such highly mobile environments, the resulting pink radicals are kinetically stabilized in the gels, probably due to limited diffusion in the extended polymer chains. Moreover, the DFSN embedded in the isotropically strained chain exhibits increased thermal reactivity, which can be reasonably explained by an entropic contribution of the FN to the dissociation. The utility of the MN polymers is demonstrated not only in terms of swelling-force-induced network modification, but also in the context of tunable reactivity of the dissociative unit through proper design of the hierarchical network architecture.

Swelling‐induced Mechanochromism in Multinetwork Polymers

AbstractWe report a novel and versatile approach to achieving swelling‐induced mechanochemistry using a multinetwork (MN) strategy that enables polymer networks to repeatedly swell with monomers and solvents. The isotropic expansion of the first network (FN) provides sufficient force to drive the mechanochemical scission of a radical‐based mechanophore, difluorenylsuccinonitrile (DFSN). Although prompt recombination generally occurs in such highly mobile environments, the resulting pink radicals are kinetically stabilized in the gels, probably due to limited diffusion in the extended polymer chains. Moreover, the DFSN embedded in the isotropically strained chain exhibits increased thermal reactivity, which can be reasonably explained by an entropic contribution of the FN to the dissociation. The utility of the MN polymers is demonstrated not only in terms of swelling‐force‐induced network modification, but also in the context of tunable reactivity of the dissociative unit through proper design of the hierarchical network architecture.

Content Caching Based on Popularity and Priority of Content Using seq2seq LSTM in ICN

Efficient content caching is essential to address the explosive growth of multimedia contents and most works on content cache placement have been proposed mainly based on the popularity of content. Since the priority of content is also an important attribute of content, we consider both popularity and priority of content together for content caching in information-centric networking (ICN). We define weighted delivery cost of content as content delivery cost multiplied by a weighted sum of popularity and priority of the content. Then, we formulate optimized content cache placement problem to minimize weighted content delivery cost for all content requests in a hierarchical network architecture with multi-access edge computing (MEC) and software-defined networking (SDN) controller. Average quality of experience (QoE), i.e., average content delivery cost, for contents with each priority is imposed as constraint. The number of content requests is predicted based on seq2seq long short-term memory (LSTM) model in MECs and this is delivered to SDN controller. Then, SDN controller obtains predicted popularity of contents and decides content placement in MECs and core routers by solving the content cache placement optimization problem based on binary particle swarm optimization (BPSO). Performance of the proposed content caching scheme is compared with conventional popularity-based, popularity prediction-based, and popularity prediction-based optimization schemes, from the aspect of QoE satisfaction ratio, average cost, weighted average cost, total cost, and weighted total cost. Numerical results show the effectiveness of the proposed scheme at caching content with high priority efficiently, at the expense of caching content with low priority.

Hierarchical Graph Pattern Understanding for Zero-Shot Video Object Segmentation.

The optical flow guidance strategy is ideal for obtaining motion information of objects in the video. It is widely utilized in video segmentation tasks. However, existing optical flow-based methods have a significant dependency on optical flow, which results in poor performance when the optical flow estimation fails for a particular scene. The temporal consistency provided by the optical flow could be effectively supplemented by modeling in a structural form. This paper proposes a new hierarchical graph neural network (GNN) architecture, dubbed hierarchical graph pattern understanding (HGPU), for zero-shot video object segmentation (ZS-VOS). Inspired by the strong ability of GNNs in capturing structural relations, HGPU innovatively leverages motion cues (i.e., optical flow) to enhance the high-order representations from the neighbors of target frames. Specifically, a hierarchical graph pattern encoder with message aggregation is introduced to acquire different levels of motion and appearance features in a sequential manner. Furthermore, a decoder is designed for hierarchically parsing and understanding the transformed multi-modal contexts to achieve more accurate and robust results. HGPU achieves state-of-the-art performance on four publicly available benchmarks (DAVIS-16, YouTube-Objects, Long-Videos and DAVIS-17). Code and pre-trained model can be found at https://github.com/NUST-Machine-Intelligence-Laboratory/HGPU.

Deep learning tomographic reconstruction through hierarchical decomposition of domain transforms.

Deep learning (DL) has shown unprecedented performance for many image analysis and image enhancement tasks. Yet, solving large-scale inverse problems like tomographic reconstruction remains challenging for DL. These problems involve non-local and space-variant integral transforms between the input and output domains, for which no efficient neural network models are readily available. A prior attempt to solve tomographic reconstruction problems with supervised learning relied on a brute-force fully connected network and only allowed reconstruction with a 1284 system matrix size. This cannot practically scale to realistic data sizes such as 5124 and 5126 for three-dimensional datasets. Here we present a novel framework to solve such problems with DL by casting the original problem as a continuum of intermediate representations between the input and output domains. The original problem is broken down into a sequence of simpler transformations that can be well mapped onto an efficient hierarchical network architecture, with exponentially fewer parameters than a fully connected network would need. We applied the approach to computed tomography (CT) image reconstruction for a 5124 system matrix size. This work introduces a new kind of data-driven DL solver for full-size CT reconstruction without relying on the structure of direct (analytical) or iterative (numerical) inversion techniques. This work presents a feasibility demonstration of full-scale learnt reconstruction, whereas more developments will be needed to demonstrate superiority relative to traditional reconstruction approaches. The proposed approach is also extendable to other imaging problems such as emission and magnetic resonance reconstruction. More broadly, hierarchical DL opens the door to a new class of solvers for general inverse problems, which could potentially lead to improved signal-to-noise ratio, spatial resolution and computational efficiency in various areas.

Intelligent Resource Allocation Schemes for UAV-Swarm-Based Cooperative Sensing

Driven by the development of the smart Internet of Things (IoT), unmanned aerial vehicle (UAV) swarms have been widely applied to implement diverse sensing tasks for many IoT applications. By integrating resources of UAVs, a UAV swarm can collaboratively collect and process massive image data for fast system response. However, it is difficult to reduce the computing delay caused by overlapping sensing operations and insufficient computing resources. In addition, transmission delays among UAVs may be intensified due to limited bandwidth resources. To address the mentioned challenges, we propose a UAV-swarm-based hierarchical network architecture to jointly schedule sensing, computing, and communication resources. Specifically, multiple computing groups that are formed by UAVs execute image processing in a pipeline manner to improve computing resource utilization. In order to reduce task execution time, we formulate a nonlinear integer optimization problem for the coordination of heterogeneous resources. A multiagent reinforcement learning (MARL)-based algorithm is designed to find the optimal joint resource allocation strategy under sensing accuracy constraints. Simulation results demonstrate that our algorithm reduces the task execution time while significantly improving the computing resource utilization.