Computational Expenditure Research Articles

Fast‐MedNeXt: Accelerating the MedNeXt Architecture to Improve Brain Tumour Segmentation Efficiency

ABSTRACTWith the rapid development of medical imaging technology, 3D image segmentation technology has gradually become a mainstream method, especially in brain tumour detection and diagnosis showing its unique advantages. The technique makes full use of 3D spatial information to locate and analyze tumours more accurately, thus playing an important role in improving diagnostic accuracy, optimising treatment planning and promoting research. However, it also suffers from significant computational expenditure and delayed processing pace. In this paper, we propose an innovative optimisation scheme to address this problem. We thoroughly investigate the MedNeXt network and propose Fast‐MedNeXt, which aims to increase the processing speed while maintaining accuracy. First, we introduce the partial convolution (PConv) technique, which replaces the deep convolutional layers in the network. This improvement effectively reduces computation and memory requirements while maintaining efficient feature extraction. Second, based on PConv, we propose PConv‐Down and PConv‐Up modules, which are applied to the up‐sampling and down‐sampling modules to further optimise the network structure and improve efficiency. To confirm the efficacy of the approach, we carried out a sequence of tests in the multimodal brain tumour segmentation challenge 2021 (BraTS2021). By comparing with the MedNeXt series network, the Fast‐MedNeXt reduced the latency by 22.1%, 20.5%, 15.8%, and 11.4% respectively, while the average accuracy also increased by 0.475% and 0.2% respectively. These significant performance improvements demonstrate the effectiveness of Fast‐MedNeXt in 3D medical image segmentation tasks and provide a new and more efficient solution for the field.

Optimizing Attention and Cognitive Control Costs Using Temporally Layered Architectures.

The current reinforcement learning framework focuses exclusively on performance, often at the expense of efficiency. In contrast, biological control achieves remarkable performance while also optimizing computational energy expenditure and decision frequency. We propose a decision-bounded Markov decision process (DB-MDP) that constrains the number of decisions and computational energy available to agents in reinforcement learning environments. Our experiments demonstrate that existing reinforcement learning algorithms struggle within this framework, leading to either failure or suboptimal performance. To address this, we introduce a biologically inspired, temporally layered architecture (TLA), enabling agents to manage computational costs through two layers with distinct timescales and energy requirements. TLA achieves optimal performance in decision-bounded environments and in continuous control environments, matching state-of-the-art performance while using a fraction of the computing cost. Compared to current reinforcement learning algorithms that solely prioritize performance, our approach significantly lowers computational energy expenditure while maintaining performance. These findings establish a benchmark and pave the way for future research on energy and time-aware control.

Edge-device collaborative computing for multi-view classification

Motivated by the proliferation of Internet-of-Thing (IoT) devices and the rapid advances in the field of deep learning, there is a growing interest in pushing deep learning computations, conventionally handled by the cloud, to the edge of the network to deliver faster responses to end users, reduce bandwidth consumption to the cloud, and address privacy concerns. However, to fully realize deep learning at the edge, two main challenges still need to be addressed: (i) how to meet the high resource requirements of deep learning on resource-constrained devices, and (ii) how to leverage the availability of multiple streams of spatially correlated data, to increase the effectiveness of deep learning and improve application-level performance. To address the above challenges, we explore collaborative inference at the edge, in which edge nodes and end devices share correlated data and the inference computational burden by leveraging different ways to split computation and fuse data. Besides traditional centralized and distributed schemes for edge-end device collaborative inference, we introduce selective schemes that decrease bandwidth resource consumption by effectively reducing data redundancy. As a reference scenario, we focus on multi-view classification in a networked system in which sensing nodes can capture overlapping fields of view. The proposed schemes are compared in terms of accuracy, computational expenditure at the nodes, communication overhead, inference latency, robustness, and noise sensitivity. Experimental results highlight that selective collaborative schemes can achieve different trade-offs between the above performance metrics, with some of them bringing substantial communication savings (from 18% to 74% of the transmitted data with respect to centralized inference) while still keeping the inference accuracy well above 90%.

Accelerated calculation of phase-variable for numerical simulation of multiphase flows

In this manuscript, we unveil an innovative acceleration technique for the simulation of multiphase flows, which builds upon the foundation laid by our previously devised multiphase flow solver. The cornerstone of this method lies in augmenting computational efficiency through the selective updating of variables solely within domains characterized by significant gradients of the phase variable. This tactic diminishes the dimensionality of the system of linear equations, thereby hastening the computational process. To pinpoint the regions warranting focused attention, a judicious criterion is indispensable to strike an optimal balance between efficiency and precision. This criterion affords a marked decrement in computational expenditure while preserving the fidelity of the original methodology. Rigorous validations corroborate that this acceleration mechanism can enhance computational efficiency by a minimum of 49.5% in resolving the phase field equation and by at least 70.2% in the computation of pragmatic two-phase flows, without compromising accuracy. Moreover, the efficacy of this acceleration technique is inversely proportional to the rate of interface evolution, becoming increasingly efficient as the interface evolves more slowly.

On efficient computation in active inference

Biological agents demonstrate a remarkable proficiency in calibrating appropriate scales of planning and evaluation when interacting with their environments. It follows logically that any decision-making algorithm aspiring to neurobiological plausibility must mirror these attributes, particularly regarding computational expenditure and the intricacy of evaluative processes. However, active inference encounters notable challenges in simulating apt behaviours within complex environments. These stem chiefly from its substantial computational demands and the intricate task of defining the agent’s behaviour preference. We address these through a two-fold approach. First, we introduce a planning algorithm by using the Bellman-optimality principle to minimise the planning cost function (i.e., expected free energy). Briefly, we recursively compute the expected free energy of actions in reverse temporal sequence to significantly reduce the computational complexity. Secondly, inspired by the Z-learning algorithm, we propose a novel method to learn time-constrained agent preferences. We face-validate the efficacy of these through grid-world simulations and demonstrate precise model learning and planning, even under uncertainty. These algorithmic advances create new opportunities for various applications—in neuroscience and machine learning.

Hybrid surrogate model combining physics and data for seismic drift estimation of shear‐wall structures

AbstractTo address the issue of costly computational expenditure related to high‐fidelity numerical models, surrogate models have been widely used in various engineering tasks, including design optimization. Despite the successful application of the existing surrogate models, physics‐based models depend largely on simplifications and assumptions, which render parameter calibration challenging; whereas data‐driven models require substantial data to reach their full potential, with their performance often being constrained in tasks when obtaining massive data is difficult. In this study, a hybrid surrogate model is proposed combining physics‐based and data‐driven models to rapidly estimate building seismic responses. The application of this model is exemplified through effective estimation of inter‐story drift ratios (IDRs), being a critical factor in shear‐wall structure design. Initially, a data augmentation technique and a parametric modeling procedure are introduced to significantly enhance the dataset diversity. Subsequently, a task decomposition strategy is proposed to effectively integrate a data‐driven graph neural network (GNN) and a physics‐based flexural‐shear model. Additionally, the output layer and the loss function of the GNN are modified to enhance the estimation accuracy by eliminating fundamental errors. Results of numerical experiments indicate that the proposed hybrid model can complete IDR estimations in an average time of 0.56 s, with a mean absolute percentage error of 12.7%. This performance significantly surpasses that of existing purely data‐driven and physics‐based models. A case study shows that the efficiency of the proposed hybrid model is approximately 100 times greater than that of conventional finite element software. This enables an accurate assessment of the design compliance with code requirements. The results of this study can be applied to the design optimization of seismic‐resistant building structures.

Low illumination fog noise image denoising method based on ACE-GPM.

The Perona-Malik (P-M) model exhibits deficiencies such as noise amplification, new noise introduction, and significant gradient effects when processing noisy images. To address these issues, this paper proposes an image-denoising algorithm, ACE-GPM, which integrates an Automatic Color Equalization (ACE) algorithm with a gradient-adjusted P-M model. Initially, the ACE algorithm is employed to enhance the contrast of low-light images obscured by fog and noise. Subsequently, the Otsu method, a technique to find the optimal threshold based on between-class variance, is applied for precise segmentation, enabling more accurate identification of different regions within the image. After that, distinct gradients enhance the image's foreground and background via an enhancement function that accentuates edge and detailed information. The denoising process is finalized by applying the gradient P-M model, employing a gradient descent approach to further emphasize image edges and details. Experimental evidence indicates that the proposed ACE-GPM algorithm not only elevates image contrast and eliminates noise more effectively than other denoising methods but also preserves image details and texture information, evidenced by an average increase of 0.42 in the information entropy value. Moreover, the proposed solution achieves these outcomes with reduced computational resource expenditures while maintaining high image quality.

Study on the hydrodynamics of structured packing: Liquid holdup and pressure drop of a novel 3D printed packing

SpiroPak is a patented innovative structured packing and can be used in process industries for chemical separation purposes. The core of its design is to ensure an uninterrupted flow path for both liquids and gases, leading to remarkably low-pressure drops but with substantial surface areas for efficient mass transfer. Experimental studies using the wire-mesh sensor technique have showcased SpiroPak's ability to eliminate the channelling effect that is observed in commercial packings. Furthermore, its wide-ranging liquid holdup capability encompasses the functionalities of conventional packings, offering clients the feasibility of diverse applications. In this study, the pressure-drop correlations for SpiroPak in this work are based on Stichlmair empirical formulas, with the coefficients of C1, C2 and C3 tuned through the curve fitting method. The primary objective of this study was to establish packing selection criteria based on the key hydraulic performance metrics, such as pressure drop and liquid holdup. By swiftly identifying suitable packing solutions, the method accelerates the pre-selection process, significantly reducing computational complexity and time expenditures at the early stages of research and development.

A deep learning approach for hydrofoil optimization of tidal turbines

Hydrofoil optimization plays a crucial role in improving the hydrodynamic performance of tidal turbines. However, the high-dimensional representations in hydrofoil design space require significant Computational Fluid Dynamics (CFD) simulations, increasing computational load and time expenditure. This paper proposes a hydrofoil optimization framework based on deep learning to achieve an effective mapping between design parameters, pressure fields, and hydrodynamic characteristics. The designed hydrofoil represented using the Signed Distance Function (SDF) is input to Convolutional Neural Networks (CNN) for fast prediction of hydrodynamic coefficients and pressure field. Due to the effective integration of the prediction model and genetic optimization algorithm, this framework can generate a large number of smooth hydrofoils and rapidly predict their hydrodynamic performance to achieve efficient optimization of hydrofoils. The optimized hydrofoil shapes obtained based on the above framework exhibit a higher lift-to-drag ratio than common hydrofoils. Furthermore, the optimized hydrofoils are applied to design tidal energy turbine blades. The simulation results demonstrate that the hydrodynamic performance of tidal turbines can be effectively improved through hydrofoil design optimization. The proposed framework enhances the design efficiency of the tidal turbine rotor and provides turbine hydrofoils with higher power coefficients.

DVSAI: Diverse View-Shared Anchors Based Incomplete Multi-View Clustering

In numerous real-world applications, it is quite common that sample information is partially available for some views due to machine breakdown or sensor failure, causing the problem of incomplete multi-view clustering (IMVC). While several IMVC approaches using view-shared anchors have successfully achieved pleasing performance improvement, (1) they generally construct anchors with only one dimension, which could deteriorate the multi-view diversity, bringing about serious information loss; (2) the constructed anchors are typically with a single size, which could not sufficiently characterize the distribution of the whole samples, leading to limited clustering performance. For generating view-shared anchors with multi-dimension and multi-size for IMVC, we design a novel framework called Diverse View-Shared Anchors based Incomplete multi-view clustering (DVSAI). Concretely, we associate each partial view with several potential spaces. In each space, we enable anchors to communicate among views and generate the view-shared anchors with space-specific dimension and size. Consequently, spaces with various scales make the generated view-shared anchors enjoy diverse dimensions and sizes. Subsequently, we devise an integration scheme with linear computational and memory expenditures to integrate the outputted multi-scale unified anchor graphs such that running spectral algorithm generates the spectral embedding. Afterwards, we theoretically demonstrate that DVSAI owns linear time and space costs, thus well-suited for tackling large-size datasets. Finally, comprehensive experiments confirm the effectiveness and advantages of DVSAI.

Convex Hull Prediction for Adaptive Video Streaming by Recurrent Learning.

Adaptive video streaming relies on the construction of efficient bitrate ladders to deliver the best possible visual quality to viewers under bandwidth constraints. The traditional method of content dependent bitrate ladder selection requires a video shot to be pre-encoded with multiple encoding parameters to find the optimal operating points given by the convex hull of the resulting rate-quality curves. However, this pre-encoding step is equivalent to an exhaustive search process over the space of possible encoding parameters, which causes significant overhead in terms of both computation and time expenditure. To reduce this overhead, we propose a deep learning based method of content aware convex hull prediction. We employ a recurrent convolutional network (RCN) to implicitly analyze the spatiotemporal complexity of video shots in order to predict their convex hulls. A two-step transfer learning scheme is adopted to train our proposed RCN-Hull model, which ensures sufficient content diversity to analyze scene complexity, while also making it possible to capture the scene statistics of pristine source videos. Our experimental results reveal that our proposed model yields better approximations of the optimal convex hulls, and offers competitive time savings as compared to existing approaches. On average, the pre-encoding time was reduced by 53.8% by our method, while the average Bjøntegaard delta bitrate (BD-rate) of the predicted convex hulls against ground truth was 0.26%, and the mean absolute deviation of the BD-rate distribution was 0.57%.

EPPTA: Efficient partially observable reinforcement learning agent for penetration testing applications

AbstractIn recent years, penetration testing (pen‐testing) has emerged as a crucial process for evaluating the security level of network infrastructures by simulating real‐world cyber‐attacks. Automating pen‐testing through reinforcement learning (RL) facilitates more frequent assessments, minimizes human effort, and enhances scalability. However, real‐world pen‐testing tasks often involve incomplete knowledge of the target network system. Effectively managing the intrinsic uncertainties via partially observable Markov decision processes (POMDPs) constitutes a persistent challenge within the realm of pen‐testing. Furthermore, RL agents are compelled to formulate intricate strategies to contend with the challenges posed by partially observable environments, thereby engendering augmented computational and temporal expenditures. To address these issues, this study introduces EPPTA (efficient POMDP‐driven penetration testing agent), an agent built on an asynchronous RL framework, designed for conducting pen‐testing tasks within partially observable environments. We incorporate an implicit belief module in EPPTA, grounded on the belief update formula of the traditional POMDP model, which represents the agent's probabilistic estimation of the current environment state. Furthermore, by integrating the algorithm with the high‐performance RL framework, sample factory, EPPTA significantly reduces convergence time compared to existing pen‐testing methods, resulting in an approximately 20‐fold acceleration. Empirical results across various pen‐testing scenarios validate EPPTA's superior task reward performance and enhanced scalability, providing substantial support for efficient and advanced evaluation of network infrastructure security.

Battle Models: Inception ResNet vs. Extreme Inception for Marine Fish Object Detection

Within the domain of deep learning applied to computer vision, there exists a significant emphasis on the competition between two prominent models, namely Inception ResNet and Xception, particularly in the field of marine fish object detection. The present study conducted a comparative analysis of two advanced neural network architectures in order to assess their efficacy in the identification and localization of marine fish species in underwater images. The two models underwent a rigorous evaluation, utilizing their capabilities in feature extraction. The findings indicate a complex performance landscape, wherein Inception ResNet exhibits remarkable accuracy in identifying marine fish objects, while Xception demonstrates superior computational efficiency. The present study elucidates the inherent trade-off between precision and computational expenditure, offering valuable perspectives on the pragmatic ramifications of choosing one model over another. Furthermore, this research underscores the significance of carefully choosing a suitable model that aligns with the particular requirements of object detection applications in the context of marine fish. This study endeavors to guide professionals and scholars in marine biology and computer vision, enabling them to make well-informed choices when utilizing deep learning techniques to detect maritime fish objects in underwater settings. The research specifically focuses on the comparison between Inception ResNet and Xception models.

Integrated optical and thermal model to investigate the performance of a solar parabolic dish collector coupled with a cavity receiver

The cavity receiver is the most promising type of receiver, which drops the heat loss and converts the incoming concentrated radiation to useful energy. In this work, the coupled optical-thermal investigation is performed for a novel cavity receiver of a 40 m2 parabolic dish collector to estimate the thermal performance. The variation of thermal performance with inclination γ, wind direction ψ, wind speed V, and inlet temperature THTF,i has been discussed. SolTrace and ANSYS® Fluent are used for optical and thermal modeling, respectively. Surface errors, heat flux at each coordinate of the receiver, and the parabolic dish effect have been considered in the study. The text user interface (TUI) over graphical has been adopted in ANSYS® to reduce the computational expenditure for variation of parameters. In thermal modeling, a user-defined function (UDF) is applied for mapping data from SolTrace simulation with 107 rays. Therminol 66, as heat transfer fluid, is used in the receiver coils. The optical performance of 84.15% is achieved at an aiming height of 4.5 m. The thermal performance of 73.11% is obtained as maximum in no-wind conditions for 90° tilt. The worst thermal performance is obtained in side-on wind. Further, the performance has been compared with other models and found to be performing well.

COST-WINNERS: COST reduction WIth Neural NEtworks-based augmented Random Search for simultaneous thermal and electrical energy storage control

The combination of local renewable energy production, dynamic loads, and multiple energy storage systems with different dynamics requires sophisticated control systems to maximize the energy cost efficiency of the combined energy system. Battery and thermal energy storage systems can be combined to increase the local use of on-site renewable energy, reduce peak power demand, and exploit time-of-use energy pricing. In this paper, we focus on how the augmented random search algorithm and artificial neural networks can be used together to solve an energy cost optimization problem involving the control of a battery energy storage system and a thermal energy storage system at the same time in a smart warehouse. As part of this work, a simulated training environment made using the data from the smart warehouse’s operations. In addition to the energy storage systems, the warehouse energy system has integrated a large roof mounted photovoltaic power plant and an industrial-scale cooling system.The developed solution is able to minimize the energy costs by modulating both energy systems, depending on the situation. Additionally, when it is tested against the state-of-the-art solutions, our developed solution at worst matches performance when the alternative algorithm is allowed to increase training time by a factor of nearly three. On average, our presented solution doubles the performance of the benchmark algorithm with much less computational resource expenditure.

A robust ECC-based authentication framework for energy internet (EI)-based vehicle to grid communication system

The integration of electric vehicles (EVs) into the smart grid system through Vehicle-to-Grid (V2G) technology enables bi-directional power delivery, allowing EVs to distribute excess electricity to the grid and recharge as needed. However, V2G networks face significant security and privacy challenges due to the involvement of large amounts of data and untrusted entities. To address these challenges, we propose a new authentication protocol based on Elliptic Curve Cryptography (ECC) that enables secure communication between EVs and charging stations in V2G networks. The proposed plan functions by going through four primary phases, which are: 1) Initialization, 2) Registration, 3) Authentication and 4) password change, vehicle revoke and new joining. Our protocol aims to maintain secure connections while minimizing computation and communication costs through the use of lightweight cryptographic operations such as one-way hash functions, concatenation, and bitwise Xor operations. We validated the security of our protocol through both informal and formal security analyses, including verification with the Scyther Verification tool, which confirms that the proposed protocol is free from security threats within bounds. Our ECC-based authentication protocol for V2G is more secure and lightweight compared to other related protocols in the same context. Our contributions are: 1) A new ECC-based authentication framework for energy Internet (EI)-based V2G communication systems, and 2) The use of lightweight cryptographic operations to reduce computational expenditure and improve resource utilization. Overall, our proposed protocol provides a robust and secure framework for V2G communication that can address the significant security and privacy challenges facing V2G networks.

Leveraging conscious and nonconscious learning for efficient AI.

Various interpretations of the literature detailing the neural basis of learning have in part led to disagreements concerning how consciousness arises. Further, artificial learning model design has suffered in replicating intelligence as it occurs in the human brain. Here, we present a novel learning model, which we term the "Recommendation Architecture (RA) Model" from prior theoretical works proposed by Coward, using a dual-learning approach featuring both consequence feedback and non-consequence feedback. The RA model is tested on a categorical learning task where no two inputs are the same throughout training and/or testing. We compare this to three consequence feedback only models based on backpropagation and reinforcement learning. Results indicate that the RA model learns novelty more efficiently and can accurately return to prior learning after new learning with less computational resources expenditure. The final results of the study show that consequence feedback as interpretation, not creation, of cortical activity creates a learning style more similar to human learning in terms of resource efficiency. Stable information meanings underlie conscious experiences. The work provided here attempts to link the neural basis of nonconscious and conscious learning while providing early results for a learning protocol more similar to human brains than is currently available.

A modified convolution model for calculating the far field directivity of a parametric array loudspeaker.

The far field directivity is a straightforward indicator to describe the radiation pattern of the audio sound generated by a parametric array loudspeaker (pal), but its accurate and computationally efficient prediction is still challenging at present. This paper derives two-dimensional (2D), three-dimensional (3D), and 3D axisymmetric convolution models for calculating the far field directivity based on the quasilinear solution of Westervelt equation. The obtained expressions are expressed as linear and spherical convolutions of the ultrasound directivity and Westervelt directivity for 2D and 3D models, respectively. To improve prediction accuracy, the obtained expression is multiplied by an effective directivity resulted from the aperture factor of audio sound. The calculated directivities are compared against the exact solution obtained using the cylindrical and spherical wave expansions for 2D and 3D models, respectively. Numerical results with piston, apodized, and steerable profiles in both 2D and 3D models show that the proposed modified convolution model agrees well with the exact solution. It is also found that sidelobes appear in the audio sound directivity at large aperture sizes and high audio frequencies due to the aperture factor of audio sound, which can be predicted with the proposed method with a relatively low computational expenditure.

Deep Unfolded Gridless DOA Estimation Networks Based on Atomic Norm Minimization

Deep unfolded networks have recently been regarded as an essential way to direction of arrival (DOA) estimation due to the fast convergence speed and high interpretability. However, few consider gridless DOA estimation. This paper proposes two deep unfolded gridless DOA estimation networks to resolve the above problem. We first consider the atomic norm-based 1D and decoupled atomic norm-based 2D gridless DOA models solved by the alternating iterative minimization of variables, respectively. Then, the corresponding deep networks are trained offline after constructing the corresponding complete training datasets. At last, the trained networks are applied to realize the 1D DOA and 2D estimation, respectively. Simulation results reveal that the proposed networks can secure higher 1D and 2D DOA estimation performances while maintaining a lower computational expenditure than typical methods.

Low-cost quasi-global optimization of expensive electromagnetic simulation models by inverse surrogates and response features

Conceptual design of contemporary high-frequency structures is typically followed by a careful tuning of their parameters, predominantly the geometry ones. The process aims at improving the relevant performance figures, and may be quite expensive. The reason is that conventional design methods, e.g., based on analytical or equivalent network models, often only yield rough initial designs. This is especially the case for miniaturized components featuring considerable electromagnetic (EM) cross couplings, or antenna systems with non-negligible radiator coupling (e.g., MIMO, closely-spaced arrays). For reliability reasons, parametric optimization is carried out using EM simulation tools, which is a time-consuming task. In many cases, designer needs to resort to a global search, especially when handling several objectives and constraints is necessary, or the high-frequency structure under design is overly complex. Combination of both aforementioned factors makes it no longer possible to rely on engineering insight, even to detect a promising region of the design space. Unfortunately, nature-inspired algorithms, commonly employed for solving these tasks typically exhibit significant computational expenditures. This paper proposes a simple yet efficient method for globalized search using a response feature approach and inverse regression surrogates. Owing to less nonlinear dependence of the feature point coordinates on the system variables (as compared to the original responses, e.g., S-parameter frequency characteristics), our methodology permits a rapid identification of the most appropriate regions of the parametric space, and further design tuning by means of local routines. At the same time, the overall optimization cost is comparable to the cost of local procedures. The proposed approach is validated using several high-frequency structures (a dual-band antenna, a microstrip coupler, an impedance matching transformer) optimized under different design scenarios. Global search capability and computational efficiency are demonstrated through comprehensive comparisons with multiple-start local search, as well as particle swarm optimizer, a representative nature-inspired algorithm.