Computational Cost Research Articles

Divisible E-Cash for Billing in Private Ad Retargeting

This paper presents new techniques for private billing in systems for privacy-preserving online advertising. In particular, we show how an ad exchange can use an e-cash scheme to bill advertisers for ad impressions without learning which client saw which ad: The exchange issues electronic coins to advertisers, advertisers pay publishers (via clients) for ad impressions, and publishers unlinkably redeem coins with the exchange. To implement this proposal, we design a new divisible e-cash scheme that uses modern zero-knowledge proofs to reduce the ad exchange's computational costs by roughly 250x compared to the previous state-of-the-art. With our new e-cash scheme, our private-billing infrastructure adds little overhead to existing private ad-retargeting systems: less than 63 ms of latency, negligible client computation, less than 3.2 KB of client communication, and a combined server operating cost (advertisers, publishers, and exchange) of less than 1% of ad spend, an over 5x savings compared to the previous state-of-the-art.

Automated fault diagnosis of rotating machinery using sub domain greedy Network Architecture search

Network Architecture Search (NAS) automates hyperparameter adjustments in deep learning models, offering a potent solution for building intelligent fault diagnosis models. Despite its potential, NAS faces challenges in this application. The primary issues include the excessive time required when searching across the full data domain and the inefficacy of the common random search strategy in yielding high-performance sub networks. This study addresses these challenges through several key initiatives: It defines a specific library of hyperparameters tailored for intelligent fault diagnosis tasks; constructs sub data domain model clusters to reduce the computational costs associated with NAS; and introduces a hyperparameter search strategy leveraging a greedy algorithm, which enhances the search efficiency for optimal diagnostic sub network and minimizes its prediction errors. When applied to datasets involving gear and bearing faults, the proposed method demonstrates a reduction in both the time cost of searches and the prediction errors of sub networks, compared to traditional Network Architecture Search. Moreover, it outperforms manually tuned deep learning models by simultaneously optimizing multiple sets of hyperparameter and automatically generating higher accuracy diagnostic sub networks.

PPDM++: Parallel Point Detection and Matching for Fast and Accurate HOI Detection.

Human-Object Interaction (HOI) detection aims to understand human activities by detecting interaction triplets. Previous HOI detection methods adopt a two-stage instance-driven paradigm. Unfortunately, many non-interactive human-object pairs generated by the first stage are the main obstacle impeding HOI detectors from high efficiency and promising performance. To remedy this, we propose a novel top-down interaction-driven paradigm, detecting interactions first and bridging interactive human-object pairs through interactions. We formulate HOI as a point triplet human point, interaction point, object point and design a Parallel Point Detection and Matching (PPDM) framework. We further take advantage of two-stage methods and propose a novel framework, PPDM++, that detects the interactive human-object pairs by PPDM, then extracts region features for each pair to predict actions. The core of PPDM/PPDM++ is to convert the instance-driven bottom-up paradigm to an interaction-driven top-down paradigm, thus avoiding additional computation costs from traversing a tremendous number of non-interactive pairs. Benefiting from the advanced paradigm, PPDM/PPDM++ has achieved significant performance gains with high efficiency. PPDM-DLA-34 has achieved 19.94 mAP with 42 FPS as the first real-time HOI detector, and PPDM++-SwinB achieves 30.1 mAP with 17 FPS on HICO-DET dataset. We also built an application-oriented database named HOI-A, a supplement to the existing datasets.

Research on Image Encryption Method based on the Chaotic Iteration of a Ternary Nonlinear Function

Objective: Considering that some image encryption algorithms have the disadvantages of complex structure and high computational cost, and there are not many commonly used chaotic systems, which are easy to crack by attacks, to solve these problems, this paper proposes an image encryption algorithm based on three-dimensional nonlinear functions to solve these problems. Methods: The algorithm mainly combines the sinusoidal chaotic map with the ternary nonlinear function system to encrypt the image. Firstly, multiple ternary nonlinear function chaotic systems are designed. Then, the function iteration system is changed to invoke the computation of a specific expression under a random number; it is a chaotic sequence generated according to a chaotic mapping such as sine, and then the value of this chaotic sequence is used to select a ternary nonlinear function for iteration to obtain a chaotic sequence. Finally, the chaotic sequence performs the XOR and scrambling operations on the grey image Results: The algorithm has a simple structure, a better encryption effect, and more incredible difficulty deciphering. Moreover, through the phase diagram and bifurcation diagram, it can be seen that the system has good chaotic characteristics Conclusion: The method in this paper is novel; this method is a random variable order composite operation, which can not only be applied to image encryption but also can be used for fractal map generation and so on, and in some other chaotic fields will have a wide range of applications. It has essential research value.

Multimodal biometric integration: Trends and insights from the past quinquennial

Recent years have seen a notable breakthrough in multimodal biometric systems, which combine numerous biometric modalities for increased security and accuracy. This paper offers a thorough analysis of recent advances in the field, addressing the many multimodal biometric strategies that have been put out by different researchers. Numerous modalities are covered by the evaluation, such as palm print, speech, iris, facial features, palm print, fingerprint, body form, gait, and more. To obtain amazing results in terms of accuracy and security, researchers have used cutting-edge methods like convolutional neural networks (CNNs), support vector machines (SVM), recurrent neural networks (RNNs), deep learning, and optimization algorithms. Many fusion strategies have been investigated to efficiently merge data from many modalities, such as feature-level fusion, decision-level fusion, and score-level fusion. Developments in template protection systems have also addressed security issues related to transmission and storage of biometric data. Even though a lot of the suggested systems have shown great accuracy rates, there are still issues with hardware restrictions, dataset biases, privacy problems, and computational costs. Prospective study avenues encompass investigating more extensive and heterogeneous datasets, crafting resilient fusion algorithms, and amalgamating cutting-edge technologies like deep learning-based biometric cryptosystems and federated learning. All things considered, the literature study demonstrates the enormous potential of multimodal biometric systems in offering safe and dependable authentication solutions for a wide range of applications, from on-line student authentication and proctoring systems to customized healthcare networks.

An automatically recursive feature elimination method based on threshold decision in random forest classification

ABSTRACT The rich feature information contained in the diverse remote sensing data has also exhibited growing potential in the field of image classification. However, the processing of multi-feature data still grapples with the challenges posed by the “curse of dimensionality” and the high computational costs. This paper proposes a remote sensing feature parameter selection method. This method is based on the Gini index of a Random Forest (RF) classifier and employs a 10% threshold decision to identify the optimal combination of remote sensing feature parameters. First, spectral features, texture features, temperature features, elevation features, and principal component features are selected to create a stack of remote sensing images. Second, multiple sets of decision trees are established to cross-validate the contributions of various feature parameters. The feature rankings are determined based on the normalized mean of feature importance. Subsequently, a threshold isset to filter out remote sensing feature parameters that meet the criteria. Finally, an iterative parameter optimization process is carried out to obtain the optimal combination of remote sensing feature parameters. The Landsat 8 image covering Hangzhou Bay in eastern China and the Sentinel-2 remote sensing image of Yancheng Nature Reserve in Jiangsu, China were selected for experiments. The results show that the feature parameters of the remote sensing image screened by the method in this paper are representative, and the proposed method demonstrates strong adaptability and generalization performance in complex environments. The study has important guiding significance for regional spatial planning and sustainable development.

Initial state diversification for efficient AlphaZero-style training

AlphaZero has achieved superhuman performance in Go, chess, and shogi with a general reinforcement learning (RL) algorithm. This achievement is remarkable because AlphaZero does not rely on any training dataset of strong players. However, AlphaZero-style training requires substantial computational resources. Gumbel AlphaZero, a recently introduced more efficient version of AlphaZero, reduces the computational cost of AlphaZero training. The goal of this study is to further improve the playing strength of Gumbel AlphaZero under a limited amount of computational resources. We focus on the diversity in training games, inspired by procedural generation and domain randomization in RL studies, and propose a novel method, initial state diversification. This method diversifies the initial states of a self-play game to encourage the RL agent to understand the game in a more general manner through diverse experiences. For example, in shogi, the initial state of each self-play game is diversified by rearranging the pieces under realistic domain constraints. Experiments demonstrated that training with initial state diversification improves the playing strength of Gumbel AlphaZero in shogi, within the same computational budget for training.

Data-driven discovery of dynamics from time-resolved coherent scattering

Coherent X-ray scattering (CXS) techniques are capable of interrogating dynamics of nano- to mesoscale materials systems at time scales spanning several orders of magnitude. However, obtaining accurate theoretical descriptions of complex dynamics is often limited by one or more factors—the ability to visualize dynamics in real space, computational cost of high-fidelity simulations, and effectiveness of approximate or phenomenological models. In this work, we develop a data-driven framework to uncover mechanistic models of dynamics directly from time-resolved CXS measurements without solving the phase reconstruction problem for the entire time series of diffraction patterns. Our approach uses neural differential equations to parameterize unknown real-space dynamics and implements a computational scattering forward model to relate real-space predictions to reciprocal-space observations. This method is shown to recover the dynamics of several computational model systems under various simulated conditions of measurement resolution and noise. Moreover, the trained model enables estimation of long-term dynamics well beyond the maximum observation time, which can be used to inform and refine experimental parameters in practice. Finally, we demonstrate an experimental proof-of-concept by applying our framework to recover the probe trajectory from a ptychographic scan. Our proposed framework bridges the wide existing gap between approximate models and complex data.

A Maneuver in the Trade-Off Space of Federated Learning Aggregation Frameworks Secured with Polymorphic Encryption: PolyFLAM and PolyFLAP Frameworks

Maintaining user privacy in machine learning is a critical concern due to the implications of data collection. Federated learning (FL) has emerged as a promising solution by sharing trained models rather than user data. However, FL still faces several challenges, particularly in terms of security and privacy, such as vulnerability to inference attacks. There is an inherent trade-off between communication traffic across the network and computational costs on the server or client, which this paper aims to address by maneuvering between these performance parameters. To tackle these issues, this paper proposes two complementary frameworks: PolyFLAM (“Polymorphic Federated Learning Aggregation of Models”) and PolyFLAP (“Polymorphic Federated Learning Aggregation of Parameters”). These frameworks provide two options to suit the needs of users, depending on whether they prioritize reducing communication across the network or lowering computational costs on the server or client. PolyFLAM reduces computational costs by exchanging entire models, eliminating the need to rebuild models from parameters. In contrast, PolyFLAP reduces communication costs by transmitting only model parameters, which are smaller in size compared to entire models. Both frameworks are supported by polymorphic encryption, ensuring privacy is maintained even in cases of key leakage. Furthermore, these frameworks offer five different machine learning models, including support vector machines, logistic regression, Gaussian naïve Bayes, stochastic gradient descent, and multi-layer perceptron, to cover as many real-life problems as possible. The evaluation of these frameworks with simulated and real-life datasets demonstrated that they can effectively withstand various attacks, including inference attacks that aim to compromise user privacy by capturing exchanged models or parameters.

Efficient entanglement purification based on noise guessing decoding

In this paper, we propose a novel bipartite entanglement purification protocol built upon hashing and upon the guessing random additive noise decoding (GRAND) approach recently devised for classical error correction codes. Our protocol offers substantial advantages over existing hashing protocols, requiring fewer qubits for purification, achieving higher fidelities, and delivering better yields with reduced computational costs. We provide numerical and semi-analytical results to corroborate our findings and provide a detailed comparison with the hashing protocol of Bennet et al. Although that pioneering work devised performance bounds, it did not offer an explicit construction for implementation. The present work fills that gap, offering both an explicit and more efficient purification method. We demonstrate that our protocol is capable of purifying states with noise on the order of 10% per Bell pair even with a small ensemble of 16 pairs. The work explores a measurement-based implementation of the protocol to address practical setups with noise. This work opens the path to practical and efficient entanglement purification using hashing-based methods with feasible computational costs. Compared to the original hashing protocol, the proposed method can achieve some desired fidelity with a number of initial resources up to one hundred times smaller. Therefore, the proposed method seems well-fit for future quantum networks with a limited number of resources and entails a relatively low computational overhead.

AHKASH: a new Hybrid particle-in-cell code for simulations of astrophysical collisionless plasma

Abstract We introduce Astrophysical Hybrid-Kinetic simulations with the flASH code ($\texttt {AHKASH}$) – a new Hybrid particle-in-cell (PIC) code developed within the framework of the multi-physics code FLASH. The new code uses a second-order accurate Boris integrator and a predictor-predictor-corrector algorithm for advancing the Hybrid-kinetic equations, using the constraint transport method to ensure that magnetic fields are divergence-free. The code supports various interpolation schemes between the particles and grid cells, with post-interpolation smoothing to reduce finite particle noise. We further implement a δf method to study instabilities in weakly collisional plasmas. The new code is tested on standard physical problems such as the motion of charged particles in uniform and spatially varying magnetic fields, the propagation of Alfvén and whistler waves, and Landau damping of ion acoustic waves. We test different interpolation kernels and demonstrate the necessity of performing post-interpolation smoothing. We couple the TurbGen turbulence driving module to the new Hybrid PIC code, allowing us to test the code on the highly complex physical problem of the turbulent dynamo. To investigate steady-state turbulence with a fixed sonic Mach number, it is important to maintain isothermal plasma conditions. Therefore, we introduce a novel cooling method for Hybrid PIC codes and provide tests and calibrations of this method to keep the plasma isothermal. We describe and test the ‘hybrid precision’ method, which significantly reduces (by a factor ∼1.5) the computational cost, without compromising the accuracy of the numerical solutions. Finally, we test the parallel scalability of the new code, showing excellent scaling up to 10,000 cores.

Proper orthogonal decomposition reduced-order model of the global oceans

AbstractA reduced-order model (ROM) of the global oceans is developed by projecting the hydrostatic Boussinesq equations of motion onto a proper orthogonal decomposition (POD) basis. Three-dimensional POD modes are calculated from the ocean fields of an ensemble climate reanalysis dataset. The coefficients in the POD ROM are calculated using a regression approach. The performance of various POD ROM configurations are assessed. Each configuration is derived from an alternate sea-water equation of state, linking the density and temperature fields. POD ROM variants incorporating an equation of state in which density is a quadratic function of temperature, are able to reproduce the statistics of the large-scale structures at a fraction of the computational cost required to numerically simulate this flow. Due to the speed and efficiency of calculation, such reduced-order models of the global geophysical system will enable researchers and policy makers to assess the physical risk for a broader range of potential future climate scenarios.

A volume‐of‐solid implicit volume penalty method for moving‐body flows

AbstractAn original Immersed Boundary Method for solving moving body flows is proposed. This method couples (i) a Lagrangian Volume‐of‐Solid description of the solid object avoiding conservation issues and (ii) a robust implicit volume penalty forcing embedded in a low‐Mach number projection method to account for the solid's impact on the fluid dynamics. A new composite velocity field is introduced to describe both solid and fluid domains in a single set of governing equations. The accuracy of the method has been assessed on several academic cases, involving stationary or moving bodies and with different mesh resolutions. The predicted forces on the solid are in excellent agreement with body‐fitted reference cases. The system of equations is also proven to be fully mass conservative. Application of the method on a two‐dimensional vertical axis turbine case shows a reduction in computational cost compared to a body‐fitted method.

Machine Learning-Based Surrogate Model for Genetic Algorithm with Aggressive Mutation for Feature Selection

The genetic algorithm with aggressive mutations GAAM, is a specialised algorithm for feature selection. This algorithm is dedicated to the selection of a small number of features and allows the user to specify the maximum number of features desired. A major obstacle to the use of this algorithm is its high computational cost, which increases significantly with the number of dimensions to be retained. To solve this problem, we introduce a surrogate model based on machine learning, which reduces the number of evaluations of the fitness function by an average of 48% on the datasets tested, using the standard parameters specified in the original paper. Additionally, we experimentally demonstrate that eliminating the crossover step in the original algorithm does not result in any visible changes in the algorithm’s results. We also demonstrate that the original algorithm uses an artificially complex mutation method that could be replaced by a simpler method without loss of efficiency. The sum of the improvements resulted in an average reduction of 53% in the number of evaluations of the fitness functions. Finally, we have shown that these outcomes apply to parameters beyond those utilized in the initial article, while still achieving a comparable decrease in the count of evaluation function calls. Tests were conducted on 9 datasets of varying dimensions, using two different classifiers.

A data‐assisted physics‐informed neural network (DA‐PINN) for fretting fatigue lifetime prediction

AbstractIn this study, we present for the first time the application of physics‐informed neural network (PINN) to fretting fatigue problems. Although PINN has recently been applied to pure fatigue lifetime prediction, it has not yet been explored in the case of fretting fatigue. We propose a data‐assisted PINN (DA‐PINN) for predicting fretting fatigue crack initiation lifetime. Unlike traditional PINN that solves partial differential equations for specific problems, DA‐PINN combines experimental or numerical data with physics equations as part of the loss function to enhance prediction accuracy. The DA‐PINN method, employed in this study, consists of two main steps. First, damage parameters are obtained from the finite element method by using critical plane method, which generates a data set used to train an artificial neural network (ANN) for predicting damage parameters in other cases. Second, the predicted damage parameters are combined with the experimental parameters to form the input data set for the DA‐PINN models, which predict fretting fatigue lifetime. The results demonstrate that DA‐PINN outperforms ANN in terms of prediction accuracy and eliminates the need for high computational costs once the damage parameter data set is constructed. Additionally, the choice of loss‐function methods in DA‐PINN models plays a crucial role in determining its performance.

Implicit hydromechanical representation of fractures using a continuum approach

Fractures control fluid flow, solute transport, and mechanical deformation in crystalline media. They can be modeled numerically either explicitly or implicitly via an equivalent continuum. The implicit framework implies lower computational cost and complexity. However, upscaling heterogeneous fracture properties for its implicit representation as an equivalent fracture layer remains an open question. In this study, we propose an approach, the Equivalent Fracture Layer (EFL), for the implicit representation of fractures surrounded by low-permeability rock matrix to accurately simulate hydromechanical coupled processes. The approach assimilates fractures as equivalent continua with a manageable scale (≫1 μm) that facilitates spatial discretization, even for large-scale models including multiple fractures. Simulation results demonstrate that a relatively thick equivalent continuum layer (in the order of cm) can represent a fracture (with aperture in the order of μm) and accurately reproduce the hydromechanical behavior (i.e., fluid flow and deformation/stress behavior). There is an upper bound restriction due to the Young's modulus because the equivalent fracture layer should have a lower Young's modulus than that of the surrounding matrix. To validate the approach, we model a hydraulic stimulation carried out at the Bedretto Underground Laboratory for Geosciences and Geoenergies in Switzerland by comparing numerical results against measured data. The method further improves the ability and simplicity of continuum methods to represent fractures in fractured media.

Efficient general method for numerically modeling laser pulse propagation, overlap, and lifetime effects in amplifiers

An efficient numerical time-dependent general method is developed to address incoherent pulse overlap and lifetime effects in laser amplifiers. The alternating propagation-population laser energetics method (APPLE) has been validated against a semi-discrete coupled rate equation numerical method (SDRE) and analytic formalisms in bounding cases. APPLE is based on decoupled rates applied to a time-dependent framework where both space-time-dependent populations and pulse energetics are consistently updated in each time step. A significant advantage of APPLE lies in its conceptual simplicity, ease of implementation, and relatively small computational cost. SDRE tracks the populations through coupled rates and uses the method of lines to discretize the hyperbolic partial differential transport equations allowing for use of ordinary differential equation solvers. With reasonably sized mesh, we report both energetic and power pulse shape relative differences on the order of one percent between the models over a large range of initial conditions.

An IGBT coupling structure with a smart service life reliability predictor using active learning

Abstract An effective approach is proposed to evaluate the service life reliability of a multi-physics coupling structure of an insulated gate bipolar transistor (IGBT) module. The node-based smoothed finite element method with stabilization terms is firstly employed to construct an electrical-thermal-mechanical (ETM) coupling structure of the IGBT module, based on which the multi-physics responses can be accurately calculated to predict the service life of the IGBT module. By using the high-quality sample data obtained through the ETM coupling model, a Monte Carlo based active learning Kriging metamodel (AK-MCS) is developed to assess the service life reliability of the IGBT module, which can greatly reduce the computational cost needed by the surrogate model construction and reliability analysis. Numerical results show that the proposed ETM coupling structure can produce high-quality sample data of the IGBT dynamics and the AK-MCS machine learning technique can accurately estimate the service life reliability of the IGBT module.

An Attribute-Based End-to-End Policy-Controlled Signcryption Scheme for Secure Group Chat Communication

Secure instant communication is an important topic of information security. A group chat is a highly convenient mode of instant communication. Increasingly, companies are adopting group chats as a daily office communication tool. However, a large volume of messages in group chat communication can lead to message overload, causing group members to miss important information. Additionally, the communication operator’s server may engage in the unreliable behavior of stealing information from the group chat. To address these issues, this paper proposes an attribute-based end-to-end policy-controlled signcryption scheme, aimed at establishing a secure and user-friendly group chat communication mode. By using the linear secret sharing scheme (LSSS) with strong expressive power to construct the access structure in the signcryption technology, the sender can precisely control the recipients of the group chat information to avoid message overload. To minimize computational cost, a signcryption step with constant computational overhead is designed. Additionally, a message-sending mechanism combining “signcryption + encryption” is employed to prevent the operator server from maliciously stealing group chat information. Rigorous analysis shows that PCE-EtoE can resist adaptive chosen-ciphertext attacks under the standard model. Simulation results demonstrate that our theoretical derivation is correct, and that the PCE-EtoE scheme outperforms existing schemes in terms of computational cost, making it suitable for group chat communication.

Comparisons of Inverse Dynamics Formulations in a Spatial Redundantly Actuated Parallel Mechanism Constrained by Two Point Contact Higher Kinematic Pairs

A spatial redundantly actuated parallel mechanism (RAPM) constrained by two point contact higher kinematic pairs (HKPs) has been designed, arising from the inspiration of mastication in human beings: the end effector is the lower jaw, the six kinematic chains are the primary chewing muscles, and the constraints at HKPs are the temporomandibular joints. In this paper, firstly, the constrained motions of the mechanism are described in detail; thereafter, five models are formulated by the well-known Newton–Euler’s law, the Lagrangian equations, and the principle of virtual work, to explore its rigid-body inverse dynamics. The symbolic results show that the model structures based on these approaches are quite different: the model via the Newton–Euler law well reflects the nature of the mechanism in terms of the constraint forces from HKPs with six equations and eight unknowns, and the existence of reaction forces at the spherical joints is tightly dependent on the number of kinematic chains. In comparison, from the latter two methods, the constraint forces and the reaction forces at spherical joints do not appear in the four models in which there are only four equations and six unknowns. Further, by using the dynamics model of the non-redundantly actuated counterpart as the core in both the second models from the energy and virtual work-related methods, their computational cost is only about 16.7% and 36.63% of the two first models, respectively. Finally, the comparisons between the dynamics models of the RAPM and its counterpart clarify that the HKP constraints greatly alter the model structures and raise the technical difficulties.