Low Computational Cost Research Articles

Phase retrieval via conjugate gradient minimization in double-plane lensless holographic microscopy

Optimization-based phase retrieval method for digital lensless holographic microscopy in the double-plane recording configuration is proposed. In our method the phase retrieval is framed as an optimization problem that can be efficiently and rigorously tackled with gradient decent tools. This is done with the conjugate gradient method that possesses excellent theoretical features such as global and fast convergence (compared to steepest descent) and relatively low computational cost (compared to second order optimizers). The proposed method is extensively tested with simulations and experimental measurements that show superiority of our method over the Gerchberg-Saxton algorithm, especially in terms of reconstruction of problematic low frequency components of viable phase information.

CIBORG: CIrcuit-Based and ORiented Graph theory permutation routing protocol for single-hop IoT networks

The Internet of Things (IoT) has emerged as a promising paradigm which facilitates the seamless integration of physical devices and digital systems, thereby transforming multiple sectors such as healthcare, transportation, and urban planning. This paradigm is also known as ad-hoc networks. IoT is characterized by several pieces of equipment called objects. These objects have different and limited capacities such as battery, memory, and computing power. These limited capabilities make it difficult to design routing protocols for IoT networks because of the high number of objects in a network. In IoT, objects often have data which does not belong to them and which should be sent to other objects, then leading to a problem known as permutation routing problems. The solution to that problem is found when each object receives its items. In this paper, we propose a new approach to addressing the permutation routing problem in single-hop IoT networks. To this end, we start by representing an IoT network as an oriented graph, and then, based on a reservation channel protocol, we first define a permutation routing protocol for an IoT in a single channel. Secondly, we generalize the previous protocol to make it work in multiple channels. Routing is done using graph theory approaches. The obtained results show that the wake-up times and activities of IoT objects are greatly reduced, thus optimizing network lifetime. This is an effective solution for the permutation routing problem in IoT networks. The proposed approach considerably reduces energy consumption and computation time. It saves 5.2 to 32.04% residual energy depending on the number of items and channels used. Low energy and low computational cost demonstrate that the performance of circuit-based and oriented graph theory is better than the state-of-the-art protocol and therefore is a better candidate for the resolution of the permutation routing problem in single-hop environment.

ReadCurrent: a VDCNN-based tool for fast and accurate nanopore selective sequencing.

Nanopore selective sequencing allows the targeted sequencing of DNA of interest using computational approaches rather than experimental methods such as targeted multiplex polymerase chain reaction or hybridization capture. Compared to sequence-alignment strategies, deep learning (DL) models for classifying target and nontarget DNA provide large speed advantages. However, the relatively low accuracy of these DL-based tools hinders their application in nanopore selective sequencing. Here, we present a DL-based tool named ReadCurrent for nanopore selective sequencing, which takes electric currents as inputs. ReadCurrent employs a modified very deep convolutional neural network (VDCNN) architecture, enabling significantly lower computational costs for training and quicker inference compared to conventional VDCNN. We evaluated the performance of ReadCurrent across 10 nanopore sequencing datasets spanning human, yeasts, bacteria, and viruses. We observed that ReadCurrent achieved a mean accuracy of 98.57% for classification, outperforming four other DL-based selective sequencing methods. In experimental validation that selectively sequenced microbial DNA from human DNA, ReadCurrent achieved an enrichment ratio of 2.85, which was higher than the 2.7 ratio achieved by MinKNOW using the sequence-alignment strategy. In summary, ReadCurrent can rapidly classify target and nontarget DNA with high accuracy, providing an alternative in the toolbox for nanopore selective sequencing. ReadCurrent is available at https://github.com/Ming-Ni-Group/ReadCurrent.

Accuracy and efficiency of finite element head models: The role of finite element formulation and material laws.

Traumatic brain injury is a significant problem worldwide. In the United States of America, around 1.7 million cases are documented annually, displaying the need for a deeper understanding of the effects on the human brain. The tests required for this assessment are very complex. Tests on cadavers may raise serious ethical questions, and in vivo crash tests are not viable. In this context, there is a great need to developing finite element head models (FEHM) to study the biomechanics of the tissues when submitted to a certain impact or acceleration/deceleration scenario. An excellent compromise between accuracy and CPU efficiency is always desirable for a FEHM, For this reason, this work focuses on the improvement of an existing head model, including the study of the behavior of the brain using distinct finite element types. The finite element type and formulation is of utmost importance for the general accuracy and efficiency of the models. Several validations were performed, comparing the simulation results against experimental data. The simulations with hexahedral elements, under specific conditions, obtained more accurate results with a lower computational cost. Using hexahedrals, a comparison was also performed using two material characterizations with more than 10 years apart, using the latest finite element head model validation experiment. Overall, the newer material model displays a less stiff response, although its implementation must always depend on the overall purpose of the model it is being applied to.

Maximizing supply chain performance leveraging machine learning to anticipate customer backorders

The complexity of global supply chains, with their multi-tiered and lengthy structures, presents significant challenges for effectively planning inventory replenishment, accurately forecasting demand, and managing customer backorders. To maintain customer loyalty and avoid extended waiting periods, companies need to have an efficient system for predicting product backlog for customers without overstocking. Traditional statistical techniques like regression analysis can forecast demand and the probability of customer backorders. Nevertheless, they are restricted to modeling linear relationships and may not be well-suited for capturing complex relationships.On the other hand, analytical methods like machine learning (ML) show great promise. However, ML algorithms can be computationally intensive, especially when dealing with many predictors, which can make models complex and increase computational costs. Thus, simpler models with fewer attributes can make data collection and model complexity more manageable. In this study we assess the efficacy and accuracy of simplified ML algorithms, the impact of utilizing limited, high-impact predictors in supply chain backorder prediction. Using publicly available data sets from Kaggle, we developed two sets of models: one with 22 predictors and another with only the top five predictors. The results demonstrate a significant decrease in computational costs, ranging from 30 % to 98 %, with only a marginal reduction in accuracy and F1-score, ranging from 0.6 % to 4.2 %. These findings underscore the potential for simpler backorder prediction models, which helps streamline data collection with lower computational cost. This study enhances the current literature by providing insights into optimizing customer backorder prediction with potential generalization for various types of supply chains. It strikes a balance between accuracy and computational efficiency, making the findings valuable for practical implementation across various industries.

Parameter-Fitting-Free Continuum Modeling of Electric Double Layer in Aqueous Electrolyte.

Electric double layers (EDLs) play fundamental roles in various electrochemical processes. Despite the extensive history of EDL modeling, there remain challenges in the accurate prediction of its structure without expensive computation. Herein, we propose a predictive multiscale continuum model of EDL that eliminates the need for parameter fitting. This model computes the distribution of the electrostatic potential, electron density, and species' concentrations by taking the extremum of the total grand potential of the system. The grand potential includes the microscopic interactions that are newly introduced in this work: polarization of solvation shells, electrostatic interaction in parallel plane toward the electrode, and ion-size-dependent entropy. The parameters that identify the electrode and electrolyte materials are obtained from independent experiments in the literature. The model reproduces the trends in the experimental differential capacitance with multiple electrode and nonadsorbing electrolyte materials (Ag(110) in NaF, Ag(110) in NaClO4, and Hg in NaF), which verifies the accuracy and predictiveness of the model and rationalizes the observed values to be due to changes in electron stability. However, our calculation on Pt(111) in KClO4 suggests the need for the incorporation of electrode/ion-specific interactions. Sensitivity analyses confirmed that effective ion radius, ion valence, the electrode's Wigner-Seitz radius, and the bulk modulus of the electrode are significant material properties that control the EDL structure. Overall, the model framework and findings provide insights into EDL structures and predictive capability at low computational cost.

Rapid, Accurate, Ranking of Protein-Ligand Binding Affinities with VM2, the Second-Generation Mining Minima Method.

The structure-based technologies most widely used to rank the affinities of candidate small molecule drugs for proteins range from faster but less reliable docking methods to slower but more accurate explicit solvent free energy methods. In recent years, we have advanced another technology, which is called mining minima because it "mines" out the main contributions to the chemical potentials of the free and bound molecular species by identifying and characterizing their main local energy minima. The present study provides systematic benchmarks of the accuracy and computational speed of mining minima, as implemented in the VeraChem Mining Minima Generation 2 (VM2) code, across two well-regarded protein-ligand benchmark data sets, for which there are already benchmark data for docking, free energy, and other computational methods. A core result is that VM2's accuracy approaches that of explicit solvent free energy methods at a far lower computational cost. In finer-grained analyses, we also examine the influence of various run settings, such as the treatment of crystallographic water molecules, on the accuracy, and define the costs in time and dollars of representative runs on Amazon Web Services (AWS) compute instances with various CPU and GPU combinations. We also use the benchmark data to determine the importance of VM2's correction from generalized Born to finite-difference Poisson-Boltzmann results for each energy well and find that this correction affords a remarkably consistent improvement in accuracy at a modest computational cost. The present results establish VM2 as a distinctive technology for early-stage drug discovery, which provides a strong combination of efficiency and predictivity.

Efficient Parameterization of Density Functional Tight-Binding for 5f-Elements: A Th-O Case Study.

Density functional tight binding (DFTB) models for f-element species are challenging to parametrize owing to the large number of adjustable parameters. The explicit optimization of the terms entering the semiempirical DFTB Hamiltonian related to f orbitals is crucial to generating a reliable parametrization for f-block elements, because they play import roles in bonding interactions. However, since the number of parameters grows quadratically with the number of orbitals, the computational cost for parameter optimization is much more expensive for the f-elements than for the main group elements. In this work we present a set of efficient approaches for mitigating the hurdle imposed by the large size of the parameter space. A novel group-by-orbital correction functions for two-center bond integrals was developed. With this approach the number of parameters is reduced, and it grows linearly with the number of elements, maintaining the accuracy and the number of parameters, in the case of f elements, by more than 40%. The parameter optimization step was accelerated by means of the mini-batch BFGS method. This method allows parameter optimizations with much larger training sets than other single batch methods. A stochastic optimizer was employed that helped overcome shallow local minima in the objective function. The proposed algorithm was used to parametrize the DFTB Hamiltonian for the Th-O system, which was subsequently applied to the study of ThO2 nanoparticles. The training set consisted of 6322 unique structures, which is barely feasible with conventional optimization methods. The optimized parameter set, LANL-ThO, displays good agreement with DFT-calculated properties such as energies, forces, and structures for both clusters and bulk ThO2. Benefiting from the fewer number of parameters and lower computational costs for objective function evaluations, this new approach shows its potential applications in DFTB parametrization for elements with high angular momentum, which present a challenge to conventional methods.

Revocable identity-based designated verifier proxy re-signature with signature evolution

Identity-based proxy re-signature (IDPRS) enables a semi-trusted proxy to convert one user’s signature to another user’s for the same message without revealing any secret information of both users’ signing keys. In IDPRS, user revocation is a necessary problem arising from privilege expiry or private key compromise. A revoked user cannot produce new signatures including re-signatures from the proxy. There is a need for some applications like a contract management system to keep expired message and signature pairs valid in the next period. The conventional solution is to require the unrevoked users to sign these messages again, which burdens users heavily. Therefore, establishing an IDPRS system with revocation and efficient signature updating is crucial. This paper introduces the concept of revocable IDPRS with signature evolution, facilitating efficient signature updating. The signature evolution enables the proxy to batch update non-revoked users’ signatures from old ones to new ones, while also offering precise execution of signature updates on specific messages. Moreover, to further protect the privacy of signature validity, we employ designated verifier in our scheme. Formal security proofs are provided, and performance evaluation demonstrates our scheme’s lower computation costs for signature updating and its comprehensive security attributes.

ICS-ResNet: A Lightweight Network for Maize Leaf Disease Classification

The accurate identification of corn leaf diseases is crucial for preventing disease spread and improving corn yield. Plant leaf images are often affected by factors such as complex backgrounds, climate, light, and sample data imbalance. To address these issues, we propose a lightweight convolutional neural network, ICS-ResNet, based on ResNet50. This network incorporates improved spatial and channel attention modules as well as a deep separable residual structure to enhance recognition accuracy. (1) The residual connections in the ResNet network prevent gradient loss during deep network training. (2) The improved channel attention (ICA) and spatial attention (ISA) modules fully utilize semantic information from different feature layers to accurately localize key features of the network. (3) To reduce the number of parameters and lower computational costs, we replace traditional convolutional computation with a depth-separable residual structure. (4) We also employ cosine annealing to dynamically adjust the learning rate, enhancing the network’s training stability, improving model convergence, and preventing local optima. Experiments on the corn dataset in Plant Village compare the proposed ICS-ResNet with eight popular networks: CSPNet, InceptionNet_v3, EfficientNet, ShuffleNet, MobileNet, ResNet50, ResNet101 and ResNet152. The results show that the ICS-ResNet achieves an accuracy of 98.87%, which is 5.03%, 3.18%, 1.13%, 1.81%, 1.13%, 0.68%, 0.44% and 0.60% higher than the other networks, respectively. Furthermore, the number of parameters and computations are reduced by 69.21% and 54.88%, respectively, compared to the original ResNet50 network, significantly improving the efficiency of corn leaf disease classification. The study provides strong technical support for sustainable agriculture and the promotion of agricultural science and technology innovation.

AQuaRef: Machine learning accelerated quantum refinement of protein structures.

Cryo-EM and X-ray crystallography provide crucial experimental data for obtaining atomic-detail models of biomacromolecules. Refining these models relies on library-based stereochemical restraints, which, in addition to being limited to known chemical entities, do not include meaningful noncovalent interactions relying solely on nonbonded repulsions. Quantum mechanical (QM) calculations could alleviate these issues but are too expensive for large molecules. We present a novel AI-enabled Quantum Refinement (AQuaRef) based on AIMNet2 neural network potential mimicking QM at substantially lower computational costs. By refining 41 cryo-EM and 30 X-ray structures, we show that this approach yields atomic models with superior geometric quality compared to standard techniques, while maintaining an equal or better fit to experimental data.

Ultrafast jet classification at the HL-LHC

Three machine learning models are used to perform jet origin classification. These models are optimized for deployment on a field-programmable gate array device. In this context, we demonstrate how latency and resource consumption scale with the input size and choice of algorithm. Moreover, the models proposed here are designed to work on the type of data and under the foreseen conditions at the CERN large hadron collider during its high-luminosity phase. Through quantization-aware training and efficient synthetization for a specific field programmable gate array, we show that O(100) ns inference of complex architectures such as Deep Sets and Interaction Networks is feasible at a relatively low computational resource cost.

Latent Space Representations for Marker-Less Realtime Hand-Eye Calibration.

Marker-less hand-eye calibration permits the acquisition of an accurate transformation between an optical sensor and a robot in unstructured environments. Single monocular cameras, despite their low cost and modest computation requirements, present difficulties for this purpose due to their incomplete correspondence of projected coordinates. In this work, we introduce a hand-eye calibration procedure based on the rotation representations inferred by an augmented autoencoder neural network. Learning-based models that attempt to directly regress the spatial transform of objects such as the links of robotic manipulators perform poorly in the orientation domain, but this can be overcome through the analysis of the latent space vectors constructed in the autoencoding process. This technique is computationally inexpensive and can be run in real time in markedly varied lighting and occlusion conditions. To evaluate the procedure, we use a color-depth camera and perform a registration step between the predicted and the captured point clouds to measure translation and orientation errors and compare the results to a baseline based on traditional checkerboard markers.

MambaReID: Exploiting Vision Mamba for Multi-Modal Object Re-Identification.

Multi-modal object re-identification (ReID) is a challenging task that seeks to identify objects across different image modalities by leveraging their complementary information. Traditional CNN-based methods are constrained by limited receptive fields, whereas Transformer-based approaches are hindered by high computational demands and a lack of convolutional biases. To overcome these limitations, we propose a novel fusion framework named MambaReID, integrating the strengths of both architectures with the effective VMamba. Specifically, our MambaReID consists of three components: Three-Stage VMamba (TSV), Dense Mamba (DM), and Consistent VMamba Fusion (CVF). TSV efficiently captures global context information and local details with low computational complexity. DM enhances feature discriminability by fully integrating inter-modality information with shallow and deep features through dense connections. Additionally, with well-aligned multi-modal images, CVF provides more granular modal aggregation, thereby improving feature robustness. The MambaReID framework, with its innovative components, not only achieves superior performance in multi-modal object ReID tasks, but also does so with fewer parameters and lower computational costs. Our proposed MambaReID's effectiveness is validated by extensive experiments conducted on three multi-modal object ReID benchmarks.

Directionality of earthquake-induced slope displacements from numerical analysis and sliding block approaches

Earthquake ground shaking intensity differs at different azimuthal orientations (ground motion directionality), which could also result in orientational variations in earthquake-induced slope displacements. Therefore, the range of seismic displacements (e.g., the median and maximum values over all orientations) that may occur by considering the directions of downslope movement and strong shaking is important to evaluate seismic landslide hazard of site-specific slopes. This study performs comprehensive directionality analyses of seismic slope displacement based on two-dimensional numerical and sliding-block approaches. Four slope models with clay and sand soil layers and slope angle of 30° and 40° are analyzed using 75 ground motion records rotated over all orientations. Different sliding-block models are considered, including rigid-block analysis and two coupled flexible-block representations of one-dimensional soil columns with only sliding mass and full site respectively. The results indicate that the orientation variation of slope displacements is much larger than the ground motion parameters. The directionality characteristics of seismic slope displacement from numerical calculation can be well captured by the sliding-block analyses, although significant differences in the computed displacement values are observed. A procedure is proposed to obtain the maximum and median values of seismic slope displacement over all orientations from advanced numerical analysis at low computational cost.

Spin-Spin Coupling Constant Based on Reference Interaction Site Model Self-Consistent Field with Constrained Spatial Electron Density.

A method for computing spin-spin coupling constants (SSCCs) using the reference interaction site model self-consistent field with constrained spatial electron density (RISM-SCF-cSED) is proposed for the first time. Describing solvents using integral equation theory allows us to reflect solvent effects at atomic resolution in SSCCs while accounting for thermal fluctuations at a low computational cost. Applying the method to water, 1,1-difluoroethylene, and 1-methylaminomethylene-2-naphthalenone revealed that the solvent shift was evaluated to a greater extent than in the continuum solvent model. The origin of this phenomenon was analyzed in terms of the physical mechanisms underlying SSCCs.

Forecasting ecological water demand of an arid oasis under a drying climate scenario based on deep learning methods

Ecological water diversion projects (EWDP) are an effective management tool for restoring oasis ecosystems in arid regions. Given the potential for drier climatic conditions in arid regions in the future, it is essential to develop water diversion strategies that can adapt to climate change and reduce the risk of oasis ecosystem degradation. Here, this study used a Bayesian optimization-based long- and short-term memory (BO-LSTM) model to determine the optimal amount of water diversion needed to maintain healthy growth of oasis vegetation under future climate change scenarios in the Qingtu Oasis, which is a typical downstream oasis of inland rivers restored by EWDP in China. The results showed that the BO-LSTM model effectively captured the response of oasis vegetation to changes in water inundation areas and drought stress with low computational cost and high accuracy. The study revealed that regional vegetation became more vulnerable than previously thought when extreme drought and a drying trend were taken into account. It was found that if the amount of water entering the oasis ranges from 10 to 15 million m3, there will be a decline in the growth of oasis vegetation as indicated by the normalized difference vegetation index (NDVI). Even if current levels of water diversion (20 million m3) are maintained, oasis vegetation may still face growth decline due to meteorological drought. The optimal amount of water diversion was determined to be 25 million m3, resulting in a 21.7% increase in NDVI regardless of drought events. This study represents an innovative approach as it couples EWDP, climate change, and oasis vegetation dynamics based on deep learning models, which unveils divergent responses of oasis vegetation to climate change and EWDP and verifies a non-linear relationship between water diversion amounts and ecological benefits generated.

Using Reinforcement Learning and Error Models for Drone Precise Landing

We propose a novel framework for achieving precision landing in drone services. The proposed framework consists of two distinct decoupled modules, each designed to address a specific aspect of landing accuracy. The first module is concerned with intrinsic errors, where new error models are introduced. This includes a spherical error model that takes into account the orientation of the drone. Additionally, we propose a live position correction algorithm that employs the error models to correct for intrinsic errors in real time. The second module focuses on external wind forces and presents an aerodynamics model with wind generation to simulate the drone’s physical environment. We utilize reinforcement learning to train the drone in simulation with the goal of landing precisely under dynamic wind conditions. Experimental results, conducted through simulations and validated in the physical world, demonstrate that our proposed framework significantly increases landing accuracy while maintaining a low onboard computational cost.

Perturbation theory in the complete degenerate active space (CDAS-PT2).

Methods based on the multireference perturbation theory (MRPT) with the one-electron zeroth-order Hamiltonian are widely used for the description of excited states, for example, due to their relatively low computational cost. However, current methods have a common drawback-use of a model space with low size. In this article, we propose the MRPT method with the model space extended to the complete active space. The one-electron zeroth-order Hamiltonian suitable for this extension is formulated. The proposed method was applied to common models, such as LiF, ethylene, and trans-butadiene. It was shown to have accuracy superior to XMCQDPT2 in most cases, especially in the case of the small active space.

A numerical oxidation model for SiC‐coated thermal protection systems in hypersonic applications

AbstractSilicon carbide (SiC) is a widely used component for thermal protection system (TPS) design. This material is often used either as a coating layer to protect beneath carbon‐based material against oxidation or as a ceramic matrix for composite materials used in the hypersonic leading edges. A numerical equilibrium model for SiC‐coated material ablation is developed in this study to compute the material response of coated TPS in hypersonic relevant aerothermal environment. The developed model is based on an equilibrium approach similar to the carbon graphite oxidation framework for ablative charring materials. The equilibrium wall composition is calculated based on the minimization of the Gibbs free energy. The proposed method can capture numerically the transition from passive to active oxidation while maintaining a low computational cost and good numerical stability. Implementation of the model in the Porous Material Analysis Toolbox based on OpenFOAM code permits to compare against previous literature test cases. By comparing the surface temperature obtained from the model with the experimental data, it is observed that the numerical model gives a good estimation providing accurate surface temperature prediction.