Computational Burden Research Articles

An agent-based method to estimate 3D cell migration trajectories from 2D measurements: Quantifying and comparing T vs CAR-T 3D cell migration

Background and objective:Immune cell migration is one of the key features that enable immune cells to find invading pathogens, control tissue damage, and eliminate primary developing tumors. Chimeric antigen receptor (CAR) T-cell therapy is a novel strategy in the battle against various cancers. It has been successful in treating hematological tumors, yet it still faces many challenges in the case of solid tumors. In this work, we evaluate the three-dimensional (3D) migration capacity of T and CAR-T cells within dense collagen-based hydrogels. Quantifying three-dimensional (3D) cell migration requires microscopy techniques that may not be readily accessible. Thus, we introduce a straightforward mathematical model designed to infer 3D trajectories of cells from two-dimensional (2D) cell trajectories. Methods:We develop a 3D agent-based model (ABM) that simulates the temporal changes in the direction of migration with an inverse transform sampling method. Then, we propose an optimization procedure to accurately orient cell migration over time to reproduce cell migration from 2D experimental cell trajectories. With this model, we simulate cell migration assays of T and CAR-T cells in microfluidic devices conducted under hydrogels with different concentrations of type I collagen and validate our 3D cell migration predictions with light-sheet microscopy. Results:Our findings indicate that CAR-T cell migration is more sensitive to collagen concentration increases than T cells, resulting in a more pronounced reduction in their invasiveness. Moreover, our computational model reveals significant differences in 3D movement patterns between T and CAR-T cells. T cells exhibit migratory behavior in 3D whereas that CAR-T cells predominantly move within the XY plane, with limited movement in the Z direction. However, upon the introduction of a CXCL12 chemical gradient, CAR-T cells present migration patterns that closely resemble those of T cells. Conclusions:This framework demonstrates that 2D projections of 3D trajectories may not accurately represent real migration patterns. Moreover, it offers a tool to estimate 3D migration patterns from 2D experimental data, which can be easily obtained with automatic quantification algorithms. This approach helps reduce the need for sophisticated and expensive microscopy equipment required in laboratories, as well as the computational burden involved in producing and analyzing 3D experimental data.

Adaptive Dynamic Programming for Robust Event-Driven Tracking Control of Nonlinear Systems With Asymmetric Input Constraints.

This article considers the robust dynamic event-driven tracking control problem of nonlinear systems having mismatched disturbances and asymmetric input constraints. Initially, to tackle the asymmetric constraints, a novel nonquadratic value function is constructed for the original system. This makes the asymmetrically constrained tracking control problem transformed into an unconstrained optimal regulation problem. Then, a dynamic event-driven mechanism is proposed. Meanwhile, the event-driven Hamilton-Jacobi-Bellman equation (ED-HJBE) is developed for the optimal regulation problem in order to acquire the optimal control with distinctly decreased computational burden. To solve the ED-HJBE, a single critic neural network (CNN) is designed in the adaptive dynamic programming framework. Meanwhile, the gradient descent method is employed to update the CNN's weights. After that, both the weight estimation error and the tracking error are proved to be uniformly ultimately bounded via Lyapunov's direct method. Finally, simulations of the spring-mass-damper system and the pendulum plant are separately utilized to validate the established theoretical claims.

Planning with mental models – Balancing explanations and explicability

Human-aware planning involves generating plans that are explicable, i.e. conform to user expectations, as well as providing explanations when such plans cannot be found. In this paper, we bring these two concepts together and show how an agent can achieve a trade-off between these two competing characteristics of a plan. To achieve this, we conceive a first-of-its-kind planner MEGA that can reason about the possibility of explaining a plan in the plan generation process itself. We will also explore how solutions to such problems can be expressed as “self-explaining plans” – and show how this representation allows us to leverage classical planning compilations of epistemic planning to reason about this trade-off at plan generation time without having to incur the computational burden of having to search in the space of differences between the agent model and the mental model of the human in the loop in order to come up with the optimal trade-off. We will illustrate these concepts in two well-known planning domains, as well as with a robot in a typical search and reconnaissance task. Human factor studies in the latter highlight the usefulness of the proposed approach.

SmartImpute: A Targeted Imputation Framework for Single-cell Transcriptome Data.

Single-cell RNA sequencing (scRNA-seq) has revolutionized our understanding of cellular heterogeneity and tissue transcriptomic complexity. However, the high frequency of dropout events in scRNA-seq data complicates downstream analyses such as cell type identification and trajectory inference. Existing imputation methods address the dropout problem but face limitations such as high computational cost and risk of over-imputation. We present SmartImpute, a novel computational framework designed for targeted imputation of scRNA-seq data. SmartImpute focuses on a predefined set of marker genes, enhancing the biological relevance and computational efficiency of the imputation process while minimizing the risk of model misspecification. Utilizing a modified Generative Adversarial Imputation Network architecture, SmartImpute accurately imputes the missing gene expression and distinguishes between true biological zeros and missing values, preventing overfitting and preserving biologically relevant zeros. To ensure reproducibility, we also provide a function based on the GPT4 model to create target gene panels depending on the tissue types and research context. Our results, based on scRNA-seq data from head and neck squamous cell carcinoma and human bone marrow, demonstrate that SmartImpute significantly enhances cell type annotation and clustering accuracy while reducing computational burden. Benchmarking against other imputation methods highlights SmartImpute's superior performance in terms of both accuracy and efficiency. Overall, SmartImpute provides a lightweight, efficient, and biologically relevant solution for addressing dropout events in scRNA-seq data, facilitating deeper insights into cellular heterogeneity and disease progression. Furthermore, SmartImpute's targeted approach can be extended to spatial omics data, which also contain many missing values.

A single-loop reliability sensitivity analysis strategy for time-dependent rare events with both random variables and stochastic processes

To deal with the time-dependent reliability sensitivity (TDRS) analysis of rare events with both random variables and stochastic processes, a single-loop sampling procedure combined with the cross entropy-based importance sampling strategy (CEIS), namely SCEIS, is derived to lighten the computational burden without loss of precision. Firstly, the TDRS indices are derived to make them implementable by a single-loop importance sampling (IS) procedure. Then, the quasi-optimal IS density is obtained through the extension of the cross-entropy method, which is an adaptive procedure that overcomes the challenges of determining the position and number of design points in traditional IS. Furthermore, the SCEIS can be easily implemented to estimate the TDRS indices for problems with small failure probability. Three examples involving numerical and engineering problems are analyzed to demonstrate the performance of the proposed method. The results obtained from comparing with other existing methods reveal that the proposed method provides satisfactory TDRS analysis for rare events while significantly reducing computational costs.

Two-stage data-driven optimal energy management and dynamic real-time operation in networked microgrid based on a deep reinforcement learning approach

Given the significant challenges posed by the vast and diverse data in energy management, this study introduces a two-stage approach: optimal energy management system (OEMS) and dynamic real-time operation (DRTOP). These stages employ a multi-agent policy-oriented deep reinforcement learning (DRL) approach, aiming to minimize operating and energy exchange costs through interactions in the networked microgrid (NMG) energy market. The primary objectives include minimizing the distribution system operator (DSO) cost and optimizing the exchanged power between DSO and NMG, and the power transmission losses and the secondary include minimizing MG’s operating cost, optimal use of renewable energy resources (RER) and energy storage systems (ESS), minimizing the exchanged power cost with the main grid and, risk analysis. The OEMS&DRTOP model is developed based on the Stackelberg game theory and the DRL structure. The DRL model is developed in two offline learning and online distributed operation phases to minimize the computational burden, time, and DRL operation process. This study’s results show the high efficiency of the presented approach to minimizing the operating cost, the exchanged power based on the price uncertainty, power transmission losses, and, RER and ESSs optimal participation. In addition, regarding computational load, the proposed concept demonstrates a 12.9% reduction compared to the dueling deep Q-network method and a 17% reduction compared to the deep Q-network method. Also regarding computational time, the proposed concept demonstrates a 17.13% reduction compared to the dueling deep Q-network method and a 25.6% reduction compared to the deep Q-network method.

A Low-Computational Burden Closed-Form Approximated Expression for MSE Applicable for PTP with gfGn Environment

The Precision Time Protocol (PTP) plays a pivotal role in achieving precise frequency and time synchronization in computer networks. However, network delays and jitter in real systems introduce uncertainties that can compromise synchronization accuracy. Three clock skew estimators designed for the PTP scenario were obtained in our earlier work, complemented by closed-form approximations for the Mean Squared Error (MSE) under the generalized fractional Gaussian noise (gfGn) model, incorporating the Hurst exponent parameter (H) and the a parameter. These expressions offer crucial insights for network designers, aiding in the strategic selection and implementation of clock skew estimators. However, substantial computational resources are required to fit each expression to the gfGn model parameters (H and a) from the MSE perspective requirement. This paper introduces new closed-form estimates that approximate the MSE tailored to match gfGn scenarios that have a lower computational burden compared to the literature-known expressions and that are easily adaptable from the computational burden point of view to different pairs of H and a parameters. Thus, the system requires less substantial computational resources and might be more cost-effective.

Efficient aerodynamic design using BézierGAN and model order reduction: A computational study

Bézier Generative Adversarial Networks (BézierGANs) is one of the most widely used Convolutional Neural Network (CNN) technique-based algorithms for generating smooth curves to analyze and optimize complex shapes and flow patterns, particularly aerodynamics. Airfoils, the basic cross-sectional shape to design air wings, are generated here by selecting the vital parameters wisely after alternating the typical BézierGANs algorithm instead of relying on pre-existing models. When these predicted airfoils are used to develop the physical structure of the adjacent air layer to observe the phenomena, they become the subject of computationally expensive simulations of turbulent airflow. Simulations of turbulent airflow that are computationally expensive are required when these anticipated airfoils are utilized to create the physical structure of the adjacent air layer to monitor the physical events. To prevail over this computational burden, the Model Order Reduction (MOR) technique is applied to develop the best-fitted lower-order model of the original turbulent air layer model around the airfoils by filtering the most effective system's singular values while preserving all the essential characteristics of the original large-scale model. As evidence of the effectiveness of the Reduced Order Model (ROM) of the turbulent air layer regarding the truncation of the design time and computational costs, the comparative analysis of the computational time is occupied by the original and prescribed system to ensure about 99.7660% of less time consumption by the ROM. The regeneration of the reduced-order physical model is also covered in this paper, which analyzes the physical behaviors regarding the air pressure and the velocity patterns of the airflow around the airfoils and compares the similarities to the physical characteristics of the Full Order Model (FOM). It was found that the ROM can be one of the best alternatives to FOM in real-life implementation.

HybridPy: The Simulation Suite for Mesoscopic and Microscopic Traffic Simulations

Mesoscopic, agent-based simulations efficiently model and assess entire regions’ daily activities and travel patterns, exemplified by smaller countries like Switzerland. The queue-based simulation represents a compromise between computational speed on the one hand and the necessity of detailed modeling infrastructure on the other hand. Thus, mesoscopic simulations enable an efficient and reasonably detailed analysis of the complex interplay between supply and demand in mobility research. Conversely, microsimulations excel at reproducing individual speed profiles and behavior by modeling the interactions between traffic participants, including pedestrians, bicycles, and scooters. Although allowing for more detailed system analysis, the downside is the high computational burden, which often prevents large-scale microscopic simulations from running in optimization or calibration loops. hybridPY, an extension of SUMOPy, aims to close the gap and benefit from both environments. The simulation suite allows the running of mesoscopic as well as microscopic traffic simulations based on the core idea: running a microscopic simulation in a smaller dedicated area, using the routes or mobility plans generated from a larger mesoscopic model. The main features of this software are: (i) import, editing and visualization of MATSim and BEAM CORE networks; (ii) conversion of MATSim plans to SUMO routes or plans within the SUMO area; (iii) configuring and running of MATSim simulations. The capability of hybridPY is demonstrated by two applications: the simulation of Schwabing, Germany, based on the MITO MATSim model, and the San Francisco municipality, USA, based on the mesoscopic BEAM CORE model of the entire San Francisco Bay area. Both scenarios demonstrate that the hybrid approach results in significant computational gains with respect to a pure microscopic approach.

Dynamic event-triggering-based distributed model predictive control of heterogeneous connected vehicle platoon under DoS attacks

This paper is concerned with the distributed model predictive control (DMPC) for heterogeneous connected vehicle platoon (CVP) under denial-of-service (DoS) attacks. Firstly, a dynamic event-triggering mechanism (DETM) based on the information interaction between vehicles is proposed to reduce the communication and computational burdens. Due to the fact that the triggering moment for each vehicle cannot be synchronized and DoS attacks can break the communication between vehicles, a packet replenishment mechanism is designed to ensure the integrity and effectiveness of information interaction. Then, the effect of external disturbance is handled by adding robustness constraints to the DMPC algorithm. In addition, the recursive feasibility of the DMPC algorithm and input-to-state practical stability (ISPS) of the CVP control system are demonstrated. Finally, the effectiveness of the algorithm is verified by simulation and comparison results.

Extended GWMA control charts: A critical evaluation

AbstractWe demonstrate that the recently introduced triple generally weighted moving average (GWMA) chart and its counterpart, the double GWMA chart, incorporate sub‐optimal weighting patterns that may assign more weight to certain historical data points at the expense of more recent ones. Moreover, these control charts, when compared to the exponentially weighted moving average (EWMA) chart, exhibit a substantial computational burden. Our findings underscore that a well‐designed EWMA chart offers superior overall performance in comparison to these control charts.

Novel Frequency Regulation Scenarios Generation Method Serving for Battery Energy Storage System Participating in PJM Market

As one of the largest frequency regulation markets, the Pennsylvania-New Jersey-Maryland Interconnection (PJM) market allows extensive access of Battery Energy Storage Systems (BESSs). The designed signal regulation D (RegD) is friendly for use with BESSs with a fast ramp rate but limited energy. Designing operating strategies and optimizing the sizing of BESSs in this market are significantly influenced by the regulation signal. To represent the inherent randomness of the RegD signal and reduce the computational burden, typical frequency regulation scenarios with lower resolution are often generated. However, due to the rapid changes and energy neutrality of the RegD signal, generating accurate and representative scenarios presents challenges for the methods based on shape similarity. This paper proposes a novel probability-based method for generating typical regulation scenarios. The method relies on the joint probability distribution of two features with a 15-min resolution, extracted from the RegD signal with a 2 s resolution. The two features can effectively portray the characteristic of RegD signal and its influence on BESS operation. Multiple regulation scenarios are generated based on the joint probability distributions of these features at first, with the final typical scenarios chosen based on their probability distribution similarity to the actual distribution. Utilizing regulation data from the PJM market in 2020, this paper validates and analyzes the performance of the generated typical scenarios in comparison to existing methods, specifically K-means clustering and the forward scenarios reduction method.

Quantum‐Noise‐Driven Generative Diffusion Models

AbstractGenerative models realized with Machine Learning (ML) techniques are powerful tools to infer complex and unknown data distributions from a finite number of training samples in order to produce new synthetic data. Diffusion Models (DMs) are an emerging framework that have recently overcome Generative Adversarial Networks (GANs) in creating high‐quality images. Here, is proposed and discussed the quantum generalization of DMs, i.e., three Quantum‐Noise‐Driven Generative Diffusion Models (QNDGDMs) that could be experimentally tested on real quantum systems. The idea is to harness unique quantum features, in particular the non‐trivial interplay among coherence, entanglement, and noise that the currently available noisy quantum processors do unavoidably suffer from, in order to overcome the main computational burdens of classical diffusion models during inference. Hence, the suggestion is to exploit quantum noise not as an issue to be detected and solved but instead as a beneficial key ingredient to generate complex probability distributions from which a quantum processor might sample more efficiently than a classical one. Three examples of the numerical simulations are also included for the proposed approaches. The results are expected to pave the way for new quantum‐inspired or quantum‐based generative diffusion algorithms addressing tasks as data generation with widespread real‐world applications.

UET4Rec: U-net encapsulated transformer for sequential recommender

Recommending a tempting sequence of items according to a user’s previous history of purchases and clicks, for instance, in the online shopping portals is challenging. And yet it is a crucial task for all service providers. One of the core components in the recommender systems is a sequential model with which an input sequence is transformed into the predicted items. Among many, deep neural networks, such as RNN, LSTM, and Transformer, have been favored for this purpose. However, improving the performance of these models remains an important task. To address this, we propose a novel sequential model by combining a U-net and Transformer, called U-net Encapsulated Transformer. This hybrid architecture places a transformer model between a convolutional encoder and its decoder, wherein each convolutional layer processes 1D signal, i.e. text. The primary benefit of this structure is that the computational burden is reduced since the embedding size of the input to the Transformer is drastically decreased as the signal has to go through the multi-layer convolutional encoder. This solution leverages recommendation lists for action predictions and uses user feedback as direct rewards for updating the model. In addition, the loss function mechanism is improved by including contrastive and reinforcement learning losses. Evaluation of the proposed model, including extensive ablation study, is carried out on four standard benchmark datasets, such as RC15, RetailRocket, MovieLens-1M, and Amazon-Beauty, demonstrating superior performance compared to state-of-the-art methods.

A two-stage distributed stochastic planning method for source-grid-load-storage flexibility resources considering flexible ramp capacity

With rapidly increasing levels of renewable energy penetration, flexibility resources play an ever more critical role in the future power system. This paper describes a two-stage stochastic mixed integer second order cone (MISOC) method for flexibility resource planning (VFRP) considering source-grid-load-storage interactions. Considering renewable energy penetration levels and flexibility resource characteristics, this study quantifies the extent to which the share of renewable energy generation affects cost performance, and assesses the priority and potential of investing in flexibility resources. Moreover, the multi-type flexibility resource portfolio allocation method and energy storage only approach are compared in terms of socio-economic benefits. A multi-cut Benders-based approach is used for decomposing large-scale stochastic expansion planning problems to reduce the computational burden. Numerical simulation tests on the modified energy test systems I39G20 and F213G20 verified the correctness and effectiveness of the proposed flexible resource planning models and solution algorithms. The simulation results show that retrofit of coal-fired units for better flexibility has a higher investment return. Still, energy storage presents a tremendous investment prospect and potential. Finally, the multi-type flexibility resource portfolio allocation method is more cost-effective than an energy storage only approach.

Real-Time Multimodal 3D Object Detection with Transformers

The accuracy and real-time performance of 3D object detection are key factors limiting its widespread application. While cameras capture detailed color and texture features, they lack depth information compared to LiDAR. Multimodal detection combining both can improve results but incurs significant computational overhead, affecting real-time performance. To address these challenges, this paper presents a real-time multimodal fusion model called Fast Transfusion that combines the benefits of LiDAR and camera sensors and reduces the computational burden of their fusion. Specifically, our Fast Transfusion method uses QConv (Quick Convolution) to replace the convolutional backbones compared to other models. QConv concentrates the convolution operations at the feature map center, where the most information resides, to expedite inference. It also utilizes deformable convolution to better match the actual shapes of detected objects, enhancing accuracy. And the model incorporates EH Decoder (Efficient and Hybrid Decoder) which decouples multiscale fusion into intra-scale interaction and cross-scale fusion, efficiently decoding and integrating features extracted from multimodal data. Furthermore, our proposed semi-dynamic query selection refines the initialization of object queries. On the KITTI 3D object detection dataset, our proposed approach reduced the inference time by 36 ms and improved 3D AP by 1.81% compared to state-of-the-art methods.

A Priori Estimation of Radar Satellite Interferometry’s Sensitivity for Landslide Monitoring in the Italian Emilia-Romagna Region

The InSAR technique is known to be a powerful tool for precise monitoring of wide areas in terms of displacements. It is conceivable to also use this technique to monitor landslide areas, but geometrical distortions due to ground morphology and land cover could make InSAR processing ineffective for such applications. Because of the computational burden of InSAR processing, it is important to have preliminary knowledge about the possible suitability of the technique for the inspected area before acquiring and processing the data. This paper aims to perform a preliminary analysis of the InSAR sensitivity for the specific case of landslide monitoring. A new approach is proposed considering aspects specific to landslide displacements, which are basically tangent to the slope direction. Pre-processed coherence maps were used to account for the impact of land cover. The whole analysis can be carried out without acquiring cumbersome SAR datasets and can be used as a preliminary step. The Italian Emilia-Romagna region has been considered as the study area, with landslide areas accounting for more than 12% of its territory. The outcomes show that the inspected area has favourable morphological conditions, mainly thanks to its mild slopes and the limited number of landslides facing north, but the land cover has a strong negative impact on the InSAR sensitivity. Nevertheless, 7.5% of the landslide areas have promising conditions for monitoring using radar interferometry.

RNDDNet: A residual nested dilated DenseNet based deep-learning model for chilli plant disease classification

The most significant peril to food safety arises from plant diseases, capable of substantially diminishing both the quantity and quality of agricultural yields. Identifying these plant diseases stands out as the foremost challenge within the agricultural sector. Convolutional and deep neural networks prove effective in resolving image classification challenges within the realm of computer vision. Numerous Deep Neural Network(DNN)-based structures have been employed to diagnose plant diseases. Many DNN models in the field make use of various iterations of Dense and DenseNet layers in order to enhance the receptive field and capture intricate features within the data. However, it is important to note that such models often come with a significant computational burden and can introduce aliasing artifacts due to their complexity and resource-intensive nature. To overcome those limitations, we proposed a novel Residual Nested Dilated DenseNet based deep-learning (RNDDNet) model in this paper. Residual Nested Dilated DenseNet model residual connections are achieving the required receptive field, and their dilation factors are effective in extracting more features. The RNDDNet model exhibits the highest level of accuracy in identifying plant diseases. This research introduces a less computational cost and compact model to detect diseases in plant leaves. The proposed model functions to identify diseases, utilizing a dataset comprising 3,800 photographs of chilli leaves, categorized into six distinct classes: five disorder classes and one healthy chilli class. Through experimentation, the outcomes indicate that the suggested model achieves an accuracy of 98.09 %, along with a precision of 97 %, a recall of 97.25 %, and an F1 score of 97.25%. The presented approach demonstrates its superiority over existing methodologies.

Compact and Low-Latency FPGA-Based Number Theoretic Transform Architecture for CRYSTALS Kyber Postquantum Cryptography Scheme

In the modern era of the Internet of Things (IoT), especially with the rapid development of quantum computers, the implementation of postquantum cryptography algorithms in numerous terminals allows them to defend against potential future quantum attack threats. Lattice-based cryptography can withstand quantum computing attacks, making it a viable substitute for the currently prevalent classical public-key cryptography technique. However, the algorithm’s significant time complexity places a substantial computational burden on the already resource-limited chip in the IoT terminal. In lattice-based cryptography algorithms, the polynomial multiplication on the finite field is well known as the most time-consuming process. Therefore, investigations into efficient methods for calculating polynomial multiplication are essential for adopting these quantum-resistant lattice-based algorithms on a low-profile IoT terminal. Number theoretic transform (NTT), a variant of fast Fourier transform (FFT), is a technique widely employed to accelerate polynomial multiplication on the finite field to achieve a subquadratic time complexity. This study presents an efficient FPGA-based implementation of number theoretic transform for the CRYSTAL Kyber, a lattice-based public-key cryptography algorithm. Our hybrid design, which supports both forward and inverse NTT, is able run at high frequencies up to 417 MHz on a low-profile Artix7-XC7A100T and achieve a low latency of 1.10μs while achieving state-of-the-art hardware efficiency, consuming only 541-LUTs, 680 FFs, and four 18 Kb BRAMs. This is made possible thanks to the newly proposed multilevel pipeline butterfly unit architecture in combination with employing an effective coefficient accessing pattern.

The United Kingdom electricity market mechanism: A tool for a battery energy storage system optimal dispatching

In this paper, the role of a Battery Energy Storage System (BESS) in the United Kingdom (UK) electricity market is investigated. Such device is selected since research works related to the topic demonstrate that BESSs help facing challenges caused by renewable energy sources increasing penetration in the power systems’ field. Indeed, in the next future, BESSs will assume a relevant role thanks to their fast response time. In the present paper, an optimization model is implemented to encode the UK market rules and output the optimal set of offers on day-ahead and intraday markets, dynamic frequency response services and imbalance settlement. The market rules are formalised in a mathematical model considering market temporal sequencing and BESS technical limitations. The tool also models the state of energy management, necessary since BESSs are energy limited assets. Objective function and constraints are linearized to define a Mixed Integer Linear Programming problem to reduce the computational burden and to integrate the model in the Energy Management System previously developed at University of Genoa. Specific tests are performed including and excluding ancillary services to evaluate the related advantages. The obtained outcomes reveal that ancillary services are the most convenient options within all the implemented scenarios.