Computational Tradeoffs Research Articles

Uncertainty in Fourier Transforms: A Fuzzy Logic Perspective

This study explores the integration of fuzzy logic with Fourier transforms to address the challenges of uncertainty, noise, and imprecision in real-world data. Fuzzy Fourier transforms extend traditional Fourier methods by incorporating fuzzy numbers, allowing for more robust frequency analysis, signal processing, and data reconstruction, particularly in noisy or incomplete datasets. The study examines the mathematical formulation of fuzzy Fourier transforms, computational trade-offs, and their performance advantages in comparison to classical Fourier methods. Real-world applications are discussed, including signal processing, image reconstruction, and time-series forecasting. Furthermore, the research highlights the increased computational complexity associated with fuzzy methods, the challenges of interpreting fuzzy results, and the limitations of handling probabilistic uncertainty. Future directions include the development of multi-dimensional fuzzy Fourier transforms, hybrid models integrating machine learning, and quantum computing applications. These advancements have broad implications for fields such as telecommunications, medical imaging, and financial forecasting, where handling uncertainty is critical.

Equipping computational pathology systems with artifact processing pipelines: a showcase for computation and performance trade-offs

BackgroundHistopathology is a gold standard for cancer diagnosis. It involves extracting tissue specimens from suspicious areas to prepare a glass slide for a microscopic examination. However, histological tissue processing procedures result in the introduction of artifacts, which are ultimately transferred to the digitized version of glass slides, known as whole slide images (WSIs). Artifacts are diagnostically irrelevant areas and may result in wrong predictions from deep learning (DL) algorithms. Therefore, detecting and excluding artifacts in the computational pathology (CPATH) system is essential for reliable automated diagnosis.MethodsIn this paper, we propose a mixture of experts (MoE) scheme for detecting five notable artifacts, including damaged tissue, blur, folded tissue, air bubbles, and histologically irrelevant blood from WSIs. First, we train independent binary DL models as experts to capture particular artifact morphology. Then, we ensemble their predictions using a fusion mechanism. We apply probabilistic thresholding over the final probability distribution to improve the sensitivity of the MoE. We developed four DL pipelines to evaluate computational and performance trade-offs. These include two MoEs and two multiclass models of state-of-the-art deep convolutional neural networks (DCNNs) and vision transformers (ViTs). These DL pipelines are quantitatively and qualitatively evaluated on external and out-of-distribution (OoD) data to assess generalizability and robustness for artifact detection application.ResultsWe extensively evaluated the proposed MoE and multiclass models. DCNNs-based MoE and ViTs-based MoE schemes outperformed simpler multiclass models and were tested on datasets from different hospitals and cancer types, where MoE using (MobileNet) DCNNs yielded the best results. The proposed MoE yields 86.15 % F1 and 97.93% sensitivity scores on unseen data, retaining less computational cost for inference than MoE using ViTs. This best performance of MoEs comes with relatively higher computational trade-offs than multiclass models. Furthermore, we apply post-processing to create an artifact segmentation mask, a potential artifact-free RoI map, a quality report, and an artifact-refined WSI for further computational analysis. During the qualitative evaluation, field experts assessed the predictive performance of MoEs over OoD WSIs. They rated artifact detection and artifact-free area preservation, where the highest agreement translated to a Cohen Kappa of 0.82, indicating substantial agreement for the overall diagnostic usability of the DCNN-based MoE scheme.ConclusionsThe proposed artifact detection pipeline will not only ensure reliable CPATH predictions but may also provide quality control. In this work, the best-performing pipeline for artifact detection is MoE with DCNNs. Our detailed experiments show that there is always a trade-off between performance and computational complexity, and no straightforward DL solution equally suits all types of data and applications. The code and HistoArtifacts dataset can be found online at Github and Zenodo, respectively.

Computational Methods for Fast Bayesian Model Assessment via Calibrated Posterior p-values

Posterior predictive p-values (ppps) have become popular tools for Bayesian model assessment, being general-purpose and easy to use. However, interpretation can be difficult because their distribution is not uniform under the hypothesis that the model did generate the data. Calibrated ppps (cppps) can be obtained via a bootstrap-like procedure, yet remain unavailable in practice due to high computational cost. This article introduces methods to enable efficient approximation of cppps and their uncertainty for fast model assessment. We first investigate the computational tradeoff between the number of calibration replicates and the number of MCMC samples per replicate. Provided that the MCMC chain from the real data has converged, using short MCMC chains per calibration replicate can save significant computation time compared to naive implementations, without significant loss in accuracy. We propose different variance estimators for the cppp approximation, which can be used to confirm quickly the lack of evidence against model misspecification. As variance estimation uses effective sample sizes of many short MCMC chains, we show these can be approximated well from the real-data MCMC chain. The procedure for cppp is implemented in NIMBLE, a flexible framework for hierarchical modeling that supports many models and discrepancy measures. Supplementary materials for this article are available online.

Optimising Marine Basic Spatial Units (MBSU) for Ocean Accounting using empirical data from Saleh Bay, Indonesia

Ocean Accounts, aligned with the UN System of Environmental Economic Accounting – Environmental Accounting (SEEA EA), bring together economic, social and environmental information in a coherent and standardised manner. Ecosystem extent is a structure to understand environmental assets and uses a basic spatial unit to facilitate the classification and measurement of ecosystems by type. This study tested the impact of grid size and method of designation per grid cell for Marine Basic Spatial Units (MBSU), using Saleh Bay, Indonesia as a case study. The extent of mangrove, seagrass and coral reefs were previously delineated in 2021 for ocean accounting activities. This study tested grids with two different cell sizes (10 x 10 m2 and 25 x 25 m2) and two different methods of designation, namely: (i) dominance (extent-based) and (ii) hierarchy (criteria-based) methods. The results indicated that a larger grid size is related to higher error in estimating both total area per ecosystem and spatial configuration within the study area. The dominance method produced more accurate results than the hierarchy method, although, when considering computational trade-offs, a larger grid size and the hierarchy method observed a much lower computational cost. These results demonstrate the need to carefully consider grid size and method when designating basic spatial units for accounting activities, as they impact linked accounting tables and, in turn, have implications when providing information for management and policy.

MsDC-DEQ-Net: Deep Equilibrium Model (DEQ) with Multiscale Dilated Convolution for Image Compressive Sensing (CS)

Compressive sensing (CS) is a technique that enables the recovery of sparse signals using fewer measurements than traditional sampling methods. To address the computational challenges of CS reconstruction, our objective is to develop an interpretable and concise neural network model for reconstructing natural images using CS. We achieve this by mapping one step of the iterative shrinkage thresholding algorithm (ISTA) to a deep network block, representing one iteration of ISTA. To enhance learning ability and incorporate structural diversity, we integrate aggregated residual transformations (ResNeXt) and squeeze-and-excitation mechanisms into the ISTA block. This block serves as a deep equilibrium layer connected to a semi-tensor product network for convenient sampling and providing an initial reconstruction. The resulting model, called MsDC-DEQ-Net, exhibits competitive performance compared to state-of-the-art network-based methods. It significantly reduces storage requirements compared to deep unrolling methods, using only one iteration block instead of multiple iterations. Unlike deep unrolling models, MsDC-DEQ-Net can be iteratively used, gradually improving reconstruction accuracy while considering computation tradeoffs. Additionally, the model benefits from multiscale dilated convolutions, further enhancing performance.

Fast Bayesian Record Linkage for Streaming Data Contexts

Record linkage is the task of combining records from multiple files which refer to overlapping sets of entities when there is no unique identifying field. In streaming record linkage, files arrive sequentially in time and estimates of links are updated after the arrival of each file. This problem arises in settings such as longitudinal surveys, electronic health records, and online events databases, among others. The challenge in streaming record linkage is to efficiently update parameter estimates as new data arrive. We approach the problem from a Bayesian perspective with estimates calculated from posterior samples of parameters and present methods for updating link estimates after the arrival of a new file that are faster than fitting a joint model with each new data file. In this article, we generalize a two-file Bayesian Fellegi-Sunter model to the multi-file case and propose two methods to perform streaming updates. We examine the effect of prior distribution on the resulting linkage accuracy as well as the computational tradeoffs between the methods when compared to a Gibbs sampler through simulated and real-world survey panel data. We achieve near-equivalent posterior inference at a small fraction of the compute time. Supplementary materials for this article are available online.

FeedFormer: Revisiting Transformer Decoder for Efficient Semantic Segmentation

With the success of Vision Transformer (ViT) in image classification, its variants have yielded great success in many downstream vision tasks. Among those, the semantic segmentation task has also benefited greatly from the advance of ViT variants. However, most studies of the transformer for semantic segmentation only focus on designing efficient transformer encoders, rarely giving attention to designing the decoder. Several studies make attempts in using the transformer decoder as the segmentation decoder with class-wise learnable query. Instead, we aim to directly use the encoder features as the queries. This paper proposes the Feature Enhancing Decoder transFormer (FeedFormer) that enhances structural information using the transformer decoder. Our goal is to decode the high-level encoder features using the lowest-level encoder feature. We do this by formulating high-level features as queries, and the lowest-level feature as the key and value. This enhances the high-level features by collecting the structural information from the lowest-level feature. Additionally, we use a simple reformation trick of pushing the encoder blocks to take the place of the existing self-attention module of the decoder to improve efficiency. We show the superiority of our decoder with various light-weight transformer-based decoders on popular semantic segmentation datasets. Despite the minute computation, our model has achieved state-of-the-art performance in the performance computation trade-off. Our model FeedFormer-B0 surpasses SegFormer-B0 with 1.8% higher mIoU and 7.1% less computation on ADE20K, and 1.7% higher mIoU and 14.4% less computation on Cityscapes, respectively. Code will be released at: https://github.com/jhshim1995/FeedFormer.

Reconstructing Kernel-Based Machine Learning Force Fields with Superlinear Convergence.

Kernel machines have sustained continuous progress in the field of quantum chemistry. In particular, they have proven to be successful in the low-data regime of force field reconstruction. This is because many equivariances and invariances due to physical symmetries can be incorporated into the kernel function to compensate for much larger data sets. So far, the scalability of kernel machines has however been hindered by its quadratic memory and cubical runtime complexity in the number of training points. While it is known that iterative Krylov subspace solvers can overcome these burdens, their convergence crucially relies on effective preconditioners, which are elusive in practice. Effective preconditioners need to partially presolve the learning problem in a computationally cheap and numerically robust manner. Here, we consider the broad class of Nyström-type methods to construct preconditioners based on successively more sophisticated low-rank approximations of the original kernel matrix, each of which provides a different set of computational trade-offs. All considered methods aim to identify a representative subset of inducing (kernel) columns to approximate the dominant kernel spectrum.

Geometric approach to robust stability analysis of Linear Parameter-Varying systems: Computational trade-offs between the exact and the simplex convex hulls

Robust stability is investigated for continuous-time, affine, Linear Parameter-Varying (LPV) models, with bounded variation rates of uncertainty. Towards this end, affine Parameter Dependent Lyapunov Functions (PDLF) are considered to certificate stability in the Lyapunov sense. In the literature, the search for PDLFs amounts to a non conservative Linear Matrix Inequality (LMI) problem, at expense of a factorial growth in its complexity. Remedies to overcome this complexity have been proposed recently, exploiting geometrical aspects of the problem, however, they can be conservative. This paper presents new stability tests that are a trade-off between the exact and the fastest solutions. We offer an analytical procedure to indicate when the proposed tests are prone to reduce conservativeness. Also, a simple procedure is introduced to possibly improve existing and new tests, without impacts on the computation effort. Numerical simulations illustrate the improvements of the proposed strategies.

What is redundant and what is not? Computational trade-offs in modelling to generate alternatives for energy infrastructure deployment

Given the urgent need to devise credible, deep strategies for carbon neutrality, approaches for ‘modelling to generate alternatives’ (MGA) are gaining popularity in the energy sector. Yet, MGA faces limitations when applied to state-of-the-art energy system models: the number of alternatives that can be generated is virtually infinite; no realistic computational effort can discover the complete technology and spatial option space. Here, based on our own SPORES method, a highly customisable and spatially-explicit advancement of MGA, we empirically test different search strategies – including some adapted from other MGA approaches – with the aim of identifying how to minimise redundant computation. With application to a model of the European power system, we show that, for a fixed number of generated alternatives, there is a clear trade-off in making use of the available computational power to unveil technology versus spatial dissimilarity across alternative system configurations. Moreover, we show that focussing on technology dissimilarity may fail to identify system configurations that appeal to real-world stakeholders, such as those in which capacity is more spread out at the local scale. Based on this evidence that no feasible alternative can be deemed redundant a priori, we propose to initially search for options in a way that balances spatial and technology dissimilarity; this can be achieved by combining the strengths of two different strategies. The resulting solution space can then be refined based on the feedback of stakeholders. More generally, we propose the adoption of ad-hoc MGA sensitivity analyses, targeted at testing a study’s central claims, as a computationally inexpensive standard to improve the quality of energy modelling analyses.

Jitter‐sensitive data communication in emerging wireless networks

AbstractJitter and time‐sensitive communications (TSC) will play an important role as part of the emerging wireless networks such as fifth‐generation (5G) and sixth‐generation (6G) cellular networks to enable applications including vehicle‐to‐everything (V2X), real‐time audio/video, real‐time industrial automation and control, and synchronized collaboration. TSC is characterized by stringent latency, jitter, and reliability requirements. This paper considers the problem of scheduling time‐sensitive data packets with stringent playout deadlines over a downlink multiple‐input multiple‐output (MIMO) channel, as part of the emerging wireless networks, to minimize the average jitter of a data session. Even though extensive research has been performed on MIMO technology, jitter and time‐sensitive data communication over a MIMO channel remains an under‐studied problem. In the proposed model, early or late arrival of the packets with respect to their playout deadlines incurs an early‐ or late‐arrival penalty, respectively. The proposed scheduling problem is proven to be NP‐hard and three computationally‐efficient scheduling heuristics are developed with varying performance and computational tradeoffs. These heuristics make packet scheduling decisions over the MIMO subchannels with the objective of minimizing the average jitter value of the data session. While making scheduling decisions, these heuristics intelligently take into consideration the playout deadlines and size of the packets, the time‐varying channel conditions, and the cost of the early or late arrival of the packets. The performance of these heuristics is evaluated from a design perspective through extensive simulations. The simulation results provide key insights into the relative performance tradeoffs among the three heuristics as a function of the various system parameters. This information could be utilized in intelligently allocating system resources to obtain the desired optimal system performance.

Carotid Intima-Media Thickness Segmentation using Attention Mechanism based Convolutional Neural Network with Domain-Specific Objective Function

CIMT (Carotid Intima-Media Thickness) has been proven to be both a significant and reliable marker for the evaluation of the risk of cardiovascular disease. Cardiovascular disease is the leading cause of mortality globally, and yet could be easily treated if detected in its early stages. A prime indicator of Cardiovascular disease, CIMT has previously been measured through manual examination of ultrasound videos for the gap between the Lumen-Intima and the Media-Adventitia interfaces, the two inner layers of the Carotid Artery. However, this method is not only inconvenient, but also time consuming. There has been a significant number of previous deep learning approaches to this issue, which have yielded substantial results. However, as this problem concerns the morality of patients directly, medical professionals have been hesitant to be dependent on these approaches, as the current accuracy of the state-of-the-art model still falls short to human observations. Furthermore, high performing models come at high computational costs. CIMT can actually be determined by a miniscule region of the Carotid Ultrasonic image, which many past researches have not taken into consideration. This paper proposes to use an attention mechanism to determine the region of interest and an encoder-decoder system which significantly reduces computational trade off while maintaining comparable accuracy. We also propose a novel connection loss to solve the disconnection problem in the prediction. The proposed model yields an unprecedented accuracy in terms of IoU and ACC of 0.78 and 0.99 respectively, substantially higher than previous state-of-the-art models by 18% and 8.8% on average.

Encoding time in neural dynamic regimes with distinct computational tradeoffs.

Converging evidence suggests the brain encodes time in dynamic patterns of neural activity, including neural sequences, ramping activity, and complex dynamics. Most temporal tasks, however, require more than just encoding time, and can have distinct computational requirements including the need to exhibit temporal scaling, generalize to novel contexts, or robustness to noise. It is not known how neural circuits can encode time and satisfy distinct computational requirements, nor is it known whether similar patterns of neural activity at the population level can exhibit dramatically different computational or generalization properties. To begin to answer these questions, we trained RNNs on two timing tasks based on behavioral studies. The tasks had different input structures but required producing identically timed output patterns. Using a novel framework we quantified whether RNNs encoded two intervals using either of three different timing strategies: scaling, absolute, or stimulus-specific dynamics. We found that similar neural dynamic patterns at the level of single intervals, could exhibit fundamentally different properties, including, generalization, the connectivity structure of the trained networks, and the contribution of excitatory and inhibitory neurons. Critically, depending on the task structure RNNs were better suited for generalization or robustness to noise. Further analysis revealed different connection patterns underlying the different regimes. Our results predict that apparently similar neural dynamic patterns at the population level (e.g., neural sequences) can exhibit fundamentally different computational properties in regards to their ability to generalize to novel stimuli and their robustness to noise—and that these differences are associated with differences in network connectivity and distinct contributions of excitatory and inhibitory neurons. We also predict that the task structure used in different experimental studies accounts for some of the experimentally observed variability in how networks encode time.

Tensor clustering with planted structures: Statistical optimality and computational limits

This paper studies the statistical and computational limits of high-order clustering with planted structures. We focus on two clustering models, constant high-order clustering (CHC) and rank-one higher-order clustering (ROHC), and study the methods and theory for testing whether a cluster exists (detection) and identifying the support of cluster (recovery). Specifically, we identify the sharp boundaries of signal-to-noise ratio for which CHC and ROHC detection/recovery are statistically possible. We also develop the tight computational thresholds: when the signal-to-noise ratio is below these thresholds, we prove that polynomial-time algorithms cannot solve these problems under the computational hardness conjectures of hypergraphic planted clique (HPC) detection and hypergraphic planted dense subgraph (HPDS) recovery. We also propose polynomial-time tensor algorithms that achieve reliable detection and recovery when the signal-to-noise ratio is above these thresholds. Both sparsity and tensor structures yield the computational barriers in high-order tensor clustering. The interplay between them results in significant differences between high-order tensor clustering and matrix clustering in literature in aspects of statistical and computational phase transition diagrams, algorithmic approaches, hardness conjecture, and proof techniques. To our best knowledge, we are the first to give a thorough characterization of the statistical and computational trade-off for such a double computational-barrier problem. Finally, we provide evidence for the computational hardness conjectures of HPC detection (via low-degree polynomial and Metropolis methods) and HPDS recovery (via low-degree polynomial method).

Core process representation in power system operational models: Gaps, challenges, and opportunities for multisector dynamics research

Power grid operations increasingly interact with environmental systems and human systems such as transportation, agriculture, the economy, and financial markets. Our objective is to discuss the modelling gaps and opportunities to advance the science for multisector adaptation and tradeoffs. We focus on power system operational models, which typically represent key physical and economic aspects of grid operations over days to a year and assume a fixed power grid infrastructure. Due to computational burden, models are typically customized to reflect regional resource opportunities, data availability, and applications of interest. We conceptualize power system operational models with four core processes: physical grid assets (generation, transmission, loads, and storage), model objectives and purpose, institutions and decision agents, and performance metrics. We taxonomize the representations of these core processes based on a review of 23 existing models. Using science questions around grid and short term uncertainties, long term global change, and multisectoral technological innovation as examples, we report on tradeoffs in process fidelity and tractability that have been adopted by the research community to represent multisectoral interactions in power system operational models. Our recommendations for research directions are model-agnostic, focusing on core processes, their interactions with other human systems, and consider computational tradeoffs.

Computational Tradeoffs of Search Methods for Minimum Constraint Removal Paths

The typical objective of path planning is to find the shortest feasible path. Many times, however, there may be no solution given the existence of constraints, such as obstacles. In these cases, the minimum constraint removal problem asks for the minimum set of constraints that need to be removed from the state space to find a solution. Unfortunately, minimum constraint removal paths do not exhibit dynamic programming properties, i.e., subsets of optimum solutions are not necessarily optimal. Thus, searching for such solutions is computationally expensive. This leads to approximate methods, which balance the cost of computing a solution and its quality. This work investigates alternatives in this context and evaluates their performance in terms of such tradeoffs. Solutions that follow a bounded-length approach, i.e., searching for paths up to a certain length, seem to provide a good balance between minimizing constraints, computational cost and path length.

Distributed optimization for degenerate loss functions arising from over-parameterization

We consider distributed optimization with degenerate loss functions, where the optimal sets of local loss functions have a non-empty intersection. This regime often arises in optimizing large-scale multi-agent AI systems (e.g., deep learning systems), where the number of trainable weights far exceeds the number of training samples, leading to highly degenerate loss surfaces. Under appropriate conditions, we prove that distributed gradient descent in this case converges even when communication is arbitrarily less frequent, which is not the case for non-degenerate loss functions. Moreover, we quantitatively analyze the convergence rate, as well as the communication and computation trade-off, providing insights into designing efficient distributed optimization algorithms. Our theoretical findings are confirmed by both distributed convex optimization and deep learning experiments.

Evaluating the Clinical Realism of Synthetic Chest X-Rays Generated Using Progressively Growing GANs.

Chest X-rays are a vital diagnostic tool in the workup of many patients. Similar to most medical imaging modalities, they are profoundly multi-modal and are capable of visualising a variety of combinations of conditions. There is an ever pressing need for greater quantities of labelled images to drive forward the development of diagnostic tools; however, this is in direct opposition to concerns regarding patient confidentiality which constrains access through permission requests and ethics approvals. Previous work has sought to address these concerns by creating class-specific generative adversarial networks (GANs) that synthesise images to augment training data. These approaches cannot be scaled as they introduce computational trade offs between model size and class number which places fixed limits on the quality that such generates can achieve. We address this concern by introducing latent class optimisation which enables efficient, multi-modal sampling from a GAN and with which we synthesise a large archive of labelled generates. We apply a Progressive Growing GAN (PGGAN) to the task of unsupervised X-ray synthesis and have radiologists evaluate the clinical realism of the resultant samples. We provide an in depth review of the properties of varying pathologies seen on generates as well as an overview of the extent of disease diversity captured by the model. We validate the application of the Fréchet Inception Distance (FID) to measure the quality of X-ray generates and find that they are similar to other high-resolution tasks. We quantify X-ray clinical realism by asking radiologists to distinguish between real and fake scans and find that generates are more likely to be classed as real than by chance, but there is still progress required to achieve true realism. We confirm these findings by evaluating synthetic classification model performance on real scans. We conclude by discussing the limitations of PGGAN generates and how to achieve controllable, realistic generates going forward. We release our source code, model weights, and an archive of labelled generates.

Strong Explanations in Abstract Argumentation

Abstract argumentation constitutes both a major research strand and a key approach that provides the core reasoning engine for a multitude of formalisms in computational argumentation in AI. Reasoning in abstract argumentation is carried out by viewing arguments and their relationships as abstract entities, with argumentation frameworks (AFs) being the most commonly used abstract formalism. Argumentation semantics then drive the reasoning by specifying formal criteria on which sets of arguments, called extensions, can be deemed as jointly acceptable. Such extensions provide a basic way of explaining argumentative acceptance. Inspired by recent research, we present a more general class of explanations: in this paper we propose and study so-called strong explanations for explaining argumentative acceptance in AFs. A strong explanation is a set of arguments such that a target set of arguments is acceptable in each subframework containing the explaining set. We formally show that strong explanations form a larger class than extensions, in particular giving the possibility of having smaller explanations. Moreover, assuming basic properties, we show that any explanation strategy, broadly construed, is a strong explanation. We show that the increase in variety of strong explanations comes with a computational trade-off: we provide an in-depth analysis of the associated complexity, showing a jump in the polynomial hierarchy compared to extensions.

Collaborative Social Internet of Things in Mobile Edge Networks

Artificial intelligence (AI) on chips has recently driven the expansion of the Social Internet of Things (SIoT), where a group of SIoT devices with social relations can collaboratively identify and handle local events without the help of remote servers. On the other hand, mobile-edge computing (MEC) is a favorable way to locally process SIoT data for reducing data transmission and computation among SIoT devices and backhaul networks. Nevertheless, the load sharing among SIoT devices, MEC, and remote servers brings about new challenges for the communication and computation tradeoff, cross-layer design in SIoT, and forwarding and aggregation tradeoff. To tackle these issues, we formulate a new optimization problem, namely, SIoT collaborative group and device selection problem (SCGDSP), and prove the NP-hardness. We first explore the intrinsic properties of a fundamental SCGDSP case by finding the optimal collaborative group for each user request. Then, we design an approximation algorithm for the general SCGDSP that first evaluates candidate collaborative groups under different social relations, and then selects the collaborative groups and SIoT devices properly. For scalability, the proposed algorithm also supports dynamic user requests and can be distributionally deployed in massive networks enabling collaborative MEC. Moreover, it also sustains local SIoT services, where the computation only involves SIoT devices and MEC servers. Simulation results demonstrate that effective SIoT and collaborative group selection (ESCGS) can reduce by more than 50% of the total communication and computation costs compared with baseline schemes in the real networks from topology zoo. Moreover, the distributed ESCGS reduces by 87% of the running time with total 16.5-MB message overhead, requiring no more than 0.05-ms transmission delay in a 100-Gb/s backbone network with eight MEC servers, 1000 SIoTs, and 800 monitored locations.