Search

Abstract Significant structural differences in DNN-based object detectors hinders the transferability of adversarial attacks. Studies show that intermediate features extracted by the detector contain more model-independent information, and disrupting these features can enhance attack transferability across different detectors. However, the challenge lies in selecting crucial features that impact detection from redundant intermediate features. To address this issue, we introduce the Deep information bottleneck universal adversarial perturbation (DIB-UAP). DIB-UAP utilizes the deep information bottleneck to establish a link between intermediate features and model output, extracting crucial intermediate features and disrupting them to generate UAP with strong attack transferability. Additionally, we propose a data augmentation method, Scale & Tile, which effectively enhances the attack performance of UAP on medium and large-scale objects. Testing on two benchmark datasets with eight comparative methods across four black-box mainstream detectors has confirmed the attack transferability of DIB-UAP. Furthermore, practical utility validation of DIB-UAP has been conducted on a commercial object detection platform.

Abstract It has been difficult to achieve a suitable balance between effectiveness and efficiency in lightweight semantic segmentation networks in recent years. The goal of this work is to present an efficient and reliable semantic segmentation method called EBUNet, which is aimed at achieving a favorable trade-off between inference speed and prediction accuracy. Initially, we develop an Efficient Bottleneck Unit (EBU) that employs depth-wise convolution and depth-wise dilated convolution to obtain adequate features with moderate computation costs. Then, we developed a novel Image Partition Attention Module (IPAM), which divides feature maps into subregions and generates attention weights based on them. As a third step, we developed a novel lightweight attention decoder with which to retrieve spatial information effectively. Extensive experiments show that our EBUNet achieves 73.4% mIou and 152 FPS on the Cityscapes dataset and 72.2% mIoU and 147 FPS on the Camvid dataset with only 1.57 M parameters. The results of the experiment confirm that the proposed model is capable of making a decent trade-off in terms of accuracy, inference, and model size. The source code of our EBUNet is available at ( https://github.com/Skybird1101/EBUNet ).

Abstract Industrial defect detection is a hot topic in the field of computer vision. It is a challenging task due to complex features and many categories of industrial defects. In this paper, a deep learning model based on the multiscale feature extraction module is introduced for steel surface defect detection. The main focus on the feature extraction capability of the model and feature fusion capability to improve the accuracy of the model for steel surface defect detection. First, to improve the feature extraction ability of the model, a multiscale feature extraction (MSFE) module is introduced. The MSFE module can effectively extract multiscale features through three branches that have different convolution kernel sizes. Second, an efficient feature fusion (EFF) module is proposed to optimize feature fusion by adding features from the backbone network to the neck network. Third, this paper puts forward a new Bottleneck module by reducing the normalization layer and activation function in the original Bottleneck module. Finally, the backbone network is deepened to further enhance the feature extraction ability of the model. Extensive experiments are conducted on the public NEU-DET dataset. The experimental results validate the effectiveness of the designed modules and the proposed model. Compared with other state-of-the-art methods, the proposed model achieves optimal accuracy(73.08% mAP@0.5) while maintaining a small number of parameters.

Abstract The development of efficient, low-cost water splitting electrocatalysts is needed to store energy by generating sustainable hydrogen from low power clean but intermittent energy sources such as solar and wind. Here, we report a highly sustained low overpotential for oxygen evolution reached by the unique combination of three metals (NiCoV) prepared from a simple low temperature auto-combustion process. The amorphous multimetal oxygen evolving catalyst could be stably coated on a stainless-steel support using a tribochemical particle blasting method to create an oxygen evolution reaction (OER) electrode with a low overpotential of 230 mV at 10 mA cm−2 and a low Tafel slope of 40 mV dec−1. In addition to their low overpotential, this oxygen evolving electrocatalyst preserved performance demonstrating a stability after 10 h at the technologically relevant current density and without any surface morphology alteration. Given the importance of sustainable hydrogen production, the development of this new OER catalyst points the way to removing a key technical bottleneck for the water splitting reaction and could offer a route to cost reduction and lowering hurdles to more widespread adaptation of electrolyser technologies for hydrogen production.

Abstract The sparsity problem remains a significant bottleneck for recommendation systems. In recent years, deep matrix factorization has shown promising results in mitigating this issue. Furthermore, many works have improved the prediction accuracy of deep matrix factorization by incorporating the user’s and/or items’ auxiliary information. However, there are still two remaining drawbacks that need to be addressed. First, the initialization of latent feature representations has a substantial impact on the performance of deep matrix factorization, and most current models utilize a uniform approach to this initialization, constraining the model’s optimization potential. Secondly, many existing recommendation models lack versatility and efficiency in transferring auxiliary information from users or items to expand the feature space. This paper proposes a novel model to address the issues mentioned above. By using a semi-autoencoder, the pre-trained initialization of the latent feature representation is realized in this paper. Simultaneously, this model assimilates auxiliary information, like item attributes or rating matrices from diverse domains, to generate their latent feature representations. These representations are then transferred to the target task through subspace projection distance. With this, this model can utilize auxiliary information from various sources more efficiently and this model has better versatility. This is called deep matrix factorization via feature subspace transfer. Numerical experiments on several real-world data show the improvement of this method compared with state-of-the-art methods of introducing auxiliary information about items. Compared with the deep matrix factorization model, the proposed model can achieve 6.5% improvement at most in the mean absolute error and root mean square error.

Abstract Hand gesture segmentation is an essential step to recognize hand gestures for human–robot interaction. However, complex backgrounds and the variety of gesture shapes cause low semantic segmentation accuracy in the existing lightweight methods because of imprecise features and imbalance between branches. To remedy the above problems, we propose a new segmentation structure for hand gestures. Based on the structure, a novel tri-branch lightweight segmentation network (BLSNet), is proposed for gesture segmentation. Corresponding to the structure parts, three branches are employed to achieve local features, boundaries and semantic hand features. In the boundary branch, to extract multiscale features of hand gesture contours, a novel multi-scale depth-wise strip convolution (MDSC) module is proposed based on gesture boundaries for directionality. For hand boundary details, we propose a new boundary weight (BW) module based on boundary attention. To identify hand location, a semantic branch with continuous downsampling is used to address complex backgrounds. We use the Ghost bottleneck as the building block for the entire BLSNet network. To verify the effectiveness of the proposed network, corresponding experiments have been conducted based on OUHANDS and HGR1 datasets, and the experimental results demonstrate that the proposed method is superior to contrast methods.

Abstract The Zig-zag eel (Mastacembelus armatus) is an economically important species in southern China. Its natural resources have declined year by year due to overfishing. Understanding its genetic diversity and population structure is very important for resource conservations. Here, we first successfully developed 28 polymorphic microsatellite markers for zig-zag eels and ten of them were used to examine the genetic diversity and differentiation of 7 populations collected from the major river systems of south China. In total, 224 alleles were found with the 10 microsatellite loci in 7 populations, ranging from 4.6 (Nandujiang: NDJ) to 11.1 (Xijiang, XJ), with an average of 8.871 alleles. The average observed and expected heterozygosity ranged from 0.550 (NDJ) to 0.964 (Yuangjiang, YJ) and from 0.537 (NDJ) to 0.775 (Tanjiang, TJ), respectively. The average polymorphism-information content ranged from 0.472 (NDJ) to 0.757 (TJ). Hardy–Weinberg equilibrium test results revealed the loci showed differing deviation in different populations. In total, low level of genetic diversity was only found in HJ (Hanjiang) and NDJ populations. Besides, evidence of recent bottleneck was found in the HJ populations. Analysis of molecular variation showed that the percent variation within individuals (75.00%) was higher than that among populations (25%). In addition, population structure and the pairwise FST revealed that there was low differentiation among XJ, TJ and YJ populations. These data provide important genetic resources for understanding the population differentiation and facilitating genetic conservation and utilization of this species.

Abstract Federated Learning (FL) is a recent Machine Learning method for training with private data locally stored in distributed machines without gathering them into one place for central learning. Because FL depends on a central server for repeated aggregation of local training models, this server is prone to become a performance bottleneck. Therefore, one can combine FL with Edge Computing: introduce a layer of edge servers to each serve as a regional aggregator to offload the main server. The scalability is thus improved, however at the cost of learning accuracy. We show that this cost can be alleviated with a proper choice of edge server assignment: which edge servers should aggregate the training models from which local machines. In this paper, we propose an assignment solution for this purpose. Our solution is especially useful for the case of non-IID training data which is well-known to hinder today’s FL performance. Our findings are substantiated with an evaluation study using real-world datasets.

Abstract Heart failure and heart disease prediction in real-time is a highly significant necessity for the patients living under the observation of Internet of Things-based Ambient Assisted Living systems because cardiovascular diseases are the most common fatal chronic diseases. Most of the solutions regarding heart disease prediction in the Internet of Things-based medical systems are relying on server-based predictive analysis which can appear to be complex for generating real-time prediction notifications and unreliable in case of any network interruption occurrences. The suggested edge-based solution for the prediction of heart disease from collected sensor data in real-time using a proposed lightweight deep learning technique called Oversampled Quinary Feed Forward Network (OQFFN) provides a less complex framework and more reliable notification system in case of network failure for the disease prediction which also reduces the need of forwarding all the data to the server resulting in reduced network bottleneck.

Abstract Various new treatments are emerging constantly in anti-tumor therapies, including chemotherapy, immunotherapy, and targeted therapy. However, the efficacy is still not satisfactory. Astragalus polysaccharide is an important bioactive component derived from the dry root of Radix astragali. Studies found that astragalus polysaccharides have gained great significance in increasing the sensitivity of anti-tumor treatment, reducing the side effects of anti-tumor treatment, reversing the drug resistance of anti-tumor drugs, etc. In this review, we focused on the role of astragalus polysaccharides in tumor immune microenvironment. We reviewed the immunomodulatory effect of astragalus polysaccharides on macrophages, dendritic cells, natural killer cells, T lymphocytes, and B lymphocytes. We found that astragalus polysaccharides can promote the activities of macrophages, dendritic cells, natural killer cells, T lymphocytes, and B lymphocytes and induce the expression of a variety of cytokines and chemokines. Furthermore, we summarized the clinical applications of astragalus polysaccharides in patients with digestive tract tumors. We summarized the effective mechanism of astragalus polysaccharides on digestive tract tumors, including apoptosis induction, proliferation inhibition, immunoactivity regulation, enhancement of the anticancer effect and chemosensitivity. Therefore, in view of the multiple functions of astragalus polysaccharides in tumor immune microenvironment and its clinical efficacy, the combination of astragalus polysaccharides with antitumor therapy such as immunotherapy may provide new sparks to the bottleneck of current treatment methods.

Abstract Federated learning (FL) draws attention in academia and industry due to its privacy-preserving capability in training machine learning models. However, there are still some critical security attacks and vulnerabilities, including gradients leakage and interference attacks. Meanwhile, communication is another bottleneck in basic FL schemes since large-scale FL parameter transmission leads to inefficient communication, latency, and slower learning processes. To overcome these shortcomings, different communication efficiency strategies and privacy-preserving cryptographic techniques have been proposed. However, a single method can only partially resist privacy attacks. This paper presents a practical, privacy-preserving scheme combining cryptographic techniques and communication networking solutions. We implement Kafka for message distribution, the Diffie–Hellman scheme for secure server aggregation, and gradient differential privacy for interference attack prevention. The proposed approach maintains training efficiency while being able to addressing gradients leakage problems and interference attacks. Meanwhile, the implementation of Kafka and Zookeeper provides asynchronous communication and anonymous authenticated computation with role-based access controls. Finally, we prove the privacy-preserving properties of the proposed solution via security analysis and empirically demonstrate its efficiency and practicality.

Abstract With the development of the Internet of Things (IoT), more and more devices are connected in edge-cloud environment, and spectrum scarcity has become a major bottleneck for the employment of IoT devices in the end side of cloud computing. In addition, due to the broadcast nature of RF link, the potential eavesdropper may be presented in the wireless environment. To improve spectrum efficiency (SE) and ensure safe transmission of the users, in this paper, the two-way relay cooperative Cognitive Radio Non-Orthogonal Multiple Access (CR-NOMA) with massive multiple-in multiple-out (MIMO) is proposed. Our objective is to maximize the security energy efficiency (SEE) of the cognitive system subject to the quality of service (QoS) of users. Specifically, the multi-objective optimization problem is decomposed into three subproblems, i.e., the optimization of transmit power, power allocation, and antenna selection, respectively. The Lagrange dual algorithm based on the first-order Taylor series expansion function is proposed to solve the non-convex problem. Moreover, a novel orthogonal antenna selection algorithm is proposed to decrease the cost of radio frequency chains and limit the interference to the primary user. The simulation results show that the proposed scheme has significant performance gain on SEE. It can effectively increase the number of access for cloud computing edge IoT users. In addition, due to the impact of correlation between the antennas, as the number of relay antennas increases, the SEE converges to a specific value, which exists the optimal number of antenna.

Abstract Blur detection is aimed to recognize the blurry pixels from a given image, which is increasingly valued in vision-centered applications. Albeit great improvement achieved by recent deep learning-based methods, the overweight model and rough boundary still pose challenges to blur detection. In this paper, we propose a Hierarchical Edge-guided Region-complemented Network (HER-Net) to tackle the above issues in quest of a favorable accuracy–complexity trade-off. First, we propose novel olive-shaped and pear-shaped inverted bottleneck structures based on large-kernel depth-wise convolutions to build a very concise architecture. Second, we provoke and exploit region-concerned and edge-concerned morphological priors to refine the boundary. To this end, we propose a reverse-region spatial attention to mine the complementary affinities between blurry and sharp regions so as to enrich the residual details around the boundary. In addition, we propose an edge spatial attention to guide the edge-concerned cues to emphasize the features related to the boundary. Both attentions are embedded into the model with hierarchical manners. Extensive experiments on three benchmark datasets demonstrate that the proposed method can achieve better detection performance using fewer parameters and lower floating-point operations compared to competitive methods. It proves the efficiency and effectiveness of our method in blur detection task.

Abstract Purpose Chordoma is a rare and aggressive bone cancer driven by the developmental transcription factor brachyury. Efforts to target brachyury are hampered by the absence of ligand-accessible small-molecule binding pockets. Genome editing with CRISPR systems provides an unprecedented opportunity to modulate undruggable transcription factor targets. However, delivery of CRISPR remains a bottleneck for in vivo therapy development. The aim was to investigate the in vivo therapeutic efficiency of Cas9/guide RNA (gRNA) ribonucleoprotein (RNP) delivery through a novel virus-like particle (VLP) by fusing an aptamer-binding protein to the lentiviral nucleocapsid protein. Methods The p24 based ELISA and transmission electron microscopy were used to determine the characterization of engineered VLP-packaged Cas9/gRNA RNP. The deletion efficiency of brachyury gene in chordoma cells and tissues was measured by genome cleavage detection assay. RT-PCR, Western blot, immunofluorescence staining, and IHC were employed to test the function of brachyury deletion. Cell growth and tumor volume were measured to evaluate the therapeutic efficiency of brachyury deletion by VLP-packaged Cas9/gRNA RNP. Results Our “all-in-one” VLP-based Cas9/gRNA RNP system allows for transient expression of Cas9 in chordoma cells, but maintains efficient editing capacity leading to approximately 85% knockdown of brachyury with subsequent inhibition of chordoma cell proliferation and tumor progression. In addition, this VLP-packaged brachyury-targeting Cas9 RNP avoids systemic toxicities in vivo. Conclusion Our preclinical studies demonstrate the potential of VLP-based Cas9/gRNA RNP gene therapy for the treatment of brachyury-dependent chordoma.

Abstract Federated learning (FL) enables clients learning a shared global model from multiple distributed devices while keeping training data locally. Due to the synchronous update mode between server and devices, the straggler problem has become a significant bottleneck for efficient FL. Existing approaches attempt to tackle this issue by using asynchronous-based model aggregation. However, these researches are only from the perspective of changing global model updating manner to mitigate straggler effect. They do not investigate the intrinsic reasons for the generation of the straggler effect, which could not fundamentally solve this problem. Furthermore, asynchronous-based approaches usually ignore those slow-responding but important local updates while frequently aggregating fast-responding ones during the whole training process, which may come with degradation in model accuracy. Thus, we propose FedTCR, a novel Federated learning approach via Taming Computing Resources. FedTCR includes a coarse-grained logical computing cluster construction algorithm (LCC) and a fine-grained intra-cluster collaborative training mechanism (ICT) as part of the FL process. The computing resource heterogeneity among devices and the communication frequency between devices and the server are indirectly tamed during this process, which substantially resolves the straggler problem and significantly improves the communication efficiency for FL. Experimental results show that FedTCR achieves much faster training performance, reducing the communication cost by up to $$8.59\,\times $$ 8.59 × while improving $$13.85\%$$ 13.85 % model accuracy, compared to state-of-the-art FL methods.

Abstract Autoencoder-based anomaly detection approaches can be used for precluding scope compliance failures of the automotive perception. However, the applicability of these approaches for the automotive domain should be thoroughly investigated. We study the capability of two autoencoder-based approaches using reconstruction errors and bottleneck-values for detecting semantic anomalies in automotive images. As a use-case, we consider a specific highway driving scenario identifying if there are any vehicles in the field of view of a front-looking camera. We conduct a series of experiments with two simulated driving scenario datasets and measure anomaly detection performance for different cases. We systematically test different autoencoders and training parameters, as well as the influence of image colors. We show that the autoencoder-based approaches demonstrate promising results for detecting semantic anomalies in highway driving scenario images in some cases. However, we also observe the variability of anomaly detection performance between different experiments. The autoencoder-based approaches are capable of detecting semantic anomalies in highway driving scenario images to some extent. However, further research with other use-cases and real datasets is needed before they can be safely applied in the automotive domain.

Abstract The computational bottleneck in distributed optimization methods, which is based on projected gradient descent, is due to the computation of a full gradient vector and projection step. This is a particular problem for large datasets. To reduce the computational complexity of existing methods, we combine the randomized block-coordinate descent and the Frank–Wolfe techniques, and then propose a distributed randomized block-coordinate projection-free algorithm over networks, where each agent randomly chooses a subset of the coordinates of its gradient vector and the projection step is eschewed in favor of a much simpler linear optimization step. Moreover, the convergence performance of the proposed algorithm is also theoretically analyzed. Specifically, we rigorously prove that the proposed algorithm can converge to optimal point at rate of $${\mathcal {O}}(1/t)$$ O ( 1 / t ) under convexity and $${\mathcal {O}}(1/t^2)$$ O ( 1 / t 2 ) under strong convexity, respectively. Here, t is the number of iterations. Furthermore, the proposed algorithm can converge to a stationary point, where the “Frank-Wolfe” gap is equal to zero, at a rate $${\mathcal {O}}(1/\sqrt{t})$$ O ( 1 / t ) under non-convexity. To evaluate the computational benefit of the proposed algorithm, we use the proposed algorithm to solve the multiclass classification problems by simulation experiments on two datasets, i.e., aloi and news20. The results shows that the proposed algorithm is faster than the existing distributed optimization algorithms due to its lower computation per iteration. Furthermore, the results also show that well-connected graphs or smaller graphs leads to faster convergence rate, which can confirm the theoretical results.

Abstract Air transport services require complementary ground and air infrastructure. Capacity constraints of an airport’s runway system create a significant bottleneck. The right of use depends on the availability of a slot. When the demand for slots exceeds the supply, an allocation process is required. In most countries of the world, there is no market mechanism for this. A grandfather clause restricts entry for new entrants at congested airports. A remedy can be found in a property rights perspective.

Abstract A total of 3.2 million people in Germany have been fully immunised so far. However, in order to achieve herd immunity, which is likely to require an immunisation rate of around 80 % due to the highly infectious mutants, approximately 65 million people need to be vaccinated. The current “imbalance” in the vaccination campaign is largely due to the limited supply of vaccines. Since J&J’s vaccine has recently been approved by the EU and additional production facilities are available, it is foreseeable that the capacities of the vaccination centres will no longer be sufficient. Moreover, at some point the existing unwillingness to vaccinate could become the final bottleneck in reaching herd immunity. This paper provides some calculations for these relationships. Against this background, it is questionable how well-prepared the present system is to cope with these foreseeable problems.

Ship detection is a canonical problem in computer vision. Motivated by the observation that the major bottleneck of ship detection lies on the different scales of ship instances in images, we focus on improving the detection rate, especially for the small-sized ships which are relatively far from the camera. We use the Smooth function combined with L1 and L2 norm to optimize the region proposal network (RPN) loss function and reduce the deviation between the prediction frame and the actual target to ensure the accurate location of the ship target. With the Two-Way sampling combined with the shared weight to generate the mask, we solve the problems of inaccurate segmentation, target loss and small interference when Mask Region Convolution Neural Network (RCNN) is used to segment an instance. We create the experimental data sets from the deep learning annotation tool—Labelme. Experiments show that the improved Mask-RCNN model has a confidence rate of 82.17%. Serving as the basic network, the test accuracy rate of ResNetXt-101 is 3.3% higher than that of the original ResNet-101, which can better realize the function of ship target location and mask generation.

As reflected by the TOP500 list, hypercubes are popular interconnection networks for massively parallel systems, the main reason being the simplicity and ease of implementation of this network topology. In order to retain performance high and avoid bottleneck situation, routing algorithms are critical for these high-performance systems. Furthermore, disjoint path routing is a very desirable property of such communication algorithms. Effectively, selecting mutually node-disjoint paths guarantees that notorious parallel processing issues such as deadlocks, livelocks and starvations shall never occur. In this paper, we describe a routing algorithm for hypercubes that, given a bit constraint, selects internally node-disjoint paths between any pair of nodes satisfying the constraint, and such that the selected paths all satisfy the constraint. The introduction of such bit constraint enables the selection of multiple sets of disjoint paths between several node pairs each satisfying a distinct bit constraint, which is impossible with conventional routing algorithms. Selecting simultaneously disjoint paths between different node pairs induces increased communication performance and system dependability. The correctness and complexities of the described algorithm are formally proved, and analysis of the algorithm performance in practice is conducted by empirical evaluation.

High utility pattern mining is an emerging research topic in the data mining field. Unlike frequent pattern mining, high utility pattern mining deals with non-binary databases, in which the information about purchased quantities of items is maintained. Due to the non-existence of anti-monotone property among the utilities of itemsets, utility mining becomes a big challenge. Moreover, discovering useful patterns from the huge number of potential patterns is a mining bottleneck. However, the compact (Closed and Maximal) high utility pattern mining moderately lessens the number of patterns, but it does not solve it. Recently, an efficient framework called GUIDE, was proposed in the literature to address this issue. Though, GUIDE effectively reduced the number of high utility patterns, yet the quality of few mined patterns and their utilities are not exact. In view of this, we propose a modified MGUIDE algorithm to improve the quality and determine exact utilities of maximal patterns.

We propose a Drawing-Frequency based Layered (DFL) Canvas approach to reduce a synchronization delay in collaborative web applications. The synchronization delay consists of a network delay and a drawing delay. The network delay is well-argued topic by various researches. We focus on the drawing delay, which is a primary bottleneck of Canvas objects synchronizations. We introduce the DFL Canvas approach in order to solve the drawing problem. The DFL Canvas divides Canvas objects by drawingfrequency. We implemented a collaborative paper editing support system with the DFL Canvas. Users can edit the same document and share Canvas objects on papers in real-time on web browsers. We present an evaluation result that indicates the approach is suited for collaborative paper editing support systems.

This paper addresses the workload balancing problem in identical parallel machines context. The problem consists of assigning different jobs to identical parallel machines in order to minimize the workload imbalance among the different machines. This problem is formulated as a linear mixed integer program to minimize the difference between the greatest and smallest workload assigned to each machine. Based on some numerical examples reported in the literature, we establish that the classical formulation which consists of minimizing the greatest machine completion time does not provide the optimal workload repartition. That is why we consider a new mathematical formulation based on the minimization of the difference between the workload of the bottleneck machine and the workload of the fastest machine. The proposed programming method is also used to provide optimal solutions in reasonable computational times for different test problems presented in the literature by Raghavendra and Murthy to test their genetic algorithm.

Fuzzy relations are simple mathematical structures that enable a very general representation of fuzzy knowledge, and fuzzy relational calculus offers a powerful machinery for approximate reasoning. However, one of the most relevant limitations of approximate reasoning is the efficiency bottleneck. In this paper, we present two implementations for fast fuzzy inference through relational composition, with the twofold objective of being general and efficient. The two implementations are capable of working on full and sparse representations respectively. Further, a wrapper procedure is capable of automatically selecting the best implementation on the basis of the input features. We implemented the code in GNU Octave because it is a high-level language targeted to numerical computations. Experimental results show the impressive performance gain when the proposed implementation is used.

All aquifers are heterogeneous to a certain degree. The spatial distribution of hydraulic conductivity K(x, y, z), or aquifer heterogeneity, significantly influences the groundwater flow movement and associated solute transport. Of particular importance in designing an in-situ remediation plan is a knowledge of low-K layers because they are less accessible to remedial agents and form a bottleneck in remediation. The characterization of aquifer heterogeneity is essential to the solution of many practical and scientific groundwater problems. This article reviews the field technique using the multilevel slug test (MLST), which determines a series of K estimates at depths of interest in a well by making use of a double-packer system. The K(z) obtained manifests the vertical variation of hydraulic conductivity in the vicinity of the test well, and the combination of K(z) from different wells gives rise to a three-dimensional description of K(x, y, z). The MLST response is rather sensitive to hydraulic conductivity variation; e.g., it is oscillatory for highly permeable conditions (K > 5 × 10-4 m s-1) and a nonoscillatory for K < 5 × 10-4 m s-1. In this article we discuss the instrumentation of the double-packer system, the implementation of the depth-specific slug test, the data analysis methods for a spectrum of response characteristics usually observed in the field, and field applications of the MLST.