Approximate Search Research Articles

Efficient Approximate Nearest Neighbor Search in Multi-dimensional Databases

Approximate nearest neighbor (ANN) search is a fundamental search in multi-dimensional databases, which has numerous real-world applications, such as image retrieval, recommendation, entity resolution, and sequence matching. Proximity graph (PG) has been the state-of-the-art index for ANN search. However, the search on existing PGs either suffers from a high time complexity or has no performance guarantee on the search result. In this paper, we propose a novel τ-monotonic graph (τ- MG) to address the limitations. The novelty of τ-MG lies in a τ-monotonic property. Based on this property, we prove that if the distance between a query q and its nearest neighbor is less than a constant τ, the search on τ-MG guarantees to find the exact nearest neighbor of q and the time complexity of the search is smaller than all existing PG-based methods. For index construction efficiency, we propose an approximate variant of τ-MG, namely τ-monotonic neighborhood graph (τ- MNG), which only requires the neighborhood of each node to be τ-monotonic. We further propose an optimization to reduce the number of distance computations in search. Our extensive experiments show that our techniques outperform all existing methods on well-known real-world datasets.

A novel cell partition method by introducing Silhouette Coefficient for fast approximate nearest neighbor search

In this paper, a novel cell partition method is proposed to accelerate the approximate nearest neighbor (ANN) search. The traditional cell-level elimination of database vectors, such as the inverted file system (IVF), has limitations on accelerating the search process. Therefore, the Silhouette Coefficient is introduced to realize more accurate and effective vector-level pruning. The value of the Silhouette Coefficient is used as a novel criterion for dividing the Voronoi cell into the center region and the border region. In addition, a dynamic threshold selection model is introduced to determine the sampling rate for each cell adaptively. The proposed method can greatly improve the search speed by reducing the vector-level distance calculation. Besides, the value of the Silhouette Coefficient is saved by offline generating a rank in each Voronoi cell, which does not bring any extra memory cost. Through experiments, it can be seen that for two widely-used datasets of SIFT1M and GIST1M, with no more than 1% accuracy reduction, the search speed can be significantly increased by about 44% compared with the standard inverted file system. And compared with state-of-the-art ANN search methods that aim at acceleration, the proposed method can also achieve higher search accuracy with almost the same computational consumption.

Accelerating massive queries of approximate nearest neighbor search on high-dimensional data

Accelerating massive queries of approximate nearest neighbor search on high-dimensional data

On approximate near-neighbors search under the (continuous) Fréchet distance in higher dimensions

Previous studies on Approximate Near-Neighbors Search (ANNS) among curves are either focused on curves in R1 or under the discrete Fréchet distance. In this paper, we propose the first data structure for curves under the (continuous) Fréchet distance in higher dimensions. Given a set P of n curves each with number of vertices at most m in Rd, and a fixed δ>0, we aim to preprocess P into a data structure so that for any given query curve Q with k vertices, we can efficiently report all curves in P whose Fréchet distances to Q are at most δ. In the case that k is given in the preprocessing stage, for any ε>0 we propose a deterministic data structure whose space is n⋅O(max⁡{(dε)kd,(Ddε2)kd}) that can answer (1+ε)δ-ANNS queries in O(kd) query time, where D is the diameter of P. Considering k as part of the query slightly changes the space to n⋅O(1/ε)md with O(kd) query time within an approximation factor of 5+ε. Moreover, we show that our generic data structure for ANNS can give an alternative treatment of the approximate subtrajectory range searching problem studied by de Berg et al. [1].

Authenticating q-Gram-Based Similarity Search Results for Outsourced String Databases

Approximate string searches have been widely applied in many fields, such as bioinformatics, text retrieval, search engines, and location-based services (LBS). However, the approximate string search results from third-party servers may be incorrect due to the possibility of malicious third parties or compromised servers. In this paper, we design an authenticated index structure (AIS) for string databases, which is based on the Merkle hash tree (MHT) method and the q-gram inverted index. Our AIS can facilitate verification object (VO) construction for approximate string searches with edit distance thresholds. We design an efficient algorithm named GS2 for VO construction at the server side and search result verification at the user side. We also introduce an optimization method called GS2-opt that can reduce VO size dramatically. Finally, we conduct extensive experiments on real datasets to show that our proposed methods are efficient and promising.

Secure Computation Offloading for Marine IoT: An Energy-Efficient Design via Cooperative Jamming

In smart ocean, the Marine Internet of Things (M-IoT) is ubiquitously used for data acquisition, communication connections, task computation, and so on. However, high-efficient energy consumption and secure yet low-latency transmission are subject to IoT devices' computation capacity and transmit power limitation, as well as wireless communication's broadcast characteristics. In this paper, we investigate the secure computation offloading in M-IoT with the assistance of unmanned surface vehicles (USVs) subject to the eavesdropping attack. In particular, we consider a scenario in which USVs are firstly scheduled to set up a high-quality communication link to the high altitude platform (HAP) and then exploited to provide cooperative jamming for the communication security at the physical layer when the HAP is performing computation offloading. We jointly optimize the USVs' positions, the duration of data uploading, the workload offloaded by HAP, the HAP's transmission power, and each USV's jamming signal power to minimize the system-wise energy consumption for completing the total workloads under the latency constraint. Despite the non-convexity, we decompose the problem vertically into a top problem that optimizes the USVs' positions and a bottom problem that optimizes the other variables. The Polyblock outer Approximation and bisection Search based algorithm (PAS-Algorithm) is proposed to alternatively optimize the variables in the bottom problem. And the Code bAsed croSs Entropy algorithm (CASE-Algorithm) is proposed to obtain the suboptimal solution to the formulated joint optimization problem. Numerical results validate the accuracy and effectiveness of our algorithms as well as the performance of optimizing the total energy consumption compared with the benchmark algorithms

Towards Efficient Index Construction and Approximate Nearest Neighbor Search in High-Dimensional Spaces

The approximate nearest neighbor (ANN) search in high-dimensional spaces is a fundamental but computationally very expensive problem. Many methods have been designed for solving the ANN problem, such as LSH-based methods and graph-based methods. The LSH-based methods can be costly to reach high query quality due to the hash-boundary issues, while the graph-based methods can achieve better query performance by greedy expansion in an approximate proximity graph (APG). However, the construction cost of these APGs can be one or two orders of magnitude higher than that for building hash-based indexes. In addition, they fail short in incrementally maintaining APGs as the underlying dataset evolves. In this paper, we propose a novel approach named LSH-APG to build APGs and facilitate fast ANN search using a lightweight LSH framework. LSH-APG builds an APG via consecutively inserting points based on their nearest neighbor relationship with an efficient and accurate LSH-based search strategy. A high-quality entry point selection technique and an LSH-based pruning condition are developed to accelerate index construction and query processing by reducing the number of points to be accessed during the search. LSH-APG supports fast maintenance of APGs in lieu of building them from scratch as dataset evolves. Its maintenance cost and query cost for a point is proven to be less affected by dataset cardinality. Extensive experiments on real-world and synthetic datasets demonstrate that LSH-APG incurs significantly less construction cost but achieves better query performance than existing graph-based methods.

Approximate Content-Addressable Memories: A Review

Content-addressable memory (CAM) has been part of the memory market for more than five decades. CAM can carry out a single clock cycle lookup based on the content rather than an address. Thanks to this attractive feature, CAM is utilized in memory systems where a high-speed content lookup technique is required. However, typical CAM applications only support exact matching, as opposed to approximate matching, where a certain Hamming distance (several mismatching characters between a query pattern and the dataset stored in CAM) needs to be tolerated. Recent interest in approximate search has led to the development of new CAM-based alternatives, accelerating the processing of large data workloads in the realm of big data, genomics, and other data-intensive applications. In this review, we provide an overview of approximate CAM and describe its current and potential applications that would benefit from approximate search computing.

Research on the Roundness Approximation Search Algorithm of Si3N4 Ceramic Balls Based on Least Square and EMD Methods

This paper aims to obtain the best shape accuracy evaluation algorithm for silicon nitride ceramic balls after lapping, and to extract the initial signal of the ball surface to improve the accuracy and reliability of the algorithm. The research methods of this paper are as follows: Firstly, an analysis of the uniform envelope of the lapping trajectory of ceramic balls is carried out to verify whether the lapping trajectory after processing can achieve a consistent envelope on the balls' surface. On this basis, it is found through experiments that the standard deviation SD between the roundness deviations of different contour sections is small. The value is maintained at approximately 0.03 μm, and the roundness deviation can approximately replace the spherical deviation. Then the different contour sections of the sphere are sampled by the Taylor roundness instrument. Considering the uncertainty, the sampling points of different contour sections are averaged and used as the original signal of the sphere surface. Then the EMD method is used to process the signal to be detected on the sphere surface. The initial signal of the sphere surface is extracted by judging whether the number of ripples Kc obtained by decomposition is greater than the critical value. Then the initial signal is used as the input value of the approximation algorithm. Through the roundness deviation approximation algorithm based on the least square method, the given minimum approximation domain range is finely processed. The divided fine points are used as the center of the circle to intersect with the initial signal. The maximum, minimum, and range of each circle are calculated to obtain the roundness error based on the minimum circumscribed circle, the maximum inscribed circle, and the minimum region method. Finally, the calculated values are compared with those obtained by the traditional algorithm. The experimental results of this paper show that the algorithm is consistent with the roundness error measured by the instrument, compared with the mainstream evaluation criteria. In summary, the conclusions can be drawn as follows: Through a large number of experimental cases and comparative experiments, the algorithm has high accuracy and reliability. The research results of this paper have essential reference significance for accurately evaluating the shape accuracy of ceramic balls in actual production.

Analysis of braid manifolds using uniform manifold approximation and projection

Autonomous sonar target recognition has been an ongoing challenge in the underwater signal processing community for many decades. Aspect-dependent target scattering and waveguide propagation causes target features to overlap, which increases the difficulty of constructing feature dictionaries. Current machine learning algorithms can suffer from an inability to track morphing target-specific features. Other ”black-box” machine learning algorithms produce results that are not explainable. We seek to extend previous work on creating feature representations using braid manifolds. Specifically, we use Uniform Manifold Approximation and Projection (UMAP) as a dimensionality reduction technique to search for underlying features that lay on a manifold. UMAP is an unsupervised learning algorithm that is built upon Riemannian geometry and fuzzy simplicial sets. The low dimensional representation embedding that UMAP produces is computed using both stochastic approximate nearest neighbor search and stochastic gradient descent with negative sampling. The performance of UMAP embeddings will be compared against other common dimension reduction algorithms and evaluated using various classifiers. [The authors would like to thank the Office of Naval Research for funding under Grant No. N00014-21-1-2420 and the University of Iowa.]

Deep Uncoupled Discrete Hashing via Similarity Matrix Decomposition

Hashing has been drawing increasing attention in the task of large-scale image retrieval owing to its storage and computation efficiency, especially the recent asymmetric deep hashing methods. These approaches treat the query and database in an asymmetric way and can take full advantage of the whole training data. Though it has achieved state-of-the-art performance, asymmetric deep hashing methods still suffer from the large quantization error and efficiency problem on large-scale datasets due to the tight coupling between the query and database. In this article, we propose a novel asymmetric hashing method, called D eep U ncoupled D iscrete H ashing (DUDH), for large-scale approximate nearest neighbor search. Instead of directly preserving the similarity between the query and database, DUDH first exploits a small similarity-transfer image set to transfer the underlying semantic structures from the database to the query and implicitly keep the desired similarity. As a result, the large similarity matrix is decomposed into two relatively small ones and the query is decoupled from the database. Then both database codes and similarity-transfer codes are directly learned during optimization. The quantization error of DUDH only exists in the process of preserving similarity between the query and similarity-transfer set. By uncoupling the query from the database, the training cost of optimizing the CNN model for the query is no longer related to the size of the database. Besides, to further accelerate the training process, we propose to optimize the similarity-transfer codes with a constant-approximation solution. In doing so, the training cost of optimizing similarity-transfer codes can be almost ignored. Extensive experiments on four widely used image retrieval benchmarks demonstrate that DUDH can achieve state-of-the-art retrieval performance with remarkable training cost reduction (30× - 50× relative).

Convergence analysis of an ALF-based nonconvex splitting algorithm with SQP structure

<p style='text-indent:20px;'>In this paper, by combining the splitting method of augmented Lagrange function (ALF) with the sequential quadratic programming (SQP) approximation, a novel ALF-based splitting algorithm with SQP structure is proposed for multi-block linear constrained nonconvex separable optimization. The new algorithm uses ALF-based splitting idea to decompose the original problem into several small-scale subproblems. Meanwhile, the SQP approximation and Armijo-type line search are used to solve some subproblems with smoothness concurrently. Under the conventional weak hypothesis, the decreasing property of ALF as merit function is obtained. Furthermore, the global convergence, strong convergence and convergence rate results of the new algorithm in general sense are given.</p>

Genetic algorithm-based rule generation for approximate keyword search

Genetic algorithm-based rule generation for approximate keyword search

Genetic algorithm-based rule generation for approximate keyword search

Genetic algorithm-based rule generation for approximate keyword search

Feature adaptive multi-view hash for image search

<abstract><p>With the rapid development of network technology and small handheld devices, the amount of data has significantly increased and various kinds of data can be supplied to us at the same time. Recently, hashing technology has become popular in executing large-scale similarity search and image matching tasks. However, most of the prior hashing methods are mainly focused on the choice of the high-dimensional feature descriptor for learning effective hashing functions. In practice, real world image data collected from multiple scenes cannot be descriptive enough by using a single type of feature. Recently, several unsupervised multi-view hashing learning methods have been proposed based on matrix factorization, anchor graph and metric learning. However, large quantization error will be introduced via a sign function and the robustness of multi-view hashing is ignored. In this paper we present a novel feature adaptive multi-view hashing (FAMVH) method based on a robust multi-view quantization framework. The proposed method is evaluated on three large-scale benchmarks CIFAR-10, CIFAR-20 and Caltech-256 for approximate nearest neighbor search task. The experimental results show that our approach can achieve the best accuracy and efficiency in the three large-scale datasets.</p></abstract>

Near-neighbor preserving dimension reduction via coverings for doubling subsets of ℓ1

Randomized dimensionality reduction has been recognized as one of the cornerstones in handling high-dimensional data, originating in various foundational works such as the celebrated Johnson-Lindenstrauss Lemma. More specifically, nearest neighbor-preserving embeddings exist for ℓ2 (Euclidean) and ℓ1 (Manhattan) metrics, as well as doubling subsets of ℓ2, where doubling dimension is today the most effective way of capturing intrinsic dimensionality, as well as input structure in various applications. These randomized embeddings bound the distortion only for distances between the query point and a point set. Motivated by the foundational character of fast Approximate Nearest Neighbor search in ℓ1, this paper settles an important missing case, namely that of doubling subsets of ℓ1. In particular, we introduce a randomized dimensionality reduction by means of a near neighbor-preserving embedding, which is related to the decision-with-witness problem. The input set gets represented with a carefully chosen covering point set; in a second step, the algorithm randomly projects the latter. In order to obtain the covering point sets, we leverage either approximate r-nets or randomly shifted grids, with different tradeoffs between preprocessing time and target dimension. We exploit Cauchy random variables, and derive a concentration bound of independent interest. Our algorithms are rather simple and should therefore be useful in practice.

Unconventional application of k-means for distributed approximate similarity search

Similarity search based on a distance function in metric spaces is a fundamental problem for many applications. Queries for similar objects lead to the well-known machine learning task of nearest-neighbours identification. Many data indexing strategies, collectively known as Metric Access Methods (MAM), have been proposed to speed up these queries. Moreover, since exact approaches to solving similarity queries can be complex and time-consuming, alternative options have emerged to reduce query execution time, such as returning approximate results or resorting to distributed computing platforms. In this paper, we introduce MASK (Multilevel Approximate Similarity search with k-means), an unconventional application of the k-means algorithm as the foundation of a multilevel index structure for approximate similarity search suitable for metric spaces. We show that this method leverages inherent properties of k-means for this purpose, like representing high-density data areas with fewer prototypes. An implementation of this new indexing procedure is evaluated using a synthetic dataset and two real-world datasets in high-dimensional and high-sparsity spaces. Experimental tests show that MASK performs better than alternative algorithms for approximate similarity search. Results are promising and underpin the applicability of this novel indexing method in multiple domains.

An approximate search algorithm for the student-internship allocation problem

This paper proposes an approximate search algorithm to solve the student-internship allocation problem. The key idea of the algorithm is that in each iteration, each student unassigned to an enterprise will be assigned to an enterprise where the student ranks it highest, and it remains at maximum capacity. If the enterprise assigned to the student overcomes its capacity, then the enterprise will remove a student to whom it ranks the worst student to allow it not to overcome its capacity. Experimental results with randomly generated datasets show that our algorithm is efficient for the problem of large sizes.

Improving Natural Language Person Description Search from Videos with Language Model Fine-Tuning and Approximate Nearest Neighbor

Due to the ubiquitous nature of CCTV cameras that record continuously, there is a large amount of video data that are unstructured. Often, when these recordings have to be reviewed, it is to look for a specific person that fits a certain description. Currently, this is achieved by manual inspection of the videos, which is both time-consuming and labor-intensive. While person description search is not a new topic, in this work, we made two contributions. First, we improve upon the existing state-of-the-art by proposing unsupervised finetuning on the language model that forms a main part of the text branch of person description search models. This led to higher recall values on the standard dataset. The second contribution is that we engineered a complete pipeline from video files to fast searchable objects. Due to the use of an approximate nearest neighbor search and some model optimizations, a person description search can be performed such that the result is available immediately when deployed on a standard PC with no GPU, allowing an interactive search. We demonstrated the effectiveness of the system on new data and showed that most people in the videos can be successfully discovered by the search.

Semi‐Paired Semi‐Supervised Deep Hashing for cross‐view retrieval

AbstractDue to its fast computational speed and low storage cost, hashing has been effectively applied to large‐scale multimedia retrieval tasks, such as medical video and security video retrieval. Most existing cross‐view hashing methods require good matching information, however, this exact pairing relationship is difficult to fully realise in practice. The association between views is incomplete, as is the label information. This task of missing paired and labelled information is very challenging, but less explored in research. In this study, a semi‐supervised semi‐paired deep hashing for large‐scale data is proposed, named Semi‐Paired Semi‐Supervised Deep Hashing (SPSDH) to solve this challenging task. SPSDH is a novel end‐to‐end deep neural network model with high‐order affinity. A non‐local higher‐order affinity measure that better considers the multimodal neighbourhood structure is proposed. A common representation to associate different modalities is introduced, which combined with the labelled information greatly maintains the consistency within the modalities. SPSDH is evaluated on three benchmark datasets for large‐scale cross‐view approximate nearest neighbour search and compared with several state‐of‐the‐art hashing methods. Extensive experimental results demonstrate the superior performance of our proposed SPSDH in semi‐supervised semi‐paired retrieval tasks.