Search

Unmanned aerial vehicles (UAVs) can be flexibly deployed to offload cellular traffic or to provide video services for emergency scenarios without infrastructure. However, the inherent resource allocation and three-dimensional (3D) aerial trajectory design have not been formally studied. In this paper, we study the joint resource allocation and 3D aerial trajectory design for dynamic adaptive streaming over HTTP (DASH)-enabled services in a UAV communication system, where a UAV is employed as a base station for multiuser video streaming. Various factors are taken into account, including video data rate, quality variation, communication outage, play interruption, etc. By adopting a video streaming utility model, two fundamental problems are formulated with different practical aims: the first problem maximizes the minimum utility for all users within a given time horizon such that max-min fairness can be provided, and the second problem minimizes the UAV operation time subject to the individual utility requirement for all users to prolong UAV endurance. To tackle the first non-convex problem, we decouple it into three sub-problems, and a three-stage iterative algorithm is proposed to obtain a suboptimal solution by solving the three sub-problems with successive convex approximation and alternating optimization techniques. An exponential search based algorithm is proposed for the second problem by utilizing the structure of the considered problem and a similar three-stage iterative algorithm. Extensive simulations are carried out to evaluate the performance, and the results show that our proposed designs significantly outperform baseline schemes. Furthermore, our results reveal new insights of UAV movement for video streaming and unveil the tradeoff between utility and quality variance.

Deep Neural Networks (DNNs) are a powerful tool for solving complex tasks in many application domains. The high performance of DNNs demands significant computational resources, which might not always be available. Network quantization with mixed-precision across the layers can alleviate this high demand. However, determining layer-wise optimal bit-widths is non-trivial, as the search space is exponential. This article proposes a novel technique for allocating layer-wise bit-widths for a DNN using a multi-layer perceptron (MLP). The Kullback-Leibler (KL) divergence of the softmax outputs between the quantized and full precision network is used as the metric to quantify the quantization quality. We explore the relationship between the KL-divergence and the network size, and from our experiments observe that more aggressive quantization leads to higher divergence, and vice versa. The MLP is trained with layer-wise bit-widths as labels and their corresponding KL-divergence as the input. The MLP training set, i.e. the pairs of the layer-wise bit-widths and their corresponding KL-divergence, is collected using a Monte Carlo sampling of the exponential search space. We introduce a penalty term in the loss to ensure that the MLP learns to predict bit-widths resulting in the smallest network size. We show that the layer-wise bit-width predictions from the trained MLP result in reduced network size without degrading accuracy while achieving better or comparable results with SOTA work but with less computational overhead. Our method achieves up to 6x, 4x, 4x compression on VGG16, ResNet50, and GoogLeNet respectively, with no accuracy drop compared to the original full precision pretrained model, on the ImageNet dataset.

The critical issue identified in recent times is the high utility itemset mining (HUIM). It may be used for showing products that are profitably utilizing considering the factors of profit and quality as opposed to the frequent itemset (FIM) or the association rule (ARM) mining. There are numerous high utility itemset mining (HUIs) algorithms, mostly designed for handling the exponential search space to discover the HUIs at the time the number of the distinct items and the database that was very large. For this purpose, a meta-heuristic algorithm is designed for HUIs mining, working based on the genetic algorithm (GA) and the Dolphin echolocation optimization (DEO). The intended purpose of this evolutionary computation (EC) techniques on the DEO, here only fewer parameters are required to be compared to the approaches that are based on the GA. As the traditional DEO mechanism had been found for handling this continuous problem, an efficient algorithm based on the DEO called the high-utility itemset mining-DEO (HUIM-DEO) was proposed. To prove that the algorithm proposed was able to outperform the other heuristic algorithms to mine the HUIs for the time taken for execution, the number of HUIs discovered, and their convergence.

This article presents a user spatio-temporal behavior aware cache selection framework to facilitate device-to-device multicast (D2MD) communication that minimizes the number of caches required while achieving a desired user load on the cellular network. Consequently, it alleviates the caching load on the cellular network. The optimization problem formulated to minimize the number of caches is combinatorial in nature with an exponential search space. Hence, a greedy algorithm for cache selection is proposed to reduce the search space. It has been shown that the greedy algorithm has a complexity $\mathcal {O}(K^2)$ where $K$ is the number of users. A real-world location-based inhomogeneous Markov chain is presented to model the joint spatio-temporal behavior of the users. The diurnal variation of observable user load on the core network as well as sum-rate of non-caching users has been demonstrated for real-world location data traces. It has been shown that the proposed framework not only achieves the desired user load but also helps in improving the sum-rate of non-caching users as compared to the mobility-unaware selection of caches.

The fidelity of applications on near-term quantum computers is limited by hardware errors. In addition to errors that occur during gate and measurement operations, a qubit is susceptible to idling errors, which occur when the qubit is idle and not actively undergoing any operations. To mitigate idling errors, prior works in the quantum devices community have proposed Dynamical Decoupling (DD), that reduces stray noise on idle qubits by continuously executing a specific sequence of single-qubit operations that effectively behave as an identity gate. Unfortunately, existing DD protocols have been primarily studied for individual qubits and their efficacy at the application-level is not yet fully understood. Our experiments show that naively enabling DD for every idle qubit does not necessarily improve fidelity. While DD reduces the idling error-rates for some qubits, it increases the overall error-rate for others due to the additional operations of the DD protocol. Furthermore, idling errors are program-specific and the set of qubits that benefit from DD changes with each program. To enable robust use of DD, we propose Adaptive Dynamical Decoupling (ADAPT), a software framework that estimates the efficacy of DD for each qubit combination and judiciously applies DD only to the subset of qubits that provide the most benefit. ADAPT employs a Decoy Circuit, which is structurally similar to the original program but with a known solution, to identify the DD sequence that maximizes the fidelity. To avoid the exponential search of all possible DD combinations, ADAPT employs a localized algorithm that has linear complexity in the number of qubits. Our experiments on IBM quantum machines (with 16-27 qubits) show that ADAPT improves the application fidelity by 1.86x on average and up-to 5.73x compared to no DD and by 1.2x compared to DD on all qubits., 16 Figures, 5 Tables

Interest management is employed in most distributed simulation to filter irrelevant messages on network and is essential for the performance and scalability of simulation systems. In High-Level Architecture (HLA), the data distribution management (DDM) adopts an interest matching algorithm to filter unnecessary communications among federates by computing the overlap statuses between the update and subscription regions. In many spatial applications, only a portion of regions changes gradually at each time step. Hence, dynamic matching for selective modified regions is necessary. The dynamic sort-based matching algorithm has proved to be the most efficient. However, the existing dynamic sort-based matching algorithms do not separate modified regions from static regions, which will produce large amounts of unnecessary comparisons and swap operations between modified and static regions. To solve the problem, we propose an exponential search enhanced dynamic sort-based interest matching algorithm called EDSIM. In EDSIM, (1) the modified regions are separated from static regions; thus, the latter are independent from the sorting operations; (2) an exponential search method is adopted to calculate the rechecking sets for modified regions, reducing unnecessary bound comparisons between modified and static regions; (3) a bubble-sort-based matching algorithm is used to update the overlap statuses between modified regions. Experimental results demonstrate that the EDSIM algorithm achieves significantly better performance than the major existing dynamic matching algorithms under various experimental scenarios.

Software reproducibility is important for re-usability and the cumulative progress of research. An important manifestation of unreproducible software is the changed outcome of software builds over time. While enhancing code reuse, the use of open-source dependency packages hosted on centralized repositories such as PyPI can have adverse effects on build reproducibility. Frequent updates to these packages often cause their latest versions to have breaking changes for applications using them. Large Python applications risk their historical builds becoming unreproducible due to the widespread usage of Python dependencies, and the lack of uniform practices for dependency version specification. Manually fixing dependency errors requires expensive developer time and effort, while automated approaches face challenges of parsing unstructured build logs, finding transitive dependencies, and exploring an exponential search space of dependency versions. In this paper, we investigate how open-source Python projects specify dependency versions, and how their reproducibility is impacted by dependency packages. We propose a tool PyDFix to detect and fix unreproducibility in Python builds caused by dependency errors. PyDFix is evaluated on two bug datasets BugSwarm and BugsInPy, both of which are built from real-world open-source projects. PyDFix analyzes a total of 2,702 builds, identifying 1,921 (71.1%) of them to be unreproducible due to dependency errors. From these, PyDFix provides a complete fix for 859 (44.7%) builds, and partial fixes for an additional 632 (32.9%) builds.

Although linear regressions are essential for learned index structures, most implementations use Simple Linear Regression, which optimizes the squared error. Since learned indexes use exponential search, regressions that optimize the logarithmic error are much better tailored for the use-case. By using this fitting optimization target, we can significantly improve learned index’s lookup performance with no architectural changes. While the log-error is harder to optimize, our novel algorithms and optimization heuristics can bring a practical performance improvement of the lookup latency. Even in cases where fast build times are paramount, log-error regressions still provide a robust fallback for degenerated leaf models. The resulting regressions are much better suited for learned indexes, and speed up lookups on data sets with outliers by over a factor of 2.

With huge design spaces for unique chemical and mechanical properties, we remove a roadblock to computational design of {high-entropy alloys} using a metaheuristic hybrid Cuckoo Search (CS) for "on-the-fly" construction of Super-Cell Random APproximates (SCRAPs) having targeted atomic site and pair probabilities on arbitrary crystal lattices. Our hybrid-CS schema overcomes large, discrete combinatorial optimization by ultrafast global solutions that scale linearly in system size and strongly in parallel, e.g. a 4-element, 128-atom model [a $10^{73+}$ space] is found in seconds -- a reduction of 13,000+ over current strategies. With model-generation eliminated as a bottleneck, computational alloy design can be performed that is currently impossible or impractical. We showcase the method for real alloys with varying short-range order. Being problem-agnostic, our hybrid-CS schema offers numerous applications in diverse fields., 10 pages, 6 figures, submitted to Nature Computational Science

Despite significant progress in general AI planning, certain domains remain out of reach of current AI planning systems. Sokoban is a PSPACE-complete planning task and represents one of the hardest domains for current AI planners. Even domain-specific specialized search methods fail quickly due to the exponential search complexity on hard instances. Our approach based on deep reinforcement learning augmented with a curriculum-driven method is the first one to solve hard instances within one day of training while other modern solvers cannot solve these instances within any reasonable time limit. In contrast to prior efforts, which use carefully handcrafted pruning techniques, our approach automatically uncovers domain structure. Our results reveal that deep RL provides a promising framework for solving previously unsolved AI planning problems, provided a proper training curriculum can be devised., Comment: 8 pages, 6 figures, accepted by IJCAI 2020

International audience; Given a knowledge base (KB), rule mining finds rules such as "If two people are married, then they live (most likely) in the same place". Due to the exponential search space, rule mining approaches still have difficulties to scale to today's large KBs. In this paper, we present AMIE 3, a system that employs a number of sophisticated pruning strategies and optimizations. This allows the system to mine rules on large KBs in a matter of minutes. Most importantly, we do not have to resort to approximations or sampling, but are able to compute the exact confidence and support of each rule. Our experiments on DBpedia, YAGO, and Wikidata show that AMIE 3 beats the state of the art by a factor of more than 15 in terms of runtime.

Higher-order mutation has the potential for improving major drawbacks of traditional first-order mutation, such as by simulating more realistic faults or improving test optimization techniques. Despite interest in studying promising higher-order mutants, such mutants are difficult to find due to the exponential search space of mutation combinations. State-of-the-art approaches rely on genetic search, which is often incomplete and expensive due to its stochastic nature. First, we propose a novel way of finding a complete set of higher-order mutants by using variational execution, a technique that can, in many cases, explore large search spaces completely and often efficiently. Second, we use the identified complete set of higher-order mutants to study their characteristics. Finally, we use the identified characteristics to design and evaluate a new search strategy, independent of variational execution, that is highly effective at finding higher-order mutants even in large code bases.

Euclidean distance optimized antenna selection (EDAS) has proven to be able to achieve high transmit diversity gains for spatial modulation (SM) multiple-input multiple-output (MIMO) systems, but the exhaustive search over all possible antenna subsets may result in an intractable search complexity, especially for large-scale MIMO (LS-MIMO) systems. In this paper, a novel EDAS-equivalent criterion, relaying on matrix dimension reduction, is proposed to reduce the search complexity. Based upon this criterion, for small-scale SM-MIMO systems, a pair of transmit antenna selection (TAS) schemes, namely, the tree search based antenna selection (TSAS), and the decremental antenna selection (D-AS) schemes, are developed. Specifically, the TSAS scheme works by iteratively splitting the original Euclidean distance (ED) matrix into submatrices with smaller dimensions. TSAS achieves the same bit error rate (BER) performance as that of the exhaustive search at the cost of exponential search complexity. To address this issue, D-AS is carried out by excluding one specific transmit antenna from all the possible candidates in each iteration based on the channel information. D-AS is able to offer a near-optimal BER performance with a further reduction in terms of the search complexity. Moreover, we extend the D-AS scheme to LS-SM-MIMO systems by sorting all the ED elements with the low-complexity heapsort algorithm. Our simulation results show that the proposed D-AS algorithm outperforms conventional TAS schemes conceived for LS-SM-MIMO systems.

Human actions are, inherently, structured patterns of body movements. We explore ensembles of hierarchical spatio-temporal trees, discovered directly from training data, to model these structures for action recognition and spatial localization. Discovery of frequent and discriminative tree structures is challenging due to the exponential search space, particularly if one allows partial matching. We address this by first building a concise action word vocabulary via discriminative clustering of the hierarchical space-time segments, which is a two-level video representation that captures both static and non-static relevant space-time segments of the video. Using this vocabulary we then utilize tree mining with subsequent tree clustering and ranking to select a compact set of discriminative tree patterns. Our experiments show that these tree patterns, alone, or in combination with shorter patterns (action words and pairwise patterns) achieve promising performance on three challenging datasets: UCF Sports, HighFive and Hollywood3D. Moreover, we perform cross-dataset validation, using trees learned on HighFive to recognize the same actions in Hollywood3D, and using trees learned on UCF-Sports to recognize and localize the similar actions in JHMDB. The results demonstrate the potential for cross-dataset generalization of the trees our approach discovers.

The worldwide energy consumption has been growing in aggregate at a tremendous rate, and a majority of the same is due to heating ventilation air conditioning (HVAC) loads in urban buildings. With the help of the recent advances in energy management and optimization techniques, the operations and functioning of these devices can now be managed and controlled efficiently for an improved energy consumption scenario and thereby reducing cost. In this article, we propose a multiobjective optimal scheduling framework based on Johnson's elementary circuit finding algorithm for controlling HVAC devices, specifically for buildings that require continuous thermal comfort maintenance. Two primary objectives addressed in this article are: minimizing power fluctuation and maximizing thermal comfortability of the users. We use standard comfortability indices to quantify thermal comfortability. To reduce the computation time, we also propose two intelligent improvement schemes that prune the exponential search space of Johnson's algorithm. Furthermore, a new greedy scheduling algorithm has been proposed to obtain near-optimal solutions efficiently. All the proposed approaches have been studied in a simulated environment depicting a real-world scenario to evaluate their efficiency and effectiveness for practical implementations, including a comparative analysis with Karp's minimum mean cycle algorithm in this problem setup.

Interpretable explanations for recommender systems and other machine learning models are crucial to gain user trust. Prior works that have focused on paths connecting users and items in a heterogeneous network have several limitations, such as discovering relationships rather than true explanations, or disregarding other users' privacy. In this work, we take a fresh perspective, and present PRINCE: a provider-side mechanism to produce tangible explanations for end-users, where an explanation is defined to be a set of minimal actions performed by the user that, if removed, changes the recommendation to a different item. Given a recommendation, PRINCE uses a polynomial-time optimal algorithm for finding this minimal set of a user's actions from an exponential search space, based on random walks over dynamic graphs. Experiments on two real-world datasets show that PRINCE provides more compact explanations than intuitive baselines, and insights from a crowdsourced user-study demonstrate the viability of such action-based explanations. We thus posit that PRINCE produces scrutable, actionable, and concise explanations, owing to its use of counterfactual evidence, a user's own actions, and minimal sets, respectively., WSDM 2020, 9 pages

Illuminating rooms is something that is taken as for granted. This paper presents an evolutionary method for solving this problem. The paper aims to find a minimalist arrangement of lamps in user-defined or random rooms such that they are optimally lit. Genetic algorithm is used to explore exponential search space of lamps placement and find those that suit the given criteria. Blender is used as a 3-D tool for this experiment. Results show promising arrangements of lamps that can be used in situations that require an emitter of signals in an obstruction filled environment.

State-of-the-art named entity recognition (NER) systems have been improving continuously using neural architectures over the past several years. However, many tasks including NER require large sets of annotated data to achieve such performance. In particular, we focus on NER from clinical notes, which is one of the most fundamental and critical problems for medical text analysis. Our work centers on effectively adapting these neural architectures towards low-resource settings using parameter transfer methods. We complement a standard hierarchical NER model with a general transfer learning framework consisting of parameter sharing between the source and target tasks, and showcase scores significantly above the baseline architecture. These sharing schemes require an exponential search over tied parameter sets to generate an optimal configuration. To mitigate the problem of exhaustively searching for model optimization, we propose the Dynamic Transfer Networks (DTN), a gated architecture which learns the appropriate parameter sharing scheme between source and target datasets. DTN achieves the improvements of the optimized transfer learning framework with just a single training setting, effectively removing the need for exponential search.

Configuring a software system to optimize non-functional properties is a hard task. There are dozens to thousands of configuration options that can affect performance, energy consumption, and other attributes of the resulting program. Even worse, options may interact, such that their combined presence (or absence) has an influence on a non-functional property.In this talk, I report on our experiences with learning different performance models based on a multitude of sampling techniques. The goal is to raise awareness of the distinct challenges in this domain: constraints among options, the exponential search space, and suitable sampling and learning techniques. I show a variety of approaches including their strengths and weaknesses and close the talk with new challenges relevant for our community: changing environments and reproducibility.

We demonstrate that agent-based simulations can exhibit results in line with classic macroeconomic theory. In particular, we present an agent-based simulation of an Arrow–Debreu economy that accurately exhibits the Stolper–Samuelson effect as an emergent property. Absent of a Walrasian auctioneer or any other central coordination, we let firm and consumer agents of different types interact in an open, money-driven market. Exogenous preference shocks result in price and wage shifts that are in accordance with the general equilibrium solution, not only qualitatively but also quantitatively with high accuracy. Key to this achievement are three independent measures. First, we overcome the poor input synchronization of conventional price finding heuristics of firms in agent-based models by introducing sensor prices, a novel approach to price finding that decouples information exploitation from information exploration. Second, we improve accuracy and convergence by employing exponential search as exploration algorithm. Third, we normalize prices indirectly by fixing dividends, thereby stabilizing the system’s dynamics.

High-utility itemset mining (HUIM) is a critical issue in recent years since it can be used to reveal the profitable products by considering both the quantity and profit factors instead of frequent itemset mining (FIM) or association-rule mining (ARM). Several algorithms have been presented to mine high-utility itemsets (HUIs) and most of them have to handle the exponential search space for discovering HUIs when the number of distinct items and the size of database are very large. In the past, a heuristic HUPE$$ _\mathrm{umu}$$umu-GRAM algorithm was proposed to mine HUIs based on genetic algorithm (GA). For the evolutionary computation (EC) techniques of particle swarm optimization (PSO), it only requires fewer parameters compared to the GA-based approaches. Since the traditional PSO mechanism is used to handle the continuous problem, in this paper, the discrete PSO is adopted to encode the particles as the binary variables. An efficient PSO-based algorithm, namely HUIM-BPSO, is proposed to efficiently find HUIs. The designed HUIM-BPSO algorithm finds the high-transaction-weighted utilization 1-itemsets (1-HTWUIs) as the size of the particles based on transaction-weighted utility (TWU) model, which can greatly reduce the combinational problem in evolution process. The sigmoid function is adopted in the updating process of the particles for the designed HUIM-BPSO algorithm. An OR/NOR-tree structure is further developed to reduce the invalid combinations for discovering HUIs. Substantial experiments on real-life datasets show that the proposed algorithm outperforms the other heuristic algorithms for mining HUIs in terms of execution time, number of discovered HUIs, and convergence.

Expensive data movement makes the optimal target-specific selection of data-locality transformations essential. Loop fusion and tiling are the most important data-locality transformations. Their optimal selection is hard since good tile size choices inherently depend on the fusion choices and vice versa. Existing approaches avoid this difficulty by optimizing independent analytical models or by reverting to heuristics. Absinthe formulates the first unified linear optimization problem to derive single shot fusion and tile size decisions for stencil codes. At the core of our optimization problem, we place a learned analytic performance model that captures the characteristics of the target system. The tuned application kernels demonstrate excellent performance within 10% of exhaustively auto-tuned versions and up to 74% faster than the results of independent optimization with max fusion heuristic and Absinthe tile size selection. While the full search space is non-linear, bounding it to relevant solutions enables the efficient exploration of the exponential search space using linear solvers. As a result, the tuning of our application kernels takes less than one minute. Our approach thus establishes the foundations for next-generation compilers, which exploit empirical information to guide target-specific code transformations.

We tackle two long-standing problems related to re-expansions in heuristic search algorithms. For graph search, A* can require $\Omega(2^{n})$ expansions, where $n$ is the number of states within the final $f$ bound. Existing algorithms that address this problem like B and B' improve this bound to $\Omega(n^2)$. For tree search, IDA* can also require $\Omega(n^2)$ expansions. We describe a new algorithmic framework that iteratively controls an expansion budget and solution cost limit, giving rise to new graph and tree search algorithms for which the number of expansions is $O(n \log C)$, where $C$ is the optimal solution cost. Our experiments show that the new algorithms are robust in scenarios where existing algorithms fail. In the case of tree search, our new algorithms have no overhead over IDA* in scenarios to which IDA* is well suited and can therefore be recommended as a general replacement for IDA*.

Health systems worldwide are confronted with enormous amounts of heterogenous data from various clinical sources. Additionally, the trend toward precision medicine, in order to tailor decisions and therapies to the individual patient, generates enormous amounts of *omics data. Modern medicine is turning into a data science. The problem is that most of these data are in arbitrarily high dimensions, making manual analysis difficult, yet impossible for humans. This requires automated methods of data science, which is what machine learning provides. Automatic machine learning methods show impressive results in dealing with big data, because such algorithms can learn from training samples to gain knowledge from experience, thus helping to make decisions and predictions. The more data, the better these algorithms perform. Unfortunately, in medicine, we are often confronted with a small number of data sets, “little data,” rare events, and complex data, where automatic approaches suffer from insufficient training sets. This calls for interactive machine learning with a doctor in the loop, who can be beneficial in solving computationally hard problems, where human expertise can help to reduce an exponential search space through heuristic selection of data samples. Therefore, what would otherwise be an NP-hard problem reduces greatly in complexity through the input and the assistance of a medical professional involved in the machine learning pipeline.

Search moves for finding novels in five search tasks and two catalogs were analyzed. Search tasks reflected the following search tactics: known-author search, topical search, open-ended browsing, search by analogy, and searching without a query. The most used search moves in both catalogs across all tasks were querying, search results inspection, and book page examination. In a traditional catalog, more effort was needed in the form of queries, search moves, and opened book pages to gain equivalent average book scores when compared with an enriched catalog. In a traditional catalog, a typical search strategy for interesting titles seemed to involve issuing queries and considering suitable entry terms carefully, and devoting more attention to search results instead of book pages. In an enriched catalog, a common approach involved time devoted to exploring the catalog's enriched front page and multiple entry points together with attention to the enriched results list.

Search algorithms or methodologies play an important role in any database, data mining, or any professional application. Given their importance, no new search algorithm was developed in the past decade. For a search methodology to be applicable, it has to be novel and faster than existing algorithms. This paper will present a new search methodology that has been awarded a patent for novelty (United States Patent no. 9009200). This search method will be called Thabit's algorithm in this paper. Moreover, the paper will show some test results to show the superiority of this methodology to several search algorithms like binary search, hashing, AVL trees and other search algorithms. For example, when searching a list of 50 million airline PNR numbers, the search time for Microsoft Hash was 18649ms, but when the list became one billion elements the search time increased to 2070318ms with a collision of 57%. Binary search was fairly stable with a search time of 116897ms in a list of 50 million and 2771639ms in the list of one billion elements. On the other hand, searching with Thabit's algorithm took 8517ms for a list of 50 million elements and 194739ms for the list of one billion elements. The AVL tree had 249269ms and 11995298 ms respectively. The comparison was done for space requirements, where binary search took 44703.6 MB for the billion elements. Thabit's algorithm, Microsoft Hash and AVL tree required 60883.3 MB, 115578.4 MB, and 104000.3 MB of space, respectively. To achieve these results, a graduate student was employed and a makerplace lab was created using a server and two laptops. In the makerplace lab the Thabit's algorithm was implemented. The implementation involved processing each item one character at a time or two, three, four, … etc characters at a time and checking which data structures led to better performance. In the Thabit's algorithm, three dimensional arrays ware replaced with one dimensional arrays. Furthermore, much study went into data size; checking the effect of increasing the size of the data and finding when thrashing starts. Finally, trying to run the programs in parallel, and checking speed gain and how this gain compares to Amdahl's law. Also, tests were done on searching through a large number of 16 digit debit card with great results using Thabit's algorithm.

In many real-world applications, data is collected in high dimensional spaces. However, not all dimensions are relevant for data analysis. Instead, interesting knowledge is hidden in correlated subsets of dimensions (i.e., subspaces of the original space). Detecting these correlated subspaces independently of the underlying mining task is an open research problem. It is challenging due to the exponential search space. Existing methods have tried to tackle this by utilizing Apriori search schemes. However, their worst case complexity is exponential in the number of dimensions; and even in practice they show poor scalability while missing high quality subspaces.This paper features a scalable subspace search scheme (4S), which overcomes the efficiency problem by departing from the traditional levelwise search. We propose a new generalized notion of correlated subspaces which gives way to transforming the search space to a correlation graph of dimensions. We perform a direct mining of correlated subspaces in this graph, and then, merge subspaces based on the MDL principle in order to obtain high dimensional subspaces with minimal redundancy. We theoretically show that our search scheme is more general than existing search schemes. Our empirical results reveal that 4S in practice scales near-linearly with both database size and dimensionality, and produces higher quality subspaces than state-of-the-art methods.

Utility mining is the study of itemset mining from the consideration of utilities. It is the utility-based itemset mining approach to find itemsets conforming to user preferences. Modern research in mining high-utility itemsets (HUI) from the databases faces two major challenges: exponential search space and database-dependent minimum utility threshold. The search space is extremely vast when the number of distinct items and the size of the database are very large. Data analysts must specify suitable minimum utility thresholds for their mining tasks, although they might have no knowledge pertaining to their databases. Moreover, a utility-mining algorithm supports only an itemset with positive item values. To evade these problems, two approaches are presented for mining HUI containing negative item values from transaction databases: with/without specifying the minimum utility threshold through a genetic algorithm with ranked mutation. To the best of our knowledge, this is the first work on mining HUI with ...

An online group testing method to search for a hidden object in a discrete search space is proposed. A relevant example is a search after a nonconforming unit in a batch, while many other applications can be related. A probability mass function is defined over the search space to represent the probability of an object (e.g., a nonconforming unit) to be located at some point or subspace. The suggested method follows a stochastic local search procedure and can be viewed as a generalization of the Learning Real-Time A* (LRTA*) search algorithm, while using informational distance measures over the searched space. It is proved that the proposed Informational LRTA* (ILRTA*) algorithm converges and always terminates. Moreover, it is shown that under relevant assumptions, the proposed algorithm generalizes known optimal information-theoretic search procedures, such as the offline Huffman search or the generalized optimum testing algorithm. However, the ILRTA* can be applied to new situations, such as a search wit...

We consider the problem of isolating the real roots of a square-free polynomial with integer coefficients using the classic variant of the continued fraction algorithm (CF), introduced by Akritas. We compute a lower bound on the positive real roots of univariate polynomials using an exponential search. This allows us to derive a worst-case bound of [email protected]?"B(d^[email protected]^2) for isolating the real roots of a polynomial with integer coefficients using the classic variant of CF, where d is the degree of the polynomial and @t the maximum bitsize of its coefficients. This improves the previous bound of Sharma by a factor of d^3 and matches the bound derived by Mehlhorn and Ray for another variant of CF which is combined with subdivision; it also matches the worst-case bound of the classical subdivision-based solvers sturm, descartes, and bernstein.

The rapid growth of World Wide Web has resulted in massive information from varied sources rising at an exponential rate. The high availability of such disparate information has precipitated the need of automatic text categorization for managing, organizing huge data and knowledge discovery. Main challenges of text classification include high dimensionality of feature space and classification accuracy. Thus, to make classifiers more accurate and efficient, there arises the need of Feature Selection. Genetic algorithms have gained much attention over traditional methods due to its simplicity and robustness to solve the optimization problem and high exponential search ability. Thus, the paper focuses on using Genetic Algorithm (GA) for Feature Selection to obtain optimal features for classifying unstructured data. We build a fuzzy rule-based classifier that automatically generates fuzzy rules for classification. The experiments are conducted on two datasets namely 20-Newsgroup and Reuters-21578 and the results indicate that GA outperforms Principal Component Analysis (PCA).

Complex queries are becoming commonplace with the growing use of decision support systems. These complex queries often have a lot of common sub-expressions, either within a single query, or across multiple such queries. The focus of this work is to speed up query execution by exploiting these common subexpressions. Given a set of queries in a batch, multi-query optimization aims at exploiting common subexpressions among these queries to reduce evaluation cost. Multi-query optimization has hitherto been viewed as impractical, since earlier algorithms were exhaustive, and explore a doubly exponential search space. We present novel heuristics for multi-query optimization, and demonstrate that optimization using these heuristics provides significant benefits over traditional optimization, at a very acceptable overhead in optimization time. In online environments, where the queries are posed as a part of an ongoing stream instead of in a batch, individual query response times can be greatly improved by caching final/intermediate results of previous queries, and using them to answer later queries. An automatic caching system that makes intelligent decisions on what results to cache would be an important step towards knobs-free operation of a database system. We describe an automatic query caching system called Exchequer which is closely coupled with the optimizer to ensure that the caching system and the optimizer make mutually consistent decisions, and experimentally illustrate the benefits of this approach. Further, because the presence of views enhances query performance, materialized views are increasingly being supported by commercial database/data warehouse systems. Whenever the data warehouse is updated, the materialized views must also be updated. We show how to find an efficient plan for maintenance of a set of views, by exploiting common subexpressions between different view maintenance expressions. These common subexpressions may be materialized temporarily during view maintenance. Our algorithms also choose additional subexpressions/indices to be materialized permanently (and maintained along with other materialized views), to speed up view maintenance. In addition to faster view maintenance, our algorithms can also be used to efficiently select materialized views to speed up query workloads.

Sentiment analysis is a critical task of extracting subjective information from online text documents. Ensemble learning can be employed to obtain more robust classification schemes. However, most approaches in the field incorporated feature engineering to build efficient sentiment classifiers. The purpose of our research is to establish an effective sentiment classification scheme by pursuing the paradigm of ensemble pruning. Ensemble pruning is a crucial method to build classifier ensembles with high predictive accuracy and efficiency. Previous studies employed exponential search, randomized search, sequential search, ranking based pruning and clustering based pruning. However, there are tradeoffs in selecting the ensemble pruning methods. In this regard, hybrid ensemble pruning schemes can be more promising. In this study, we propose a hybrid ensemble pruning scheme based on clustering and randomized search for text sentiment classification. Furthermore, a consensus clustering scheme is presented to deal with the instability of clustering results. The classifiers of the ensemble are initially clustered into groups according to their predictive characteristics. Then, two classifiers from each cluster are selected as candidate classifiers based on their pairwise diversity. The search space of candidate classifiers is explored by the elitist Pareto-based multi-objective evolutionary algorithm. For the evaluation task, the proposed scheme is tested on twelve balanced and unbalanced benchmark text classification tasks. In addition, the proposed approach is experimentally compared with three ensemble methods (AdaBoost, Bagging and Random Subspace) and three ensemble pruning algorithms (ensemble selection from libraries of models, Bagging ensemble selection and LibD3C algorithm). Results demonstrate that the consensus clustering and the elitist pareto-based multi-objective evolutionary algorithm can be effectively used in ensemble pruning. The experimental analysis with conventional ensemble methods and pruning algorithms indicates the validity and effectiveness of the proposed scheme. © 2017 Elsevier Ltd

We present a novel 3D reassembly method for fragmented, thin objects with unknown geometry. Unlike past methods, we do not make any restrictive assumptions about the overall shape of the object, or its painted texture. Our key observation is that regardless of the object's shape, matching fragments will have similar geometry and photometry along and across their adjoining regions. We begin by encoding the scale variability of each fragment's boundary contour in a multichannel, 2D image representation. Using this multichannel boundary contour representation, we identify matching sub-contours via 2D partial image registration. We then align the fragments by minimizing the distance between their adjoining regions while simultaneously ensuring geometric continuity across them. The configuration of the fragments as they are incrementally matched and aligned form a graph structure that we use to improve subsequent matches. By detecting cycles in this graph, we identify subsets of fragments with interdependent alignments. We then minimize the error within the subsets to achieve a globally optimal alignment. We leverage user feedback to cull the otherwise exponential search space; after each new match is found and aligned, it is presented to a user for confirmation or rejection. Using ceramic pottery as the driving example, we demonstrate the accuracy and efficiency of our method on six real-world datasets.

The inference of gene regulatory networks (GRNs) from expression data can mine the direct regulations among genes and gain deep insights into biological processes at a network level. During past decades, numerous computational approaches have been introduced for inferring the GRNs. However, many of them still suffer from various problems, e.g., Bayesian network (BN) methods cannot handle large-scale networks due to their high computational complexity, while information theory-based methods cannot identify the directions of regulatory interactions and also suffer from false positive/negative problems. To overcome the limitations, in this work we present a novel algorithm, namely local Bayesian network (LBN), to infer GRNs from gene expression data by using the network decomposition strategy and false-positive edge elimination scheme. Specifically, LBN algorithm first uses conditional mutual information (CMI) to construct an initial network or GRN, which is decomposed into a number of local networks or GRNs. Then, BN method is employed to generate a series of local BNs by selecting the k-nearest neighbors of each gene as its candidate regulatory genes, which significantly reduces the exponential search space from all possible GRN structures. Integrating these local BNs forms a tentative network or GRN by performing CMI, which reduces redundant regulations in the GRN and thus alleviates the false positive problem. The final network or GRN can be obtained by iteratively performing CMI and local BN on the tentative network. In the iterative process, the false or redundant regulations are gradually removed. When tested on the benchmark GRN datasets from DREAM challenge as well as the SOS DNA repair network in E.coli, our results suggest that LBN outperforms other state-of-the-art methods (ARACNE, GENIE3 and NARROMI) significantly, with more accurate and robust performance. In particular, the decomposition strategy with local Bayesian networks not only effectively reduce the computational cost of BN due to much smaller sizes of local GRNs, but also identify the directions of the regulations., Author Summary Gene regulatory network (GRN) represents how some genes encode regulatory molecules such as transcription factors or microRNAs for regulating the expression of other genes. Accurate inference of GRN is an important task to understand the biological activity from signal emulsion to metabolic dynamics, prioritize potential drug targets of various diseases, devise effective therapeutics, and discover the novel pathways. In this paper, we propose a novel local Bayesian network (LBN) algorithm to improve the accuracy of GRN inference from gene expression data by exploring advantages of Bayesian network (BN) and conditional mutual information (CMI) methods. BNs with kNN network decomposition and CMI are respectively introduced to reduce the high computational complexity of BN and remove the false or redundant regulation interactions. The superior performance of the proposed LBN approach is demonstrated on GRN datasets from DREAM challenge as well as the SOS DNA repair network in E. coli.

The ability to detect and analyze interesting subgraph patterns on large and dynamic graph-structured data in near-real time is crucial for many applications; example includes anomaly detection in phone call networks, advertisement targeting in social networks, malware detection in file download graphs, and many more. Such patterns often need to reason about how the nodes are connected to each other (i.e., the structural component) as well as how the nodes behave in the network (i.e., the activity component). An example of such an activity-driven subgraph pattern is a clique of users in a social network (the structural predicate), who each have posted more than 10 messages in last 2 hours (the activity-based predicate). In this paper, we present Casqd, a system for continuous detection and analysis of such active subgraph pattern queries over large dynamic graphs. Some of key challenges in executing such queries include: handling a wide variety of user-specified activities of interest, low selectivities of activity-based predicates and the resultant exponential search space, and high ingestion rates. A key abstraction in Casqd is a notion called graph-view, which acts as an independence layer between the query language and the underlying physical representation of the graph and the active attributes. This abstraction is aimed at simplifying the query language, while empowering the query optimizer. Considering the balance between expressibility (i.e., patterns that cover many real-world use cases) and optimizability of such patterns, we primarily focus on efficient continuous detection of the active regular structures (specifically, active cliques, active stars, and active bi-cliques). We develop a series of optimization techniques including model-based neighborhood explorations, lazy evaluation of the activity predicates, neighborhood-based search space pruning, and others, for efficient query evaluation. We perform a thorough comparative study of the execution strategies under various settings, and show that our system is capable of achieving event processing throughputs over 800k/s using a single, powerful machine.

Logarithmic search data structures, such as search trees and skiplists, are fundamental building blocks of many applications. Although the self-balancing binary search trees are among the most ubiquitous sequential search data structures, designing non-blocking rebalancing algorithms is challenging due to the required structural alternation, which may stall other concurrent operations. Skiplists, which probabilistically create multiple levels of shortcuts in an ordered list, provide practical alternatives to balanced search trees. The use of skiplists eliminates the need of rebalancing and ensures amortized logarithmic sequential search time, but concurrency is limited under write-dominated workload because the linkage between multiple distant nodes must be updated. In this paper, we present a linearizable lock-free dictionary design based on a multi-dimensional list (MDList). A node in an MDList arranges its child nodes by their dimensionality and order them by coordinate prefixes. The search operation works by first generating a one-to-one mapping from the scalar keys to a high-dimensional vectors space, then uniquely locating the target position by using the vector as coordinates. Our algorithm guarantees worst-case search time of O(log N) where N is the size of key space. Moreover, the ordering property of the data structure is readily maintained during mutations without rebalancing nor randomization. In our experimental evaluation using a micro-benchmark, our dictionary outperforms the state of the art approaches by as much as 100% when the key universe is large and an average of 30% across all scenarios.

Temporal abstraction and data mining are two research fields that have tried to synthesis time oriented data and bring out an understanding on the hidden relationships that may exist between time oriented events. In clinical settings, having the ability to know the hidden relationships on patient data as they unfold could help save a life by aiding in detection of conditions that are not obvious to clinicians and healthcare workers. Understanding the hidden patterns is a huge challenge due to the exponential search space unique to time-series data. In this paper, we propose a temporal pattern recognition model based on dimension reduction and similarity measures thereby maintaining the temporal nature of the raw data.

Authors present the concept of image mining, an extension of data mining for discovering semantically meaningful information and image data relationship from a large collection of images. Association rule mining is the process of discovering useful and interesting rules, representing frequent patterns from large datasets, depends on user specified minimum support and confidence values. These constraints lead to exponential search space and dataset dependent minimum support and confidence values. The authors propose an optimization technique for overcoming these problems using multi-fitness function Genetic algorithm and constrained nonlinear minimization and minimax optimization method. Synthetic image set containing geometric shapes and standard MIAS medical image dataset are used to validate the proposed optimization algorithm. Experimental results show that, Genetic algorithm generates more efficient, effective and strong association rules than constrained nonlinear minimization and minimax optimization method. Genetic algorithm achieves 50% and 90%, constrained nonlinear minimization and minimax optimization method achieves 22% and 74 %, reduction in association rules for synthetic image set and standard MIAS medical image dataset respectively.

Reconstructing the topology of a signaling network by means of RNA interference (RNAi) technology is an underdetermined problem especially when a single gene in the network is knocked down or observed. In addition, the exponential search space limits the existing methods to small signaling networks of size 10-15 genes. In this paper, we propose integrating RNAi data with a reference physical interaction network. We formulate the problem of signaling network reconstruction as finding the minimum number of edit operations on a given reference network. The edit operations transform the reference network to a network that satisfies the RNAi observations. We show that using a reference network does not simplify the computational complexity of the problem. Therefore, we propose two methods which provide near optimal results and can scale well for reconstructing networks up to hundreds of components. We validate the proposed methods on synthetic and real data sets. Comparison with the state of the art on real signaling networks shows that the proposed methodology can scale better and generates biologically significant results.

system is a computability model which is biochemically inspired, it is a general distributed model, highly parallel, nondeterministic, based on the notion of a membrane structure. Till this moment, there is no exact idea about the real implementation of P systems. P systems are used in solving NP-complete problems in polynomial time, but with building the whole exponential search space. Cloud computing assume infinite memory and infinite processing power. This paper proposes an algorithm that uses the cloud resources in a fully parallel manner as a step towards P systems implementation, the nondeterminism property of P systems is certainly not maintained. The paper used the SAT problem as the case study.

Major challenges in mining high utility itemsets from the transaction databases requires exponential search space and database-dependent minimum utility threshold. The search space is very large because of the large number of distinct items and size of the database. Data analysts need to specifying appropriate minimum utility thresholds for their data mining tasks though they may have no knowledge pertaining to their databases. To get rid of these problems, Stellar mass black hole optimisation (SBO) method is proposed to mine Top-K HUIs from the transaction database without specifying minimum utility threshold. To know the performance of SBO, the experiment results are compared with GA.

A fundamental issue in Web search is ranking search results based on user logs, since different users may have different preferences and intents with regards to a search query. Also, in many search query applications, users tend to look at only the top part of the ranked result list in order to find relevant documents. The setting we consider contains various types of users, each of which is interested in a subset of the search results. The goal is to rank the search results of a query providing highly ranked relevant results. Our performance measure is the discounted cumulative gain which offers a graded relevance scale of documents in a search engine result set, and measures the usefulness (gain) of a document based on its position in the result list. Based on this measure, we suggest a general approach to developing approximation algorithms for ranking search results that captures different aspects of users’ intents. We also take into account that the relevance of one document cannot be treated independently of the relevance of other documents in a collection returned by a search engine. We first consider the scenario where users are interested in only a single search result (e.g., navigational queries). We then develop a polynomial time approximation scheme for this case. We further consider the general case where users have different requirements on the number of search results, and develop efficient approximation algorithms. Finally, we consider the problem of choosing the top k out of n search results and show that for this problem (1¿-¿1/e) is indeed the best approximation factor achievable, thus separating the approximability of the two versions of the problem.

This is an algorithm which has the same time complexity as that of linear search of “O (n)”. But still it is better than “linear search” in terms of execution time. Let A[ ] be the array of some size N. If the element which we want to search is at any position before “N/2” than “my-search and linear-search” both will have execution time, but the magic happens when the search element is after “N/2” position. Suppose the element want to search is at nth position, then using the linear search will find the element after nth iteration, but using “my-search” we can search the element after 1st iteration itself. Elements in (N-i)th position can be found in the (I+1)th iteration i.e. suppose size is 1000 than element in 1000th position can be found in 1st iteration, similarly 999 in 2nd iteration and process goes on like this.

Part 1: The International Cross Domain Conference (CD-ARES 2016); International audience; Most Machine Learning (ML) researchers focus on automatic Machine Learning (aML) where great advances have been made, for example, in speech recognition, recommender systems, or autonomous vehicles. Automatic approaches greatly benefit from the availability of “big data”. However, sometimes, for example in health informatics, we are confronted not a small number of data sets or rare events, and with complex problems where aML-approaches fail or deliver unsatisfactory results. Here, interactive Machine Learning (iML) may be of help and the “human-in-the-loop” approach may be beneficial in solving computationally hard problems, where human expertise can help to reduce an exponential search space through heuristics.In this paper, experiments are discussed which help to evaluate the effectiveness of the iML-“human-in-the-loop” approach, particularly in opening the “black box”, thereby enabling a human to directly and indirectly manipulating and interacting with an algorithm. For this purpose, we selected the Ant Colony Optimization (ACO) framework, and use it on the Traveling Salesman Problem (TSP) which is of high importance in solving many practical problems in health informatics, e.g. in the study of proteins.

Our new algorithm, priority search, was created with the help of skip list data structure and algorithms. Skip list data structure consists of linked lists formed in layers, which were linked in a pyramidal way. The time complexity of searching algorithm is equal to O(lgN) in an N-element skip list data structure. The new developed searching algorithm was based on the hit search number for each searched data. If a datum has greater hit search number, then it was upgraded in the skip list data structure to the upper level. That is, the mostly searched data were located in the upper levels of the skip list data structure and rarely searched data were located in the lower levels of the skip list data structure. The pyramidal structure of data was constructed by using the hit search numbers, in another word, frequency of each data. Thus, the time complexity of searching was almost ?(1) for N records data set. In this paper, the applications of searching algorithms like linear search, binary search, and priority search were realized, and the obtained results were compared. The results demonstrated that priority search algorithm was better than the binary search algorithm.

The paper introduces an AND/OR search space perspective for graphical models that include probabilistic networks (directed or undirected) and constraint networks. In contrast to the traditional (OR) search space view, the AND/OR search tree displays some of the independencies present in the graphical model explicitly and may sometimes reduce the search space exponentially. Indeed, most algorithmic advances in search-based constraint processing and probabilistic inference can be viewed as searching an AND/OR search tree or graph. Familiar parameters such as the depth of a spanning tree, treewidth and pathwidth are shown to play a key role in characterizing the effect of AND/OR search graphs vs. the traditional OR search graphs. We compare memory intensive AND/OR graph search with inference methods, and place various existing algorithms within the AND/OR search space.

In pattern recognition field, there are many optimization problems having exponential search spaces. To solve of sequential search algorithms seeking sub-optimal solutions have been used. The algorithms have limitations of stopping at local optimums. Recently lots of researches attempt to solve the problems using genetic algorithms. This paper explains the huge search spaces of typical problems such as feature selection, classifier ensemble selection, neural network pruning, and clustering, and it reviews the genetic algorithms for solving them. Additionally we present several subjects worthy of noting as future researches.

Human actions are, inherently, structured patterns of body movements. We explore ensembles of hierarchical spatio-temporal trees, discovered directly from training data, to model these structures for action recognition. The hierarchical spatio-temporal trees provide a robust mid-level representation for actions. However, discovery of frequent and discriminative tree structures is challenging due to the exponential search space, particularly if one allows partial matching. We address this by first building a concise action vocabulary via discriminative clustering. Using the action vocabulary we then utilize tree mining with subsequent tree clustering and ranking to select a compact set of highly discriminative tree patterns. We show that these tree patterns, alone, or in combination with shorter patterns (action words and pairwise patterns) achieve state-of-the-art performance on two challenging datasets: UCF Sports and HighFive. Moreover, trees learned on HighFive are used in recognizing two action classes in a different dataset, Hollywood3D, demonstrating the potential for cross-dataset generality of the trees our approach discovers.

The critical resource that limits the application of best-first search is memory. We present a new class of best-first search algorithms that reduce the space complexity. The key idea is to store only the Open list of generated nodes, but not the Closed list of expanded nodes. The solution path can be recovered by a divide-and-conquer technique, either as a bidirectional or unidirectional search. For many problems, frontier search dramatically reduces the memory required by best-first search. We apply frontier search to breadth-first search of sliding-tile puzzles and the 4-peg Towers of Hanoi problem, Dijkstra's algorithm on a grid with random edge costs, and the A* algorithm on the Fifteen Puzzle, the four-peg Towers of Hanoi Problem, and optimal sequence alignment in computational biology.