Search results for: weighted search

Authors: Gemma García López

Abstract:

The goal of this study is to analyze if search queries carried out in search engines such as Google, can offer emotional information about the user that performs them. Knowing the emotional state in which the Internet user is located can be a key to achieve the maximum personalization of content and the detection of worrying behaviors. For this, two studies were carried out using tools with advanced natural language processing techniques. The first study determines if a query can be classified as positive, negative or neutral, while the second study extracts emotional content from words and applies the categorical and dimensional models for the representation of emotions. In addition, we use search queries in Spanish and English to establish similarities and differences between two languages. The results revealed that text search queries performed by users on the Internet can be classified emotionally. This allows us to better understand the emotional state of the user at the time of the search, which could involve adapting the technology and personalizing the responses to different emotional states.

Keywords: emotion classification, text search queries, emotional analysis, sentiment analysis in text, natural language processing

Procedia PDF Downloads 115

Authors: Kang-Mo Jung

Abstract:

The use of regularization for statistical methods has become popular. The least absolute shrinkage and selection operator (LASSO) framework has become the standard tool for sparse regression. However, it is well known that the LASSO is sensitive to outliers or leverage points. We consider a new robust estimation which is composed of the weighted loss function of the pairwise difference of residuals and the adaptive penalty function regulating the tuning parameter for each variable. Rank regression is resistant to regression outliers, but not to leverage points. By adopting a weighted loss function, the proposed method is robust to leverage points of the predictor variable. Furthermore, the adaptive penalty function gives us good statistical properties in variable selection such as oracle property and consistency. We develop an efficient algorithm to compute the proposed estimator using basic functions in program R. We used an optimal tuning parameter based on the Bayesian information criterion (BIC). Numerical simulation shows that the proposed estimator is effective for analyzing real data set and contaminated data.

Keywords: adaptive penalty function, robust penalized regression, variable selection, weighted rank regression

Procedia PDF Downloads 431

Authors: Mohammed Belloufi, Rachid Benzine, Badreddine Sellami

Abstract:

Conjugate gradient methods is useful for solving large scale optimization problems in scientific and engineering computation, characterized by the simplicity of their iteration and their low memory requirements. It is well known that the search direction plays a main role in the line search method. In this paper, we propose a search direction with the Wolfe line search technique for solving unconstrained optimization problems. Under the above line searches and some assumptions, the global convergence properties of the given methods are discussed. Numerical results and comparisons with other CG methods are given.

Keywords: unconstrained optimization, conjugate gradient method, strong Wolfe line search, global convergence

Procedia PDF Downloads 385

Authors: Fatma Abdedayem

Abstract:

The most effective and efficient methods in image categorization are almost based on bag-of-words (BOW) which presents image by a histogram of occurrence of visual words. In this paper, we propose a novel extension to this method. Firstly, we extract features in multi-scales by applying a color local descriptor named opponent-SIFT. Secondly, in order to represent image we use Spatial Pyramid Representation (SPR) and an extension to the BOW method which based on weighting visual words. Typically, the visual words are weighted during histogram assignment by computing the ratio of their occurrences in the image to the occurrences in the background. Finally, according to classical BOW retrieval framework, only a few words of the vocabulary is useful for image representation. Therefore, we select the useful weighted visual words that respect the threshold value. Experimentally, the algorithm is tested by using different image classes of PASCAL VOC 2007 and is compared against the classical bag-of-visual-words algorithm.

Keywords: BOW, useful visual words, weighted visual words, bag of visual words

Procedia PDF Downloads 410

Authors: Nicanor Garcia Alvarez, Belarmino Adenso-Diaz, Laura Calzada Infante

Abstract:

In recent years, researchers have shown an interest in modelling maritime traffic as a complex network. In this paper, we propose a bipartite weighted network to model maritime traffic and detect port maritime communities. The bipartite weighted network considers two different types of nodes. The first one represents Spanish ports, while the second one represents the countries with which there is major import/export activity. The flow among both types of nodes is modeled by weighting the volume of product transported. To illustrate the model, the data is segmented by each type of traffic. This will allow fine tuning and the creation of communities for each type of traffic and therefore finding similar ports for a specific type of traffic, which will provide decision-makers with tools to search for alliances or identify their competitors. The traffic with the greatest impact on the Spanish gross domestic product is selected, and the evolution of the communities formed by the most important ports and their differences between 2019 and 2009 will be analyzed. Finally, the set of communities formed by the ports of the Spanish port system will be inspected to determine global similarities between them, analyzing the sum of the membership of the different ports in communities formed for each type of traffic in particular.

Keywords: bipartite networks, competition, infomap, maritime traffic, port communities

Procedia PDF Downloads 116

Authors: Laicheng Cao, Xiangqian Su, Youxiao Wu

Abstract:

Feature selection is one of the important problems in network security, pattern recognition, data mining and other fields. In order to remove redundant features, effectively improve the detection speed of intrusion detection system, proposes a new feature selection method, which is based on the invasive weed optimization (IWO) algorithm and tabu search algorithm(TS). Use IWO as a global search, tabu search algorithm for local search, to improve the results of IWO algorithm. The experimental results show that the feature selection method can effectively remove the redundant features of network data information in feature selection, reduction time, and to guarantee accurate detection rate, effectively improve the speed of detection system.

Keywords: intrusion detection, feature selection, iwo, tabu search

Procedia PDF Downloads 498

Authors: Mohammed M. Siddeq, Mohammed H. Rasheed, Omar M. Salih, Marcos A. Rodrigues

Abstract:

This research proposes a data encoding and decoding method based on the Matrix Minimization algorithm. This algorithm is applied to high-frequency coefficients for compression/encoding. The algorithm starts by converting every three coefficients to a single value; this is accomplished based on three different keys. The decoding/decompression uses a search method called QSS (Quick Sequential Search) Decoding Algorithm presented in this research based on the sequential search to recover the exact coefficients. In the next step, the decoded data are saved in an auxiliary array. The basic idea behind the auxiliary array is to save all possible decoded coefficients; this is because another algorithm, such as conventional sequential search, could retrieve encoded/compressed data independently from the proposed algorithm. The experimental results showed that our proposed decoding algorithm retrieves original data faster than conventional sequential search algorithms.

Keywords: matrix minimization algorithm, decoding sequential search algorithm, image compression, DCT, DWT

Procedia PDF Downloads 112

Authors: Meng-Hui Chen, Chiao-Wei Yu, Pei-Chann Chang

Abstract:

Imperial competitive algorithm (ICA) simulates a multi-agent algorithm. Each agent is like a kingdom has its country, and the strongest country in each agent is called imperialist, others are colony. Countries are competitive with imperialist which in the same kingdom by evolving. So this country will move in the search space to find better solutions with higher fitness to be a new imperialist. The main idea in this paper is using the peculiarity of ICA to explore the search space to solve the kinds of combinational problems. Otherwise, we also study to use the greed search to increase the local search ability. To verify the proposed algorithm in this paper, the experimental results of traveling salesman problem (TSP) is according to the traveling salesman problem library (TSPLIB). The results show that the proposed algorithm has higher performance than the other known methods.

Keywords: traveling salesman problem, artificial chromosomes, greedy search, imperial competitive algorithm

Procedia PDF Downloads 425

Authors: Seema Vaidya

Abstract:

Mining association rule is a key issue in data mining. In any case, the standard models ignore the distinction among the exchanges, and the weighted association rule mining does not transform on databases with just binary attributes. This paper proposes a novel continuous example and executes a tree (FP-tree) structure, which is an increased prefix-tree structure for securing compacted, discriminating data about examples, and makes a fit FP-tree-based mining system, FP enhanced capacity algorithm is used, for mining the complete game plan of examples by illustration incessant development. Here, this paper handles the motivation behind making remarkable and weighted item sets, i.e. rare weighted item set mining issue. The two novel brightness measures are proposed for figuring the infrequent weighted item set mining issue. Also, the algorithm are handled which perform IWI which is more insignificant IWI mining. Moreover we utilized the rare item set for choice based structure. The general issue of the start of reliable definite rules is troublesome for the grounds that hypothetically no inciting technique with no other person can promise the rightness of influenced theories. In this way, this framework expects the disorder with the uncommon signs. Usage study demonstrates that proposed algorithm upgrades the structure which is successful and versatile for mining both long and short diagnostics rules. Structure upgrades aftereffects of foreseeing rare diseases of patient.

Keywords: association rule, data mining, IWI mining, infrequent item set, frequent pattern growth

Procedia PDF Downloads 377

Authors: I. Nižetić Kosović, T. Jagušt

Abstract:

Lately, with the increasing number of location-based applications, demand for highly accurate and reliable indoor localization became urgent. This is a challenging problem, due to the measurement variance which is the consequence of various factors like obstacles, equipment properties and environmental changes in complex nature of indoor environments. In this paper we propose low-cost custom-setup infrastructure solution and localization algorithm based on the Weighted Centroid Localization (WCL) method. Localization accuracy is increased by several enhancements: calibration of RSSI values gained from wireless nodes, repetitive measurements of RSSI to exclude deviating values from the position estimation, and by considering orientation of the device according to the wireless nodes. We conducted several experiments to evaluate the proposed algorithm. High accuracy of ~1m was achieved.

Keywords: indoor environment, received signal strength indicator, weighted centroid localization, wireless localization

Procedia PDF Downloads 208

Authors: Feng Tao, Ma Jing, Guo Xian, Wang Jing

Abstract:

Named Data Networking (NDN) replaces IP address of traditional network with data name, and adopts dynamic cache mechanism. In the existing mechanism, however, only one-to-one search can be achieved because every data has a unique name corresponding to it. There is a certain mapping relationship between data content and data name, so if the data name is intercepted by an adversary, the privacy of the data content and user’s interest can hardly be guaranteed. In order to solve this problem, this paper proposes a one-to-many fuzzy search scheme based on order-preserving encryption to reduce the query overhead by optimizing the caching strategy. In this scheme, we use hash value to ensure the user’s query safe from each node in the process of search, so does the privacy of the requiring data content.

Keywords: NDN, order-preserving encryption, fuzzy search, privacy

Procedia PDF Downloads 448

Authors: Khaled Moh. Alhamad

Abstract:

This paper describes a tabu search heuristic for a ship routing and scheduling problem (SRSP). The method was developed to address the problem of loading cargos for many customers using heterogeneous vessels. Constraints relate to delivery time windows imposed by customers, the time horizon by which all deliveries must be made and vessel capacities. The results of a computational investigation are presented. Solution quality and execution time are explored with respect to problem size and parameters controlling the tabu search such as tenure and neighbourhood size.

Keywords: heuristic, scheduling, tabu search, transportation

Procedia PDF Downloads 478

Authors: Tapas Acharya, Monalisa Mitra

Abstract:

Overlay modeling is the most widely used conventional analysis for spatial decision support system. Overlay modeling requires a set of themes with different weightage computed in varied manners, which gives a resultant input for further integrated analysis. In spite of the popularity and most widely used technique; it gives inconsistent and erroneous results for similar inputs while processed in various GIS overlay techniques. This study is an attempt to compare and analyse the differences in the outputs of different overlay methods using GIS platform with same set of themes of the Precambrian metamorphic to obtain groundwater prospecting in Precambrian metamorphic rocks. The objective of the study is to emphasize the most suitable overlay method for groundwater prospecting in older Precambrian metamorphics. Seven input thematic layers like slope, Digital Elevation Model (DEM), soil thickness, lineament intersection density, average groundwater table fluctuation, stream density and lithology have been used in the spatial overlay models of fuzzy overlay, weighted overlay and weighted sum overlay methods to yield the suitable groundwater prospective zones. Spatial concurrence analysis with high yielding wells of the study area and the statistical comparative studies among the outputs of various overlay models using RStudio reveal that the Weighted Overlay model is the most efficient GIS overlay model to delineate the groundwater prospecting zones in the Precambrian metamorphic rocks.

Keywords: fuzzy overlay, GIS overlay model, groundwater prospecting, Precambrian metamorphics, weighted overlay, weighted sum overlay

Procedia PDF Downloads 99

Authors: Aysun Sezer

Abstract:

Magnetic Resonance Imaging (MRI) is one of the advanced diagnostic tools for evaluating shoulder pathologies. Proton Density (PD)-weighted MRI sequences prove highly effective in detecting edema. However, they are deficient in the anatomical identification of bones due to a trauma-induced decrease in signal-to-noise ratio and blur in the traumatized cortices. Computer-based diagnostic systems require precise segmentation, identification, and localization of anatomical regions in medical imagery. Deep learning-based object detection algorithms exhibit remarkable proficiency in real-time object identification and localization. In this study, the YOLOv8 model was employed to detect humeral head and scapular regions in 665 axial PD-weighted MR images. The YOLOv8 configuration achieved an overall success rate of 99.60% and 89.90% for detecting the humeral head and scapula, respectively, with an intersection over union (IoU) of 0.5. Our findings indicate a significant promise of employing YOLOv8-based detection for the humerus and scapula regions, particularly in the context of PD-weighted images affected by both noise and intensity inhomogeneity.

Keywords: YOLOv8, object detection, humerus, scapula, IRM

Procedia PDF Downloads 35

Authors: Rutuja Desai

Abstract:

Search engine is fundamentally a framework used to search the data which is pertinent to the client via WWW. Looking close-by spot identified with the keywords is an imperative concept in developing web advances. For such kind of searching, extent pursuit or closest neighbor is utilized. In range search the forecast is made whether the objects meet to query object. Nearest neighbor is the forecast of the focuses close to the query set by the client. Here, the nearest neighbor methodology is utilized where Data recovery R+ tree is utilized rather than IR2 tree. The disadvantages of IR2 tree is: The false hit number can surpass the limit and the mark in Information Retrieval R-tree must have Voice over IP bit for each one of a kind word in W set is recouped by Data recovery R+ tree. The inquiry is fundamentally subordinate upon the key words and the geometric directions.

Keywords: information retrieval, nearest neighbor search, keyword search, R+ tree

Procedia PDF Downloads 261

Authors: Parag Bhalchandra, S. D. Khamitkar

Abstract:

Searching trees is a most interesting application of Artificial Intelligence. Over the period of time, many innovative methods have been evolved to better search trees with respect to computational complexities. Tree searches are difficult to understand due to the exponential growth of possibilities when increasing the number of nodes or levels in the tree. Usually it is understood when we traverse down in the tree, traverse down to greater depth, in the search of a solution or a goal. However, this does not happen in reality as explicit enumeration is not a very efficient method and there are many algorithmic speedups that will find the optimal solution without the burden of evaluating all possible trees. It was a common question before all researchers where they often wonder what algorithms will yield the best and fastest result The intention of this paper is two folds, one to review selected tree search algorithms and search strategies that can be applied to a problem space and the second objective is to stimulate to implement recent developments in the complexity behavior of search strategies. The algorithms discussed here apply in general to both brute force and heuristic searches.

Keywords: trees search, asymptotic complexity, brute force, heuristics algorithms

Procedia PDF Downloads 282

Authors: Noha Seddik, Sherine Youssef, Mohamed Kholeif

Abstract:

The automated epileptic seizure detection research field has emerged in the recent years; this involves analyzing the Electroencephalogram (EEG) signals instead of the traditional visual inspection performed by expert neurologists. In this study, a Support Vector Machine (SVM) that uses Weighted Permutation Entropy (WPE) as the input feature is proposed for classifying normal and seizure EEG records. WPE is a modified statistical parameter of the permutation entropy (PE) that measures the complexity and irregularity of a time series. It incorporates both the mapped ordinal pattern of the time series and the information contained in the amplitude of its sample points. The proposed system utilizes the fact that entropy based measures for the EEG segments during epileptic seizure are lower than in normal EEG.

Keywords: electroencephalogram (EEG), epileptic seizure detection, weighted permutation entropy (WPE), support vector machine (SVM)

Procedia PDF Downloads 341

Authors: Ebert Brea

Abstract:

We propose a novel direct search algorithm for identifying at least a local minimum of mixed integer nonlinear unconstrained optimization problems. The Mixed Integer Randomized Pattern Search Algorithm (MIRPSA), so-called by the author, is based on a randomized pattern search, which is modified by the MIRPSA for finding at least a local minimum of our problem. The MIRPSA has two main operations over the randomized pattern search: moving operation and shrinking operation. Each operation is carried out by the algorithm when a set of conditions is held. The convergence properties of the MIRPSA is analyzed using a Markov chain approach, which is represented by an infinite countable set of state space λ, where each state d(q) is defined by a measure of the qth randomized pattern search Hq, for all q in N. According to the algorithm, when a moving operation is carried out on the qth randomized pattern search Hq, the MIRPSA holds its state. Meanwhile, if the MIRPSA carries out a shrinking operation over the qth randomized pattern search Hq, the algorithm will visit the next state, this is, a shrinking operation at the qth state causes a changing of the qth state into (q+1)th state. It is worthwhile pointing out that the MIRPSA never goes back to any visited states because the MIRPSA only visits any qth by shrinking operations. In this article, we describe the MIRPSA for mixed integer nonlinear unconstrained optimization problems for doing a deep study of its convergence properties using Markov chain viewpoint. We herein include a low dimension case for showing more details of the MIRPSA, when the algorithm is used for identifying the minimum of a mixed integer quadratic function. Besides, numerical examples are also shown in order to measure the performance of the MIRPSA.

Keywords: direct search, mixed integer optimization, random search, convergence, Markov chain

Procedia PDF Downloads 434

Authors: Pakinee Aimmanee, Phoom Chokratsamesiri

Abstract:

Search engine has become an essential tool nowadays for people to search for their needed information on the internet. In this work, we evaluate the performance of the search engine from three factors: the keyword frequency, the number of inbound links, and the difficulty of the keyword. The evaluations are based on the ranking position and the number of days that Google has seen or detect the webpage. We find that the keyword frequency and the difficulty of the keyword do not affect the Google ranking where the number of inbound links gives remarkable improvement of the ranking position. The optimal number of inbound links found in the experiment is 10.

Keywords: SEO, information retrieval, web search, knowledge technologies

Procedia PDF Downloads 259

Authors: Daniel Panario, Daniel Santana de Freitas, Brett Stevens

Abstract:

Almost Perfect Nonlinear (APN) permutations with optimal resistance against differential cryptanalysis can be found in several domains. The permutation used in the standard for symmetric cryptography (the AES), for example, is based on a special kind of inversion in GF(28). Although very close to APN (2-uniform), this permutation still contains one number 4 in its differential spectrum, which means that, rigorously, it must be classified as 4-uniform. This fact motivates the search for fully APN permutations in other domains of definition. The extremely high complexity associated to this kind of problem precludes an exhaustive search for an APN permutation with 256 elements to be performed without the support of a suitable mathematical structure. On the other hand, in principle, there is nothing to indicate which mathematically structured domains can effectively help the search, and it is necessary to test several domains. In this work, the search for APN permutations in rings ℤ2×ℤ2k is investigated. After a full, exhaustive search with k=2 and k=3, all possible APN permutations in those rings were recorded, together with their differential profiles. Some very promising heuristics in these cases were collected so that, when used as a basis to prune backtracking for the same search in ℤ2×ℤ8 (search space with size 16! ≅244), just a few tenths of a second were enough to produce an APN permutation in a single CPU. Those heuristics were empirically extrapolated so that they could be applied to a backtracking search for APNs over ℤ2×ℤ16 (search space with size 32! ≅2117). The best permutations found in this search were further refined through Simulated Annealing, with a definition of neighbors suitable to this domain. The best result produced with this scheme was a 3-uniform permutation over ℤ2×ℤ16 with only 24 values equal to 3 in the differential spectrum (all the other 968 values were less than or equal 2, as it should be the case for an APN permutation). Although far from being fully APN, this result is technically better than a 4-uniform permutation and demanded only a few seconds in a single CPU. This is a strong indication that the use of mathematically structured domains, like the rings described in this work, together with heuristics based on smaller cases, can lead to dramatic cuts in the computational resources involved in the complexity of the search for APN permutations in extremely large domains.

Keywords: APN permutations, heuristic searches, symmetric cryptography, S-box design

Procedia PDF Downloads 129

Authors: Ariful Islam, Showkat Ahmad Lone

Abstract:

The Mukherjee-Islam Model is commonly used as a simple life time distribution to assess system reliability. The model exhibits a better fit for failure information and provides more appropriate information about hazard rate and other reliability measures as shown by various authors. It is possible to introduce a location parameter at a time (i.e., a time before which failure cannot occur) which makes it a more useful failure distribution than the existing ones. Even after shifting the location of the distribution, it represents a decreasing, constant and increasing failure rate. It has been shown to represent the appropriate lower tail of the distribution of random variables having fixed lower bound. This study presents the reliability computations and probability weighted moment estimation of three parameter model. A comparative analysis is carried out between three parameters finite range model and some existing bathtub shaped curve fitting models. Since probability weighted moment method is used, the results obtained can also be applied on small sample cases. Maximum likelihood estimation method is also applied in this study.

Keywords: comparative analysis, maximum likelihood estimation, Mukherjee-Islam failure model, probability weighted moment estimation, reliability

Procedia PDF Downloads 244

Authors: S. M. Kulkarni, D. S. Bormane, S. L. Nalbalwar

Abstract:

The rapid development in the technology have dramatic impact on the medical health care field. Medical data base obtained with latest machines like CT Machine, MRI scanner requires large amount of memory storage and also it requires large bandwidth for transmission of data in telemedicine applications. Thus, there is need for video compression. As the database of medical images contain number of frames (slices), hence while coding of these images there is need of motion estimation. Motion estimation finds out movement of objects in an image sequence and gets motion vectors which represents estimated motion of object in the frame. In order to reduce temporal redundancy between successive frames of video sequence, motion compensation is preformed. In this paper three step search (TSS) block matching algorithm is implemented on different types of video sequences. It is shown that three step search algorithm produces better quality performance and less computational time compared with exhaustive full search algorithm.

Keywords: block matching, exhaustive search motion estimation, three step search, video compression

Procedia PDF Downloads 455

Authors: Yao-Hong Tsai

Abstract:

This paper presents a weighted approach to unconstrained iris recognition. Nowadays, commercial systems are usually characterized by strong acquisition constraints based on the subject’s cooperation. However, it is not always achievable for real scenarios in our daily life. Researchers have been focused on reducing these constraints and maintaining the performance of the system by new techniques at the same time. With large variation in the environment, there are two main improvements to develop the proposed iris recognition system. For solving extremely uneven lighting condition, statistic based illumination normalization is first used on eye region to increase the accuracy of iris feature. The detection of the iris image is based on Adaboost algorithm. Secondly, the weighted approach is designed by Gaussian functions according to the distance to the center of the iris. Furthermore, local binary pattern (LBP) histogram is then applied to texture classification with the weight. Experiment showed that the proposed system provided users a more flexible and feasible way to interact with the verification system through iris recognition.

Keywords: authentication, iris recognition, adaboost, local binary pattern

Procedia PDF Downloads 192

Authors: Derya Karagöz

Abstract:

A Shewhart chart based on normality assumption is not appropriate for skewed distributions since its Type-I error rate is inflated. This study presents X̄ and S control charts for monitoring the process variability for skewed distributions. We propose Weighted Standard Deviation (WSD) X̄ and S control charts. Standard deviation estimator is applied to monitor the process variability for estimating the process standard deviation, in the case of the W SD X̄ and S control charts as this estimator is simple and easy to compute. Unlike the Shewhart control chart, the proposed charts provide asymmetric limits in accordance with the direction and degree of skewness to construct the upper and lower limits. The performances of the proposed charts are compared with other heuristic charts for skewed distributions by using Simulation study. The Simulation studies show that the proposed control charts have good properties for skewed distributions and large sample sizes.

Keywords: weighted standard deviation, MAD, skewed distributions, S control charts

Procedia PDF Downloads 368

Authors: Watcharapan Sukkerd, Teeradej Wuttipornpun

Abstract:

This paper aims to present non-population search algorithms called tabu search (TS), simulated annealing (SA) and variable neighborhood search (VNS) to minimize the total cost of capacitated MRP problem in multi-stage assembly flow shop with two alternative machines. There are three main steps for the algorithm. Firstly, an initial sequence of orders is constructed by a simple due date-based dispatching rule. Secondly, the sequence of orders is repeatedly improved to reduce the total cost by applying TS, SA and VNS separately. Finally, the total cost is further reduced by optimizing the start time of each operation using the linear programming (LP) model. Parameters of the algorithm are tuned by using real data from automotive companies. The result shows that VNS significantly outperforms TS, SA and the existing algorithm.

Keywords: capacitated MRP, tabu search, simulated annealing, variable neighborhood search, linear programming, assembly flow shop, application in industry

Procedia PDF Downloads 211

Authors: Afroze Ibrahim Baqapuri, Ilya Trofimov

Abstract:

Sponsored search is a multi-billion dollar industry and makes up a major source of revenue for search engines (SE). Click-through-rate (CTR) estimation plays a crucial role for ads selection, and greatly affects the SE revenue, advertiser traffic and user experience. We propose a novel architecture of solving CTR prediction problem by combining artificial neural networks (ANN) with decision trees. First, we compare ANN with respect to other popular machine learning models being used for this task. Then we go on to combine ANN with MatrixNet (proprietary implementation of boosted trees) and evaluate the performance of the system as a whole. The results show that our approach provides a significant improvement over existing models.

Keywords: neural networks, sponsored search, web advertisement, click prediction, click-through rate

Procedia PDF Downloads 547

Authors: Soha A. Bahanshal, Byung G. Kim

Abstract:

Identification of patients at high risk for hospital readmission is of crucial importance for quality health care and cost reduction. Predicting hospital readmissions among diabetic patients has been of great interest to many researchers and health decision makers. We build a prediction model to predict hospital readmission for diabetic patients within 30 days of discharge. The core of the prediction model is a modified k Nearest Neighbor called Hybrid Fuzzy Weighted k Nearest Neighbor algorithm. The prediction is performed on a patient dataset which consists of more than 70,000 patients with 50 attributes. We applied data preprocessing using different techniques in order to handle data imbalance and to fuzzify the data to suit the prediction algorithm. The model so far achieved classification accuracy of 80% compared to other models that only use k Nearest Neighbor.

Keywords: machine learning, prediction, classification, hybrid fuzzy weighted k-nearest neighbor, diabetic hospital readmission

Procedia PDF Downloads 156

Authors: Doyin Afolabi, Phillip Adewole, Oladipupo Sennaike

Abstract:

Most well-known classifiers, including the decision tree algorithm, can make predictions on balanced datasets efficiently. However, the decision tree algorithm tends to be biased towards imbalanced datasets because of the skewness of the distribution of such datasets. To overcome this problem, this study proposes a weighted decision tree algorithm that aims to remove the bias toward the majority class and prevents the reduction of majority observations in imbalance datasets classification. The proposed weighted decision tree algorithm was tested on three imbalanced datasets- cancer dataset, german credit dataset, and banknote dataset. The specificity, sensitivity, and accuracy metrics were used to evaluate the performance of the proposed decision tree algorithm on the datasets. The evaluation results show that for some of the weights of our proposed decision tree, the specificity, sensitivity, and accuracy metrics gave better results compared to that of the ID3 decision tree and decision tree induced with minority entropy for all three datasets.

Keywords: data mining, decision tree, classification, imbalance dataset

Procedia PDF Downloads 93

Authors: Phong Nguyen, Phap Nguyen, Thang Nguyen

Abstract:

High Efficiency Video Coding (HEVC) or H.265 Standard fulfills the demand of high resolution video storage and transmission since it achieves high compression ratio. However, it requires a huge amount of calculation. Since Motion Estimation (ME) block composes about 80 % of calculation load of HEVC, there are a lot of researches to reduce the computation cost. In this paper, we propose a new algorithm to lower the number of Motion Estimation’s searching points. The number of computing points in search pattern is down from 77 for Diamond Pattern and 81 for Square Pattern to only 31. Meanwhile, the Peak Signal to Noise Ratio (PSNR) and bit rate are almost equal to those of conventional patterns. The motion estimation time of new algorithm reduces by at 68.23%, 65.83%compared to the recommended search pattern of diamond pattern, square pattern, respectively.

Keywords: motion estimation, wide diamond, search pattern, H.265, test zone search, HM software

Procedia PDF Downloads 565

Authors: Ian A. McCulloh, Madelaine Zinser, Jesse Patsolic, Michael Ramos

Abstract:

Recommender systems are a common artificial intelligence (AI) application. For any given input, a search system will return a rank-ordered list of similar items. As users review returned items, they must decide when to halt the search and either revise search terms or conclude their requirement is novel with no similar items in the database. We present a statistically designed experiment that investigates the impact of similarity ratings on human judgement to conclude a search item is novel and halt the search. 450 participants were recruited from Amazon Mechanical Turk to render judgement across 12 decision tasks. We find the inclusion of ratings increases the human perception that items are novel. Percent similarity increases novelty discernment when compared with star-rated similarity or the absence of a rating. Ratings reduce the time to decide and improve decision confidence. This suggests the inclusion of similarity ratings can aid human decision-makers in knowledge search tasks.

Keywords: ratings, rankings, crowdsourcing, empirical studies, user studies, similarity measures, human-centered computing, novelty in information retrieval

Procedia PDF Downloads 91