Vector Quantization Method Research Articles

Android based rice pest detection system using learning vector quantization method

In Indonesia, rice was still an essential commodity. Aside from being a staple food producer of rice, as well as a primary source of livelihood for farmers. The problem often experienced by farmers was the threat of pest attack and cause massive losses; it could even lead to crop failures and harm the farmers. During this time to suppress the attacks of pest, farmers give pesticides which if it handles inappropriate would interfere with health. The best solution was to make detection system of pest attacks so could minimise the pest aggression. The limited knowledge and skills of farmers in the detection of rice crops made farmers prefer to spray pesticides. To facilitate farmers in detecting pest attacks, developed Android-based applications that were simple and easy to use device directly in the field using smartphone devices and Learning Vector Quantization (LVQ) Method. Farmers could use the application by taking pictures of rice plants that allegedly exposed to attacks with smartphones. The app would recommend alternative solutions to pest-infected plants without the use of pesticides.

Application of Digital Sound Recognition on Piano Tones Using MFCC and LVQ

In playing musical instruments especially the piano, a musician needs a score as the guidance to play the song. A score is a music sheet used by an artist to store or deliver a song. Many pianists do not have the expertise to create a score, especially in the making of spontaneous music. Therefore, an app is required to assist the pianists in the creation process of scores for piano music sounds. Mel-Frequency Cepstral Coefficient (MFCC) extraction method and Learning Vector Quantization method were implemented to build the application. The MFCC method was used to extract the vectors that reside in a song, while LVQ method was applied to match the testing data with the trained data. The output of this system is a musical score of the inputted song.

Perbandingan Metode Backpropagation dan Learning Vector Quantization (LVQ) Dalam Menggali Potensi Mahasiswa Baru di STMIK PalComTech

The Research aimst to compare backpropagation and Learning Vector Quantization (LVQ) methods in exploring the potential of new students at STMIK PalComTech. Comparisons in this study involve four input variables used which consist of four basic subjects of informatics engineering and information systems (math, basic programming, computer networks and management bases) which then make informatics techniques and information systems as outputs, to get the accuracy level high in this study, the researchers used several variations of parameters which eventually produced the best accuracy of the two methods. From 120 data tested using variations in test data and training data which are then processed using variations in the learning rate parameters and epochs. From the test results obtained the level of accuracy of pattern recognition in the backpropagation method is 99.17% with a learning rate variation of 0.1 and epoch 100, the learning vector quantization method has an accuracy rate of 96.67% with a variation of learning rate 1 and epoch 20 From the results of the comparison the Backpropagation method is superior in terms of accuracy so that it becomes the right method to use in exploring the potential of new students at STMIK PalComTech.

Vector and line quantization for billion-scale similarity search on GPUs

Billion-scale high-dimensional approximate nearest neighbour (ANN) search has become an important problem for searching similar objects among the vast amount of images and videos available online. The existing ANN methods are usually characterized by their specific indexing structures, including the inverted index and the inverted multi-index structure. The inverted index structure is amenable to GPU-based implementations, and the state-of-the-art systems such as Faiss are able to exploit the massive parallelism offered by GPUs. However, the inverted index requires high memory overhead to index the dataset effectively. The inverted multi-index structure is difficult to implement for GPUs, and also ineffective in dealing with database with different data distributions. In this paper we propose a novel hierarchical inverted index structure generated by vector and line quantization methods. Our quantization method improves both search efficiency and accuracy, while maintaining comparable memory consumption. This is achieved by reducing search space and increasing the number of indexed regions. We introduce a new ANN search system, VLQ-ADC, that is based on the proposed inverted index, and perform extensive evaluation on two public billion-scale benchmark datasets SIFT1B and DEEP1B. Our evaluation shows that VLQ-ADC significantly outperforms the state-of-the-art GPU- and CPU-based systems in terms of both accuracy and search speed. The source code of VLQ-ADC is available at https://github.com/zjuchenwei/vector-line-quantization.

Quarter-Point Product Quantization for approximate nearest neighbor search

Vector quantization can be used to achieve high accuracy for fast approximate nearest neighbor (ANN) search. Most existing vector quantization methods use codeword expansion to enhance the representative capacity of codebooks learned by clustering methods, which may decrease the quantization distortion and increase the accuracy. In this paper, we propose a novel codeword expansion method named Quarter-point Product Quantization (QPQ) to further minimize quantization distortions, by estimating the distances from the query points to the database points using the quarter points instead of the clustering centers. The distances can be computed more precisely and it will result in a lower distortion using QPQ, which is also a general method could be used to improve all PQ-based methods. Extensive experiments on several approximate nearest neighbor datasets show that our QPQ method can significantly improve the accuracy of PQ-based methods and consistently outperform other state-of-the-art methods.

Field scene recognition for self-localization of autonomous agricultural vehicle

In this study, a field scene recognition system was developed to estimate a self-position of a traveling vehicle along a farm road by using an original capture system with three cameras, a vector quantization method to express the features of field scenes, a machine learning based scene recognition algorithm, and a vehicle position estimation algorithm with an original voting method. The potential of our system was demonstrated through five experiments performed over four months. In the experiments, the system could robustly estimate the vehicle position with the accuracy less than 1 m at the processing speed of approximately 2.0 Hz when the vehicle was driven straight along a traveling line on the targeted two types of roads: a surfaced road and an unsurfaced road, at the driving speed of 0.5 m/s. The results demonstrated an applicability of our system to navigate an autonomous agricultural robot vehicle without using GNSS.

Vector quantization: a review

Vector quantization (VQ) is a very effective way to save bandwidth and storage for speech coding and image coding. Traditional vector quantization methods can be divided into mainly seven types, tree-structured VQ, direct sum VQ, Cartesian product VQ, lattice VQ, classified VQ, feedback VQ, and fuzzy VQ, according to their codebook generation procedures. Over the past decade, quantization-based approximate nearest neighbor (ANN) search has been developing very fast and many methods have emerged for searching images with binary codes in the memory for large-scale datasets. Their most impressive characteristics are the use of multiple codebooks. This leads to the appearance of two kinds of codebook: the linear combination codebook and the joint codebook. This may be a trend for the future. However, these methods are just finding a balance among speed, accuracy, and memory consumption for ANN search, and sometimes one of these three suffers. So, finding a vector quantization method that can strike a balance between speed and accuracy and consume moderately sized memory, is still a problem requiring study.

APLIKASI JARINGAN SYARAF TIRUAN DALAM PENGENALAN POLA HURUF PEGON JAWA

Artificial Neural Network (ANN) Technology) can help humans in processing data into information with design resembling the performance of the human brain. ANN adopts 5 aspects of human capability: Memorization, Generalization, Efficiency, Accuracy, and Tolerance in its application. ANN proves to be effective in pattern recognition. Researchers developed an application implementing ANN to recognize Java Pegon Letter pattern. The research uses 160 image data, divided into 100 training data (consisting of 5 normal images for each character) and 60 test data (consisting of 1 normal data, 1 data is not complete/corrupt, and 1 data with noise) for each character. The data obtained from the processed captures, so all of data have the same dimensions and size: 100x100 pixels. All data is processed through preprocessing and extraction stages. Furthermore, the data result is used in training stage to recognize the pattern of Java Pegon by applying the Learning Vector Quantization method. The application can recognize Java pegon pattern very well. The application can recognize 100% of training data and test data. This application also has the ability to recognize abnormal data very well, such as data with noise or corrupted data.

Classification of Pineapple Fruit Comosus Merr (Nanas) Quality Using Learning Vector Quantization Method

The demands of publics for these fruits Ananas Comosus Merr (Pineapple) became higher years to years because of the fruit has so many virtues for human healthy and the taste of this fruit is sweet and fresh. Therefore the pineapple farmers have to protect the quality and quantity of this plant in order to get high produce. This research help the pineapple farmers to classify to quality of pineapple fruits by using neural network with Learning Vector Quantization method which has 2 classes, such as: First quality (1st) and Second quality (2nd) quality. This method has 2 process they are : training process and testing process. To input data in the training and testing process are using uniformity, characteristic of varieties, the rate of aging, hardness, size, stem, crown, manure, destroyer, spoilage, rotten and the total solid content of the least was taken by observed the crop of pineapple farmers in the Teluk Batil village Sungai Apit district Siak Riau province. Learning Vector Quantization method automatically will classify the pineapple into their class. The result of the testing classification has gotten the accuracy 65.56% for the first (1st) quality and 34.44% for the second (2nd) quality. At the second testing has gotten 66.67% the accuracy for the first (1st) quality and 33.33% for the second (2nd) quality. At the third (3rd) testing has gotten 64.44% the accuracy for first (1st) quality and 35.56% for the second (2nd) quality.

Processing a measured vibroacoustic signal for rock type recognition in rotary drilling technology

Rotary drilling technology with diamond bits places high demands on expertise and experience. It plays an important role in the search, preparation, and extraction of raw materials in geological and engineering research. A large amount of hard-to-disintegrate raw materials are currently disintegrated directly by drilling diamond bits. The technology of the disintegration of rocks by rotary drilling represents a significant point and it is a central focus in the field of research in order to optimize the process. Optimization of the process currently considers both the energy and economic aspects, the efficiency of the drilling process, plus the protection of the working environment and the environment. In the case of rotary drilling technology, an interaction between the disintegrating tool and disintegrated rock or rock massif occurs. The interaction of the tool with the rock is a rudimentary source of vibration and noise. The generated vibroacoustic signal is an integrating information source. Information about the current state of the drilling process and the drilling device is obtained indirectly by processing this signal. The basic aim is to use the vibroacoustic signal to identify the type of drilling rock, i.e. its classification to the relevant category from a geomechanical point of view in relation to the optimal drilling mode. This paper deals with the use of a vibroacoustic signal in order to identify the rock in terms of the efficiency of the set mode (i.e., pressure force, revolutions and drilling bit) under current geotechnical conditions. The paper also points to the possibility of recognizing and classifying rock in the drilling process by use of a vector quantization method utilising the vibroacoustic signal.

Improving the operational efficiency of outbound retail logistics using clustering of retailers and consumers

PurposeThe purpose of this paper is to investigate ways to improve operational efficiency of outbound retail logistics considering retailers and consumers by using clustering approach. The retailers are allocated to serve a cluster of consumers. This study demonstrates economic and environment benefits that are achieved in terms of reduced delivery time, transportation cost and carbon emissions.Design/methodology/approachThis study is based on modeling the outbound logistics of a retail chain by using Kohonen self-organizing map (KSOM). KSOM is an unsupervised learning and data analysis method for vector quantization, which is based on Euclidean distance method to form clusters.FindingsAppropriate clustering of retailers and consumers provides efficient locations of retailers that are identified using the KSOM training algorithm. It provides optimum distance with lesser delivery time, transportation cost and carbon emissions.Research limitations/implicationsThe implication of research includes modeling of operational procedures in a retail supply chain, which is a crucial task for a business. These operations positively affect the reduction in inventory and distribution costs, improvement in customer service and responsiveness to the ever-changing markets of consumer durables. Overall results are insightful and practical in the sense that implementation would result in consumer convenience, eco-friendly environment, etc.Originality/valueThere is not enough research available on outbound retail logistics considering retailers and consumers using clustering approach.

A new fuzzy learning vector quantization method for classification problems based on a granular approach

In this paper, a new Fuzzy Learning Vector Quantization (FuzzLVQ) method for classification is presented. FuzzLVQ is a hybrid method based on LVQ neural networks and fuzzy systems. FuzzLVQ was implemented using modular architectures based on a granular approach, to further improve its performance in complex classification problems. The contribution of this research work is the development of the new fuzzy learning quantization method (FuzzLVQ) for classification problems, which is a hybrid method based on LVQ neural networks and fuzzy inference systems. In this work, a set of 15 experiments were performed for each presented architecture and a comparison between the classical LVQ algorithm and the new FuzzLVQ is presented. The obtained results are favorable, and the classification accuracy is slightly higher with the FuzzLVQ method. The hybridization proved to be beneficial also in other aspects of the method, such as in the training times of the neural network, and the number of cluster centers, which are reduced in respect to the classical LVQ performance. A proper optimization method could help improve the classification accuracy even more.

Joint image coding and lossless data hiding in VQ indices using adaptive coding techniques

In this paper, we propose a novel lossless data hiding method for vector quantization (VQ) compressed images. The existing methods sort the codebook based on the intensity of codewords to obtain rearranged indices (RI). The indices in RI are predicted, and the secret bits, indicators, and prediction errors are then encoded to construct an output code stream (CS). However, the neighboring indices of RI might not have similar intensity, which could impede the coding efficiency. Moreover, the equal-length indicators used in the existing methods might not efficiently represent the cases categorized in the encoding stage. As a result, the size of CS in these methods is likely to increase unnecessarily. The proposed method rearranges the indices to make their spatial correlations stronger, so that the obtained RI is more advantageous for predictive coding. A weight-controlled least square estimator is employed for prediction, and the indicators are better assigned according to the occurrence frequencies of prediction errors. The run-length coding technique is adopted to further reduce the code length. Experimental results show that the proposed method effectively reduces the bitrate for various codebook sizes.

Iqro Reading Learning System through Speech Recognition Using Mel Frequency Cepstral Coefficient (MFCC) and Vector Quantization (VQ) Method

Historically, the study of Qur'an in Indonesia evolved along with the spread of Islam. Learning methods of reading the Qur'an have been found ranging from al-Baghdadi, al-Barqi, Qiraati, Iqro', Human, Tartila, and others, which can make it easier to learn to read the Qur'an. Currently, the development of speech recognition technology can be used for the detection of Iqro vol 3 reading pronunciations. Speech recognition consists of two general stages of feature extraction and speech matching. The feature extraction step is used to derive speech-feature and speech-matching stages to compare compatibility between test sound and train voice. The speech recognition method used to recognize Iqro readings is extracting speech signal features using Mel Frequency Cepstral Coefficient (MFCC) and classifying them using Vector Quantization (VQ) to get the appropriate speech results. The result of testing for speech recognition system of Iqro reading has been tested for 30 peoples as a sample of data and there are 6 utterances indicating the information failed, so the system has a success rate of 80%.

Ultra-fast global homology detection with Discrete Cosine Transform and Dynamic Time Warping.

Evolutionary information is crucial for the annotation of proteins in bioinformatics. The amount of retrieved homologs often correlates with the quality of predicted protein annotations related to structure or function. With a growing amount of sequences available, fast and reliable methods for homology detection are essential, as they have a direct impact on predicted protein annotations. We developed a discriminative, alignment-free algorithm for homology detection with quasi-linear complexity, enabling theoretically much faster homology searches. To reach this goal, we convert the protein sequence into numeric biophysical representations. These are shrunk to a fixed length using a novel vector quantization method which uses a Discrete Cosine Transform compression. We then compute, for each compressed representation, similarity scores between proteins with the Dynamic Time Warping algorithm and we feed them into a Random Forest. The WARP performances are comparable with state of the art methods. The method is available at http://ibsquare.be/warp. Supplementary data are available at Bioinformatics online.

Navajo Nation, USA, Precipitation Variability from 2002 to 2015

AbstractDue to its semi‐arid climate, the Navajo Nation, situated in the southwestern United States, is sensitive to small changes in precipitation. However, little information on patterns and causes of rainfall variation is available for this sparsely populated region. In order to study stability and variability over time, this study characterized hydroclimatic changes for the Navajo Nation over timescales of months to years based on data from 90 sites from 2002 to 2015. This research will help local water managers identify related precipitation areas within the region, compare Navajo Nation precipitation with climate indices to ascertain larger‐scale atmospheric contributors to precipitation in the Four Corners region, and support future water planning in this understudied region. A vector quantization method, called k‐means clustering, identified five sub‐regions of contrasting precipitation climatology. The regions differed in the timing, magnitude, and relative importance of the winter and summer peaks comprising the bimodal precipitation regime of the area. Correlation examination of spatial and temporal trends of precipitation variability with three climate indices revealed strong winter precipitation relationships to the Pacific North American teleconnection pattern for all regions; summer precipitation teleconnections were weaker and more variable; however, modest correlations with Pacific Decadal Oscillation were observed. Climate field analysis indicates that cold‐season precipitation is enhanced by intensification of the Aleutian Low with a storm trajectory into the southwest United States; warm season precipitation is enhanced by poleward shift of the North American monsoon ridge.

Analyzing Coverage Probability of Multitier Heterogeneous Networks Under Quantized Multiuser ZF Beamforming

In this paper, we analyze the coverage probability of a large-scale heterogenous cellular network (HetNet) where base stations (BSs) are equipped with multiantennas, serving a number of single-antenna user equipments (UEs) in the downlink—space division multiple access (SDMA). Adopting the tool of stochastic geometry, we analyze the coverage probability under the quantized zero-forcing beamforming (ZFBF). This analysis, however, intricate as a typical UE in a given cell experiences: i) a residual inter-user interference (IUI) rendered by the quantization error; ii) quantization-driven statistical correlation between its attending signal power and IUI; iii) Nakagami-type fading environment, induced by postprocessing ZFBF; and iv) intercell interference (ICI). Current related literature has not effectively accounted for issues (i), (ii), and (iv) and commonly resorted to simplifying scenarios of full SDMA (i.e., exponential fading), independency of IUI and attending signal power, approximating IUI via an exponential random variable, and the quantization cell approximation (QCA). Our analysis relaxes such limiting assumptions; it includes the more relevant random vector quantization (RVQ) method, and compared to the literature, does not resort to the evaluation of a higher order differentiation for the evaluation of the coverage probability by exploiting our previously developed method. We further introduce three novel approximations of the coverage probability: RVQ-M , RVG-G , and QCA , which the first two approximations are given in semi-closed-form formulas (requiring a simple integration step) and the last approximation is in a closed-form expression. All approximations are numerically affordable, and we confirm the RVG-G approximation is the most accurate one. Finally, we exploit our analysis to study several practically insightful issues to shed some lights on: 1) for given feedback capacity and number of UEs, when desification becomes beneficial; 2) how multitierness of HetNets can be exploited for substantially reducing feedback capacity; and 3) an adaptive share of feedback capacity for the purpose of CDI as well as channel quality information (CQI) feedbacks, when there is an uplink penalty subjected for the feedback procedure and the scheduled data rate at each BS is specified adaptively based on the quantized CQI.

A general codebook design method for vector quantization

Vector quantization (VQ) is widely used in image processing applications, the primary focus of VQ is to determine a codebook to represent the original image well. In order to make a codebook perform better on both distortion and bit rate (BR), a general codebook (GCB) for VQ is proposed in this paper. Unlike common codebook (CCB) or private codebook (PCB), GCB is a new structure of codebook where the codewords can either come from CCB or by training the input image. By applying the codewords in CCB that perform well and updating inactive codewords, only the new generated codewords and flags of codewords to be replaced are transmitted along with index table (IT). Therefore,the BR can be significantly reduced while the performance of distortion can be efficiently improved. The experimental results demonstrate that our proposed GCB has a better performance than CCB and various kinds of PCB-based methods.

Vector Quantization based Pairwise Joint Distance Maps (VQ-PJDM) for 3D Action Recognition

This paper presents an approach for 3D action recognition using vector quantization with pairwise joint distance maps. We name this approach as VQ-PJDM. The main problem for 3D action recognition using skeleton data is that dealing with the variable length of action sequences. We solve this problem by approximation of each action sequence as a codebook, which is the output of Vector Quantization (VQ) method. The codebook size is fixed for any length of the action sequence. After all actions in the data set are approximated by VQ method, the Pairwise Distance Joint Distance Maps(PJDM) are calculated from approximated actions. The voting classifier is employed for action classification. The empirical results on the UT Kinect dataset prove that the proposed method gives better results than that of state of the art.

Application of Learning Vector Quantization Method in Kinect Device AS a Base of the Development of Behaviour Detection System

One of the lessons for mental disorder students in Special Schools is practicum lessons in the form of vocational education. This lesson uses equipment that requires prudence. Mental disorder students have characteristics that are low memory and move based on intuition. Teachers should pay extra attention especially to detect student behavior during the learning. This detection is needed for learning to take place smoothly and students are safe from the dangers around the practicum place. Teacher's feedback on the detection obtained in the form of a warning from the teacher. This study is expected to be useful for providing a special detection pattern for students to assist teachers by providing feedback in the form of warnings using natural motion detection technology. This research was conducted using Kinect as data input and data was processed using artificial neural network and Learning Vector Quantization method. The dangerous attitude used in the test is the attitude of standing at the time of drilling position. The data used by training is 126 data and do training using LVQ. At the LVQ training stage, the training was conducted with parameter of Learning Rate 0,05, maximum Iteration 44, reduction of learning rate 0.01, and Learning rate minimum 0,02.