Abstract: As biometric systems become widely deployed, the security of identification systems can be easily attacked by various spoof materials. This paper contributes to finding a reliable and practical anti-spoofing method using Convolutional Neural Networks (CNNs) based on the types of loss functions and optimizers. The types of CNNs used in this paper include AlexNet, VGGNet, and ResNet. By using various loss functions including Cross-Entropy, Center Loss, Cosine Proximity, and Hinge Loss, and various loss optimizers which include Adam, SGD, RMSProp, Adadelta, Adagrad, and Nadam, we obtained significant performance changes. We realize that choosing the correct loss function for each model is crucial since different loss functions lead to different errors on the same evaluation. By using a subset of the Livdet 2017 database, we validate our approach to compare the generalization power. It is important to note that we use a subset of LiveDet and the database is the same across all training and testing for each model. This way, we can compare the performance, in terms of generalization, for the unseen data across all different models. The best CNN (AlexNet) with the appropriate loss function and optimizers result in more than 3% of performance gain over the other CNN models with the default loss function and optimizer. In addition to the highest generalization performance, this paper also contains the models with high accuracy associated with parameters and mean average error rates to find the model that consumes the least memory and computation time for training and testing. Although AlexNet has less complexity over other CNN models, it is proven to be very efficient. For practical anti-spoofing systems, the deployed version should use a small amount of memory and should run very fast with high anti-spoofing performance. For our deployed version on smartphones, additional processing steps, such as quantization and pruning algorithms, have been applied in our final model.
Abstract: Intrusion detection systems (IDS) are the main components of network security. These systems analyze the network events for intrusion detection. The design of an IDS is through the training of normal traffic data or attack. The methods of machine learning are the best ways to design IDSs. In the method presented in this article, the pruning algorithm of C5.0 decision tree is being used to reduce the features of traffic data used and training IDS by the least square vector algorithm (LS-SVM). Then, the remaining features are arranged according to the predictor importance criterion. The least important features are eliminated in the order. The remaining features of this stage, which have created the highest level of accuracy in LS-SVM, are selected as the final features. The features obtained, compared to other similar articles which have examined the selected features in the least squared support vector machine model, are better in the accuracy, true positive rate, and false positive. The results are tested by the UNSW-NB15 dataset.
Abstract: In order to prove the applicability of a conical spouted bed combustor for the thermal exploitation of vineyard pruning wastes, the flow regimes of beds consisting of vine shoot beds and an inert bed were established under different operating conditions. The effect of inlet air temperature on the minimum spouted velocity was evaluated. Batch combustion of vine shoots in a conical spouted bed combustor was conducted at temperatures in the range 425-550 ºC with an inert bed. The experimental values of combustion efficiency of vine shoot calculated from the concentration the exhaust gases were assessed. The high experimental combustion efficiency obtained evidenced the proper suitability of the conical spouted bed combustor for the thermal combustion of vine shoots.
Abstract: Hand gesture recognition is a technique used to locate, detect, and recognize a hand gesture. Detection and recognition are concepts of Artificial Intelligence (AI). AI concepts are applicable in Human Computer Interaction (HCI), Expert systems (ES), etc. Hand gesture recognition can be used in sign language interpretation. Sign language is a visual communication tool. This tool is used mostly by deaf societies and those with speech disorder. Communication barriers exist when societies with speech disorder interact with others. This research aims to build a hand recognition system for Lesotho’s Sesotho and English language interpretation. The system will help to bridge the communication problems encountered by the mentioned societies. The system has various processing modules. The modules consist of a hand detection engine, image processing engine, feature extraction, and sign recognition. Detection is a process of identifying an object. The proposed system uses Canny pruning Haar and Haarcascade detection algorithms. Canny pruning implements the Canny edge detection. This is an optimal image processing algorithm. It is used to detect edges of an object. The system employs a skin detection algorithm. The skin detection performs background subtraction, computes the convex hull, and the centroid to assist in the detection process. Recognition is a process of gesture classification. Template matching classifies each hand gesture in real-time. The system was tested using various experiments. The results obtained show that time, distance, and light are factors that affect the rate of detection and ultimately recognition. Detection rate is directly proportional to the distance of the hand from the camera. Different lighting conditions were considered. The more the light intensity, the faster the detection rate. Based on the results obtained from this research, the applied methodologies are efficient and provide a plausible solution towards a light-weight, inexpensive system which can be used for sign language interpretation.
Abstract: Although most of the existing skyline queries algorithms focused basically on querying static points through static databases; with the expanding number of sensors, wireless communications and mobile applications, the demand for continuous skyline queries has increased. Unlike traditional skyline queries which only consider static attributes, continuous skyline queries include dynamic attributes, as well as the static ones. However, as skyline queries computation is based on checking the domination of skyline points over all dimensions, considering both the static and dynamic attributes without separation is required. In this paper, we present an efficient algorithm for computing continuous skyline queries without discriminating between static and dynamic attributes. Our algorithm in brief proceeds as follows: First, it excludes the points which will not be in the initial skyline result; this pruning phase reduces the required number of comparisons. Second, the association between the spatial positions of data points is examined; this phase gives an idea of where changes in the result might occur and consequently enables us to efficiently update the skyline result (continuous update) rather than computing the skyline from scratch. Finally, experimental evaluation is provided which demonstrates the accuracy, performance and efficiency of our algorithm over other existing approaches.
Abstract: Most of self-tuning fuzzy systems, which are
automatically constructed from learning data, are based on the
steepest descent method (SDM). However, this approach often
requires a large convergence time and gets stuck into a shallow
local minimum. One of its solutions is to use fuzzy rule modules
with a small number of inputs such as DIRMs (Double-Input Rule
Modules) and SIRMs (Single-Input Rule Modules). In this paper,
we consider a (generalized) DIRMs model composed of double
and single-input rule modules. Further, in order to reduce the
redundant modules for the (generalized) DIRMs model, pruning and
generative learning algorithms for the model are suggested. In order
to show the effectiveness of them, numerical simulations for function
approximation, Box-Jenkins and obstacle avoidance problems are
performed.
Abstract: Association rule mining is one of the most important fields of data mining and knowledge discovery. In this paper, we propose an efficient multiple support frequent pattern growth algorithm which we called “MSFP-growth” that enhancing the FPgrowth algorithm by making infrequent child node pruning step with multiple minimum support using maximum constrains. The algorithm is implemented, and it is compared with other common algorithms: Apriori-multiple minimum supports using maximum constraints and FP-growth. The experimental results show that the rule mining from the proposed algorithm are interesting and our algorithm achieved better performance than other algorithms without scarifying the accuracy.
Abstract: In this paper we consider the rule reduct generation
problem. Rule Reduct Generation (RG) and Modified Rule
Generation (MRG) algorithms, that are used to solve this problem,
are well-known. Alternative to these algorithms, we develop Pruning
Rule Generation (PRG) algorithm. We compare the PRG algorithm
with RG and MRG.
Abstract: This paper combines the branch-and-bound method and the petri net to solve the two-sided assembly line balancing problem, thus facilitating effective branching and pruning of tasks. By integrating features of the petri net, such as reachability graph and incidence matrix, the propose method can support the branch-and-bound to effectively reduce poor branches with systematic graphs. Test results suggest that using petri net in the branching process can effectively guide the system trigger process, and thus, lead to consistent results.
Abstract: Associative classification (AC) is a data mining approach that combines association rule and classification to build classification models (classifiers). AC has attracted a significant attention from several researchers mainly because it derives accurate classifiers that contain simple yet effective rules. In the last decade, a number of associative classification algorithms have been proposed such as Classification based Association (CBA), Classification based on Multiple Association Rules (CMAR), Class based Associative Classification (CACA), and Classification based on Predicted Association Rule (CPAR). This paper surveys major AC algorithms and compares the steps and methods performed in each algorithm including: rule learning, rule sorting, rule pruning, classifier building, and class prediction.
Abstract: Decision tree algorithms have very important place at
classification model of data mining. In literature, algorithms use
entropy concept or gini index to form the tree. The shape of the
classes and their closeness to each other some of the factors that
affect the performance of the algorithm. In this paper we introduce a
new decision tree algorithm which employs data (attribute) folding
method and variation of the class variables over the branches to be
created. A comparative performance analysis has been held between
the proposed algorithm and C4.5.
Abstract: Artificial Intelligence based gaming is an interesting topic in the state-of-art technology. This paper presents an automation of a tradition Omani game, called Al-Hawalees. Its related issues are resolved and implemented using artificial intelligence approach. An AI approach called mini-max procedure is incorporated to make a diverse budges of the on-line gaming. If number of moves increase, time complexity will be increased in terms of propositionally. In order to tackle the time and space complexities, we have employed a back propagation neural network (BPNN) to train in off-line to make a decision for resources required to fulfill the automation of the game. We have utilized Leverberg- Marquardt training in order to get the rapid response during the gaming. A set of optimal moves is determined by the on-line back propagation training fashioned with alpha-beta pruning. The results and analyses reveal that the proposed scheme will be easily incorporated in the on-line scenario with one player against the system.
Abstract: This paper presents an exact pruning algorithm with
adaptive pruning interval for general dynamic neural networks
(GDNN). GDNNs are artificial neural networks with internal dynamics.
All layers have feedback connections with time delays to the
same and to all other layers. The structure of the plant is unknown, so
the identification process is started with a larger network architecture
than necessary. During parameter optimization with the Levenberg-
Marquardt (LM) algorithm irrelevant weights of the dynamic neural
network are deleted in order to find a model for the plant as
simple as possible. The weights to be pruned are found by direct
evaluation of the training data within a sliding time window. The
influence of pruning on the identification system depends on the
network architecture at pruning time and the selected weight to be
deleted. As the architecture of the model is changed drastically during
the identification and pruning process, it is suggested to adapt the
pruning interval online. Two system identification examples show
the architecture selection ability of the proposed pruning approach.
Abstract: Data mining, which is the exploration of
knowledge from the large set of data, generated as a result of
the various data processing activities. Frequent Pattern Mining
is a very important task in data mining. The previous
approaches applied to generate frequent set generally adopt
candidate generation and pruning techniques for the
satisfaction of the desired objective. This paper shows how
the different approaches achieve the objective of frequent
mining along with the complexities required to perform the
job. This paper will also look for hardware approach of cache
coherence to improve efficiency of the above process. The
process of data mining is helpful in generation of support
systems that can help in Management, Bioinformatics,
Biotechnology, Medical Science, Statistics, Mathematics,
Banking, Networking and other Computer related
applications. This paper proposes the use of both upward and
downward closure property for the extraction of frequent item
sets which reduces the total number of scans required for the
generation of Candidate Sets.
Abstract: An effective visual error concealment method has been presented by employing a robust rotation, scale, and translation (RST) invariant partial patch matching model (RSTI-PPMM) and
exemplar-based inpainting. While the proposed robust and inherently
feature-enhanced texture synthesis approach ensures the generation
of excellent and perceptually plausible visual error concealment results, the outlier pruning property guarantees the significant quality improvements, both quantitatively and qualitatively. No intermediate
user-interaction is required for the pre-segmented media and the
presented method follows a bootstrapping approach for an automatic
visual loss recovery and the image and video error concealment.
Abstract: This paper sets forth the possibility and importance about applying Data Mining in Web logs mining and shows some problems in the conventional searching engines. Then it offers an improved algorithm based on the original AprioriAll algorithm which has been used in Web logs mining widely. The new algorithm adds the property of the User ID during the every step of producing the candidate set and every step of scanning the database by which to decide whether an item in the candidate set should be put into the large set which will be used to produce next candidate set. At the meantime, in order to reduce the number of the database scanning, the new algorithm, by using the property of the Apriori algorithm, limits the size of the candidate set in time whenever it is produced. Test results show the improved algorithm has a more lower complexity of time and space, better restrain noise and fit the capacity of memory.
Abstract: Although backpropagation ANNs generally predict
better than decision trees do for pattern classification problems, they
are often regarded as black boxes, i.e., their predictions cannot be
explained as those of decision trees. In many applications, it is
desirable to extract knowledge from trained ANNs for the users to
gain a better understanding of how the networks solve the problems.
A new rule extraction algorithm, called rule extraction from artificial
neural networks (REANN) is proposed and implemented to extract
symbolic rules from ANNs. A standard three-layer feedforward ANN
is the basis of the algorithm. A four-phase training algorithm is
proposed for backpropagation learning. Explicitness of the extracted
rules is supported by comparing them to the symbolic rules generated
by other methods. Extracted rules are comparable with other methods
in terms of number of rules, average number of conditions for a rule,
and predictive accuracy. Extensive experimental studies on several
benchmarks classification problems, such as breast cancer, iris,
diabetes, and season classification problems, demonstrate the
effectiveness of the proposed approach with good generalization
ability.
Abstract: In this paper, we propose a novel fast search algorithm for short MPEG video clips from video database. This algorithm is based on the adjacent pixel intensity difference quantization (APIDQ) algorithm, which had been reliably applied to human face recognition previously. An APIDQ histogram is utilized as the feature vector of the frame image. Instead of fully decompressed video frames, partially decoded data, namely DC images are utilized. Combined with active search [4], a temporal pruning algorithm, fast and robust video search can be realized. The proposed search algorithm has been evaluated by 6 hours of video to search for given 200 MPEG video clips which each length is 15 seconds. Experimental results show the proposed algorithm can detect the similar video clip in merely 80ms, and Equal Error Rate (ERR) of 3 % is achieved, which is more accurately and robust than conventional fast video search algorithm.
Abstract: This paper presents a multi-objective order allocation
planning problem with the consideration of various real-world
production features. A novel hybrid intelligent optimization model,
integrating a multi-objective memetic optimization process, a Monte
Carlo simulation technique and a heuristic pruning technique, is
proposed to handle this problem. Experiments based on industrial data
are conducted to validate the proposed model. Results show that (1)
the proposed model can effectively solve the investigated problem by
providing effective production decision-making solutions, which
outperformsan NSGA-II-based optimization process and an industrial
method.
Abstract: In 1990 [1] the subband-DFT (SB-DFT) technique was proposed. This technique used the Hadamard filters in the decomposition step to split the input sequence into low- and highpass sequences. In the next step, either two DFTs are needed on both bands to compute the full-band DFT or one DFT on one of the two bands to compute an approximate DFT. A combination network with correction factors was to be applied after the DFTs. Another approach was proposed in 1997 [2] for using a special discrete wavelet transform (DWT) to compute the discrete Fourier transform (DFT). In the first step of the algorithm, the input sequence is decomposed in a similar manner to the SB-DFT into two sequences using wavelet decomposition with Haar filters. The second step is to perform DFTs on both bands to obtain the full-band DFT or to obtain a fast approximate DFT by implementing pruning at both input and output sides. In this paper, the wavelet-based DFT (W-DFT) with Haar filters is interpreted as SB-DFT with Hadamard filters. The only difference is in a constant factor in the combination network. This result is very important to complete the analysis of the W-DFT, since all the results concerning the accuracy and approximation errors in the SB-DFT are applicable. An application example in spectral analysis is given for both SB-DFT and W-DFT (with different filters). The adaptive capability of the SB-DFT is included in the W-DFT algorithm to select the band of most energy as the band to be computed. Finally, the W-DFT is extended to the two-dimensional case. An application in image transformation is given using two different types of wavelet filters.