Abstract: In the Hierarchical Temporal Memory (HTM) paradigm
the effect of overlap between inputs on the activation of columns in
the spatial pooler is studied. Numerical results suggest that similar
inputs are represented by similar sets of columns and dissimilar inputs
are represented by dissimilar sets of columns. It is shown that the
spatial pooler produces these results under certain conditions for
the connectivity and proximal thresholds. Following the discussion
of the initialization of parameters for the thresholds, corresponding
qualitative arguments about the learning dynamics of the spatial
pooler are discussed.
Abstract: In the past few years, the amount of malicious software
increased exponentially and, therefore, machine learning algorithms
became instrumental in identifying clean and malware files through
(semi)-automated classification. When working with very large
datasets, the major challenge is to reach both a very high malware
detection rate and a very low false positive rate. Another challenge
is to minimize the time needed for the machine learning algorithm to
do so. This paper presents a comparative study between different
machine learning techniques such as linear classifiers, ensembles,
decision trees or various hybrids thereof. The training dataset consists
of approximately 2 million clean files and 200.000 infected files,
which is a realistic quantitative mixture. The paper investigates the
above mentioned methods with respect to both their performance
(detection rate and false positive rate) and their practicability.
Abstract: This paper introduces a proposal scheme for an
Intelligent System applied to Pedagogical Advising using Case-Based
Reasoning, to find consolidated solutions before used for the new
problems, making easier the task of advising students to the
pedagogical staff. We do intend, through this work, introduce the
motivation behind the choices for this system structure, justifying the
development of an incremental and smart web system who learns
bests solutions for new cases when it’s used, showing technics and
technology.
Abstract: Customer churn prediction is one of the most useful
areas of study in customer analytics. Due to the enormous amount
of data available for such predictions, machine learning and data
mining have been heavily used in this domain. There exist many
machine learning algorithms directly applicable for the problem of
customer churn prediction, and here, we attempt to experiment on
a novel approach by using a cognitive learning based technique in
an attempt to improve the results obtained by using a combination
of supervised learning methods, with cognitive unsupervised learning
methods.
Abstract: Due to the rapid increase of Internet, web opinion
sources dynamically emerge which is useful for both potential
customers and product manufacturers for prediction and decision
purposes. These are the user generated contents written in natural
languages and are unstructured-free-texts scheme. Therefore, opinion
mining techniques become popular to automatically process customer
reviews for extracting product features and user opinions expressed
over them. Since customer reviews may contain both opinionated and
factual sentences, a supervised machine learning technique applies
for subjectivity classification to improve the mining performance. In
this paper, we dedicate our work is the task of opinion
summarization. Therefore, product feature and opinion extraction is
critical to opinion summarization, because its effectiveness
significantly affects the identification of semantic relationships. The
polarity and numeric score of all the features are determined by
Senti-WordNet Lexicon. The problem of opinion summarization
refers how to relate the opinion words with respect to a certain
feature. Probabilistic based model of supervised learning will
improve the result that is more flexible and effective.
Abstract: Development of a method to estimate gene functions is
an important task in bioinformatics. One of the approaches for the
annotation is the identification of the metabolic pathway that genes are
involved in. Since gene expression data reflect various intracellular
phenomena, those data are considered to be related with genes’
functions. However, it has been difficult to estimate the gene function
with high accuracy. It is considered that the low accuracy of the
estimation is caused by the difficulty of accurately measuring a gene
expression. Even though they are measured under the same condition,
the gene expressions will vary usually. In this study, we proposed a
feature extraction method focusing on the variability of gene
expressions to estimate the genes' metabolic pathway accurately. First,
we estimated the distribution of each gene expression from replicate
data. Next, we calculated the similarity between all gene pairs by KL
divergence, which is a method for calculating the similarity between
distributions. Finally, we utilized the similarity vectors as feature
vectors and trained the multiclass SVM for identifying the genes'
metabolic pathway. To evaluate our developed method, we applied the
method to budding yeast and trained the multiclass SVM for
identifying the seven metabolic pathways. As a result, the accuracy
that calculated by our developed method was higher than the one that
calculated from the raw gene expression data. Thus, our developed
method combined with KL divergence is useful for identifying the
genes' metabolic pathway.
Abstract: Load modeling is one of the central functions in
power systems operations. Electricity cannot be stored, which means
that for electric utility, the estimate of the future demand is necessary
in managing the production and purchasing in an economically
reasonable way. A majority of the recently reported approaches are
based on neural network. The attraction of the methods lies in the
assumption that neural networks are able to learn properties of the
load. However, the development of the methods is not finished, and
the lack of comparative results on different model variations is a
problem. This paper presents a new approach in order to predict the
Tunisia daily peak load. The proposed method employs a
computational intelligence scheme based on the Fuzzy neural
network (FNN) and support vector regression (SVR). Experimental
results obtained indicate that our proposed FNN-SVR technique gives
significantly good prediction accuracy compared to some classical
techniques.
Abstract: Analyzing DNA microarray data sets is a great
challenge, which faces the bioinformaticians due to the complication
of using statistical and machine learning techniques. The challenge
will be doubled if the microarray data sets contain missing data,
which happens regularly because these techniques cannot deal with
missing data. One of the most important data analysis process on
the microarray data set is feature selection. This process finds the
most important genes that affect certain disease. In this paper, we
introduce a technique for imputing the missing data in microarray
data sets while performing feature selection.
Abstract: A Distributed Denial of Service (DDoS) attack is a
major threat to cyber security. It originates from the network layer or
the application layer of compromised/attacker systems which are
connected to the network. The impact of this attack ranges from the
simple inconvenience to use a particular service to causing major
failures at the targeted server. When there is heavy traffic flow to a
target server, it is necessary to classify the legitimate access and
attacks. In this paper, a novel method is proposed to detect DDoS
attacks from the traces of traffic flow. An access matrix is created
from the traces. As the access matrix is multi dimensional, Principle
Component Analysis (PCA) is used to reduce the attributes used for
detection. Two classifiers Naive Bayes and K-Nearest neighborhood
are used to classify the traffic as normal or abnormal. The
performance of the classifier with PCA selected attributes and actual
attributes of access matrix is compared by the detection rate and
False Positive Rate (FPR).
Abstract: Red blood cells (RBC) are the most common types of
blood cells and are the most intensively studied in cell biology. The
lack of RBCs is a condition in which the amount of hemoglobin level
is lower than normal and is referred to as “anemia”. Abnormalities in
RBCs will affect the exchange of oxygen. This paper presents a
comparative study for various techniques for classifying the RBCs as
normal or abnormal (anemic) using WEKA. WEKA is an open
source consists of different machine learning algorithms for data
mining applications. The algorithms tested are Radial Basis Function
neural network, Support vector machine, and K-Nearest Neighbors
algorithm. Two sets of combined features were utilized for
classification of blood cells images. The first set, exclusively consist
of geometrical features, was used to identify whether the tested blood
cell has a spherical shape or non-spherical cells. While the second
set, consist mainly of textural features was used to recognize the
types of the spherical cells. We have provided an evaluation based on
applying these classification methods to our RBCs image dataset
which were obtained from Serdang Hospital - Malaysia, and
measuring the accuracy of test results. The best achieved
classification rates are 97%, 98%, and 79% for Support vector
machines, Radial Basis Function neural network, and K-Nearest
Neighbors algorithm respectively.
Abstract: The dramatic rise in the use of Social Media (SM)
platforms such as Facebook and Twitter provide access to an
unprecedented amount of user data. Users may post reviews on
products and services they bought, write about their interests, share
ideas or give their opinions and views on political issues. There is a
growing interest in the analysis of SM data from organisations for
detecting new trends, obtaining user opinions on their products and
services or finding out about their online reputations. A recent
research trend in SM analysis is making predictions based on
sentiment analysis of SM. Often indicators of historic SM data are
represented as time series and correlated with a variety of real world
phenomena like the outcome of elections, the development of
financial indicators, box office revenue and disease outbreaks. This
paper examines the current state of research in the area of SM mining
and predictive analysis and gives an overview of the analysis
methods using opinion mining and machine learning techniques.
Abstract: There have been a lot of efforts and researches undertaken in developing efficient tools for performing several tasks in data mining. Due to the massive amount of information embedded in huge data warehouses maintained in several domains, the extraction of meaningful pattern is no longer feasible. This issue turns to be more obligatory for developing several tools in data mining. Furthermore the major aspire of data mining software is to build a resourceful predictive or descriptive model for handling large amount of information more efficiently and user friendly. Data mining mainly contracts with excessive collection of data that inflicts huge rigorous computational constraints. These out coming challenges lead to the emergence of powerful data mining technologies. In this survey a diverse collection of data mining tools are exemplified and also contrasted with the salient features and performance behavior of each tool.
Abstract: Machine learning represents a set of topics dealing with the creation and evaluation of algorithms that facilitate pattern recognition, classification, and prediction, based on models derived from existing data. The data can present identification patterns which are used to classify into groups. The result of the analysis is the pattern which can be used for identification of data set without the need to obtain input data used for creation of this pattern. An important requirement in this process is careful data preparation validation of model used and its suitable interpretation. For breeders, it is important to know the origin of animals from the point of the genetic diversity. In case of missing pedigree information, other methods can be used for traceability of animal´s origin. Genetic diversity written in genetic data is holding relatively useful information to identify animals originated from individual countries. We can conclude that the application of data mining for molecular genetic data using supervised learning is an appropriate tool for hypothesis testing and identifying an individual.
Abstract: The amount of energy the world uses doubles every 20 years. Green homes play an important role in reducing the residential energy demand. This paper presents a platform that is intended to learn the behavior of home residents and build a profile about their habits and actions. The proposed resident aware home controller intervenes in the operation of home appliances in order to save energy without compromising the convenience of the residents. The presented platform can be used to simulate the actions and movements happening inside a home. The paper includes several optimization techniques that are meant to save energy in the home. In addition, several test scenarios are presented that show how the controller works. Moreover, this paper shows the computed actual savings when each of the presented techniques is implemented in a typical home. The test scenarios have validated that the techniques developed are capable of effectively saving energy at homes.
Abstract: Fully reusable spaceplanes do not exist as yet. This implies that design-qualification for optimized highly-integrated forebody-inlet configuration of booster-stage vehicle cannot be based on archival data of other spaceplanes. Therefore, this paper proposes a novel TIPSO-SVM expert system methodology. A non-trivial problem related to optimization and classification of hypersonic forebody-inlet configuration in conjunction with mass-model of the two-stage-to-orbit (TSTO) vehicle is solved. The hybrid-heuristic machine learning methodology is based on two-step improved particle swarm optimizer (TIPSO) algorithm and two-step support vector machine (SVM) data classification method. The efficacy of method is tested by first evolving an optimal configuration for hypersonic compression system using TIPSO algorithm; thereafter, classifying the results using two-step SVM method. In the first step extensive but non-classified mass-model training data for multiple optimized configurations is segregated and pre-classified for learning of SVM algorithm. In second step the TIPSO optimized mass-model data is classified using the SVM classification. Results showed remarkable improvement in configuration and mass-model along with sizing parameters.
Abstract: Since big data has become substantially more accessible and manageable due to the development of powerful tools for dealing with unstructured data, people are eager to mine information from social media resources that could not be handled in the past. Sentiment analysis, as a novel branch of text mining, has in the last decade become increasingly important in marketing analysis, customer risk prediction and other fields. Scientists and researchers have undertaken significant work in creating and improving their sentiment models. In this paper, we present a concept of selecting appropriate classifiers based on the features and qualities of data sources by comparing the performances of five classifiers with three popular social media data sources: Twitter, Amazon Customer Reviews, and Movie Reviews. We introduced a couple of innovative models that outperform traditional sentiment classifiers for these data sources, and provide insights on how to further improve the predictive power of sentiment analysis. The modeling and testing work was done in R and Greenplum in-database analytic tools.
Abstract: As internet continues to expand its usage with an
enormous number of applications, cyber-threats have significantly
increased accordingly. Thus, accurate detection of malicious traffic in
a timely manner is a critical concern in today’s Internet for security.
One approach for intrusion detection is to use Machine Learning (ML)
techniques. Several methods based on ML algorithms have been
introduced over the past years, but they are largely limited in terms of
detection accuracy and/or time and space complexity to run. In this
work, we present a novel method for intrusion detection that
incorporates a set of supervised learning algorithms. The proposed
technique provides high accuracy and outperforms existing techniques
that simply utilizes a single learning method. In addition, our
technique relies on partial flow information (rather than full
information) for detection, and thus, it is light-weight and desirable for
online operations with the property of early identification. With the
mid-Atlantic CCDC intrusion dataset publicly available, we show that
our proposed technique yields a high degree of detection rate over 99%
with a very low false alarm rate (0.4%).
Abstract: Prediction of future research topics by using time series analysis either statistical or machine learning has been conducted previously by several researchers. Several methods have been proposed to combine the forecasting results into single forecast. These methods use fixed combination of individual forecast to get the final forecast result. In this paper, quite different approach is employed to select the forecasting methods, in which every point to forecast is calculated by using the best methods used by similar validation dataset. The dataset used in the experiment is time series derived from research report in Garuda, which is an online sites belongs to the Ministry of Education in Indonesia, over the past 20 years. The experimental result demonstrates that the proposed method may perform better compared to the fix combination of predictors. In addition, based on the prediction result, we can forecast emerging research topics for the next few years.
Abstract: Anomaly detection techniques have been focused on two main components: data extraction and selection and the second one is the analysis performed over the obtained data. The goal of this paper is to analyze the influence that each of these components has over the system performance by evaluating detection over network scenarios with different setups. The independent variables are as follows: the number of system inputs, the way the inputs are codified and the complexity of the analysis techniques. For the analysis, some approaches of artificial neural networks are implemented with different number of layers. The obtained results show the influence that each of these variables has in the system performance.
Abstract: Real-time or in-line process monitoring frameworks are designed to give early warnings for a fault along with meaningful identification of its assignable causes. In artificial intelligence and machine learning fields of pattern recognition various promising approaches have been proposed such as kernel-based nonlinear machine learning techniques. This work presents a kernel-based empirical monitoring scheme for batch type production processes with small sample size problem of partially unbalanced data. Measurement data of normal operations are easy to collect whilst special events or faults data are difficult to collect. In such situations, noise filtering techniques can be helpful in enhancing process monitoring performance. Furthermore, preprocessing of raw process data is used to get rid of unwanted variation of data. The performance of the monitoring scheme was demonstrated using three-dimensional batch data. The results showed that the monitoring performance was improved significantly in terms of detection success rate of process fault.