Contributions to Deep Learning Models - RiuNet
... The goal of this thesis is to present some contributions to the Deep Learning framework, particularly focused on computer vision problems dealing with images. These
contributions can be summarized in two novel methods proposed: a new regularization technique for Restricted Boltzmann Machines called ...
Financial Time Series Forecasting Using Improved Wavelet Neural
...  proposes a GRANN-ARIMA hybrid model which combines non-linear Grey
Relational Artificial Neural Network and linear ARIMA model for time series
forecasting. The experimental results indicate that the hybrid method outperforms ARIMA, Multiple Regression, GRANN, MARMA, MR ANN, ARIMA
and traditiona ...
How We`re Predicting AI—or Failing To
... they are diverse. Starting with Turing’s initial estimation of a 30% pass rate on Turing
test by the year 2000 (Turing 1950), computer scientists, philosophers and journalists
have never been shy to oﬀer their own definite prognostics, claiming AI to be impossible
(Jacquette 1987) or just around the ...
Wrappers for feature subset selection
An optimal feature subset need not be unique because it may be possible to achieve
the same accuracy using different sets of features (e.g., when two features are perfectly
correlated, one can be replaced by the other). By definition, to get the highest possible
accuracy, the best subset th ...
link - Worcester Polytechnic Institute
... relevance. At their best they have been shown to achieve the same educational gain as one on one
human tutoring (Koedinger et al., 1997). They have also received the attention of White House,
which mentioned a tutoring platform named ASSISTments in its National Educational
Technology Plan (Departmen ...
Handling the Class Imbalance Problem in Binary Classification
... Natural processes often generate some observations more frequently than others.
These processes result in an unbalanced distributions which cause the classifiers to
bias toward the majority class especially because most classifiers assume a normal
distribution. The quantity and the diversity of imba ...
Consensus group stable feature selection
... We study SVM-RFE , an algorithm well known for
its excellent generalization performance on high-dimensional
small-sample data. The main process of SVM-RFE is to
recursively eliminate features based on SVM, using the coefficients of the optimal decision boundary to measure the
relevance of each fe ...
Optimal Ensemble Construction via Meta-Evolutionary
... that trains each classifier on a randomly drawn training set. Each classifier’s
training set consists of the same number of examples randomly drawn from
the original training set, with the probability of drawing any given example
being equal. Samples are drawn with replacement, so that some examples ...
Online Full Text
... Various techniques exist for filtering spam. These
methods can be generally categorized into techniques
that have been influenced by artificial intelligence and
machine learning, and other techniques. These other
techniques tend to be older and less robust. For example,
use of white lists, black lis ...
Noise Tolerant Data Mining
... changed data entries make the succeeding data mining algorithms insufficient to discover the
genuine knowledge models. For many content sensitive domains, such as medical, financial,
or security databases, this kind of methods is simply not a good option. Second, most noise
handling methods take th ...
A Bayes Optimal Approach for Partitioning the Values of Categorical
... algorithm first sorts the categories according to the probability of the first class value, and then
searches for the best split in this sorted list. This algorithm has a time complexity of O(I log(I)),
where I is the number of categories. Based on the ideas presented in (Lechevallier, 1990; Fulton
Title An Evolutionary Approach to Automatic Kernel Construction
... This particular kernel tree was generated from experiments on the Ionosphere dataset. The diagram shows that the kernel tree is split into two parts, the vector and the
scalar tree. The inputs to the vector tree are the two samples, x and z, for which the kernel is being evaluated. These inputs are ...
Ensemble Learning Techniques for Structured
... classification models such as decision trees, artificial neural networks, Naïve Bayes, as well as many
other classifiers (Kim, 2009). Ensemble learning, based on aggregating the results from multiple
models, is a more sophisticated approach for increasing model accuracy as compared to the
no - CENG464
... This attribute minimizes the information needed to classify the tuples in the
resulting partitions and reflects the least randomness or impurity in these
Matching Conflicts: Functional Validation of Agents
... maybe only able to transform the input vector of some
certain dimensions. The actual numerical computations
carried out vary from algorithm to algorithm so that
different round-off errors are accumulated leading to
slightly different answers. Moreoverdifferent numerical
implementations of some basi ...
Combining Clustering with Classification for Spam Detection in
... clustering to reduce the training time of a classifier when dealing with large data
sets. In particular, while SVM classifiers (see  for a tutorial) have proved to be
a great success in many areas, their training time is at least O(N 2 ) for training
data of size N , which makes them non favourab ...
A Stochastic Algorithm for Feature Selection in Pattern Recognition
... The second approach (wrapper methods) is computationally demanding, but often is more accurate. A wrapper algorithm explores the space of features subsets to optimize the induction algorithm
that uses the subset for classification. These methods based on penalization face a combinatorial
challenge w ...
Inductive Intrusion Detection in Flow-Based
... Sequential forward selection which resulted in the selection of
three features, namely BAD. After each iteration, the feature
yielding the best intermediate error rate is added to the list
of features. . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Sequential backward elimination which resu ...
sociallocker - Projectsgoal
... Thereafter, we propose a dual training (DT) algorithm
and a dual prediction (DP) algorithm respectively, to
make use of the original and reversed samples in pairs for
training a statistical classifier and make predictions. In
DT, the classifier is learnt by maximizing a combination
of likelihoods of ...
Baseball Prediction Using Ensemble Learning by Arlo Lyle (Under
... In this thesis, I explored the use of machine learning techniques for baseball predictions
as opposed to the purely statistical methods that currently dominate the landscape. While
the idea of using machine learning to predict players’ statistics is not a new one, as seen by
the Vladimir system, the ...
Technical Note Naive Bayes for Regression
... Why does naive Bayes perform well even when the independence assumption is
seriously violated? Most likely it owes its good performance to the zero-one loss
function used in classification (Domingos & Pazzani, 1997). This function defines
the error as the number of incorrect predictions. Unlike othe ...
Medical Diagnosis with C4.5 Rule Preceded by Artificial
... artificial neural network ensemble, it is still impressive that
Table 1 indicates the generalization ability of C4.5 Rule-PANE
is about 23% (((.2726-.2306)/.2726 + (.1581-.1034)/.1581 +
(.0567-.0460)/.0567) / 3 = .2296) better than that of the wellestablished method C4.5 Rule on these three case stu ...
A Comparative Analysis of Classification with Unlabelled Data using
... reality, data is always in short supply. On one aspect, we would
like to use as much of the data as possible for training. But one
the other aspect, we want to use as much of it as possible for
testing. There already exist some technologies to deal with this
issue and it is still controversial till ...
portable document (.pdf) format
... Technical Details and an Applied Example
Many researchers have experienced nonconvergence errors where, for one reason
or another, a maximum likelihood solution can not be calculated or does not exist.
Conditions for nonconvergence include sparseness of data, multiple maximas,
unspecified boundary c ...
Cross-validation, sometimes called rotation estimation, is a model validation technique for assessing how the results of a statistical analysis will generalize to an independent data set. It is mainly used in settings where the goal is prediction, and one wants to estimate how accurately a predictive model will perform in practice. In a prediction problem, a model is usually given a dataset of known data on which training is run (training dataset), and a dataset of unknown data (or first seen data) against which the model is tested (testing dataset). The goal of cross validation is to define a dataset to ""test"" the model in the training phase (i.e., the validation dataset), in order to limit problems like overfitting, give an insight on how the model will generalize to an independent dataset (i.e., an unknown dataset, for instance from a real problem), etc.One round of cross-validation involves partitioning a sample of data into complementary subsets, performing the analysis on one subset (called the training set), and validating the analysis on the other subset (called the validation set or testing set). To reduce variability, multiple rounds of cross-validation are performed using different partitions, and the validation results are averaged over the rounds.Cross-validation is important in guarding against testing hypotheses suggested by the data (called ""Type III errors""), especially where further samples are hazardous, costly or impossible to collect.Furthermore, one of the main reasons for using cross-validation instead of using the conventional validation (e.g. partitioning the data set into two sets of 70% for training and 30% for test) is that the error (e.g. Root Mean Square Error) on the training set in the conventional validation is not a useful estimator of model performance and thus the error on the test data set does not properly represent the assessment of model performance. This may be because there is not enough data available or there is not a good distribution and spread of data to partition it into separate training and test sets in the conventional validation method. In these cases, a fair way to properly estimate model prediction performance is to use cross-validation as a powerful general technique.In summary, cross-validation combines (averages) measures of fit (prediction error) to correct for the optimistic nature of training error and derive a more accurate estimate of model prediction performance.