Machine learning methods can be used for solving important binary classification tasks in domains such as display advertising and recommender systems. In many of these domains categorical features are common and often of high cardinality. Using one-hot encoding in such circumstances lead to very high dimensional vector representations, causing memory and computability concerns for machine learning models. This thesis investigated the viability of a binary encoding scheme in which categorical values were mapped to integers that were then encoded in a binary format. This binary scheme allowed for representing categorical features using log2(d)-dimensional vectors, where d is the dimension associated with a one-hot encoding. To evaluate the performance of the binary encoding, it was compared against one-hot and feature hashed representations with the use of linear logistic regression and neural networks based models. These models were trained and evaluated using data from two publicly available datasets: Criteo and Census. The results showed that a one-hot encoding with a linear logistic regression model gave the best performance according to the PR-AUC metric. This was, however, at the expense of using 118 and 65,953 dimensional vector representations for Census and Criteo respectively. A binary encoding led to a lower performance but used only 35 and 316 dimensions respectively. For Criteo, binary encoding suffered significantly in performance and feature hashing was perceived as a more viable alternative. It was also found that employing a neural network helped mitigate any loss in performance associated with using binary and feature hashed representations. / Maskininlärningsmetoder kan användas för att lösa viktiga binära klassificeringsuppgifter i domäner som displayannonsering och rekommendationssystem. I många av dessa domäner är kategoriska variabler vanliga och ofta av hög kardinalitet. Användning av one-hot-kodning under sådana omständigheter leder till väldigt högdimensionella vektorrepresentationer. Detta orsakar minnesoch beräkningsproblem för maskininlärningsmodeller. Denna uppsats undersökte användbarheten för ett binärt kodningsschema där kategoriska värden var avbildade på heltalvärden som sedan kodades i ett binärt format. Detta binära system tillät att representera kategoriska värden med hjälp av log2(d) -dimensionella vektorer, där d är dimensionen förknippad med en one-hot kodning. För att utvärdera prestandan för den binära kodningen jämfördes den mot one-hot och en hashbaserad kodning. En linjär logistikregression och ett neuralt nätverk tränades med hjälp av data från två offentligt tillgängliga dataset: Criteo och Census, och den slutliga prestandan jämfördes. Resultaten visade att en one-hot kodning med en linjär logistisk regressionsmodell gav den bästa prestandan enligt PR-AUC måttet. Denna metod använde dock 118 och 65,953 dimensionella vektorrepresentationer för Census respektive Criteo. En binär kodning ledde till en lägre prestanda generellt, men använde endast 35 respektive 316 dimensioner. Den binära kodningen presterade väsentligt sämre specifikt för Criteo datan, istället var hashbaserade kodningen en mer attraktiv lösning. Försämringen i prestationen associerad med binär och hashbaserad kodning kunde mildras av att använda ett neuralt nätverk.

Software security is a big data problem. The volume of new software artifacts created far outpaces the current capacity of software analysis. This gap has brought an urgent challenge to our security community—scalability. If our techniques cannot cope with an ever increasing volume of software, we will always be one step behind attackers. Thus developing scalable analysis to bridge the gap is essential. In this dissertation, we argue that automatic code reuse detection enables an efficient data reduction of a high volume of incoming malware for downstream analysis and enhances software security by efficiently finding known vulnerabilities across large code bases. In order to demonstrate the benefits of automatic software similarity detection, we discuss two representative problems that are remedied by scalable analysis: malware triage and unpatched code clone detection. First, we tackle the onslaught of malware. Although over one million new malware are reported each day, existing research shows that most malware are not written from scratch; instead, they are automatically generated variants of existing malware. When groups of highly similar variants are clustered together, new malware more easily stands out. Unfortunately, current systems struggle with handling this high volume of malware. We scale clustering using feature hashing and perform semantic analysis using co-clustering. Our evaluation demonstrates that these techniques are an order of magnitude faster than previous systems and automatically discover highly correlated features and malware groups. Furthermore, we design algorithms to infer evolutionary relationships among malware, which helps analysts understand trends over time and make informed decisions about which malware to analyze first. Second, we address the problem of detecting unpatched code clones at scale. When buggy code gets copied from project to project, eventually all projects will need to be patched. We call clones of buggy code that have been fixed in only a subset of projects unpatched code clones. Unfortunately, code copying is usually ad-hoc and is often not tracked, which makes it challenging to identify all unpatched vulnerabilities in code basesat the scale of entire OS distributions. We scale unpatched code clone detection to spot over15,000 latent security vulnerabilities in 2.1 billion lines of code from the Linux kernel, allDebian and Ubuntu packages, and all C/C++ projects in SourceForge in three hours on asingle machine. To the best of our knowledge, this is the largest set of bugs ever reported in a single paper.

This thesis deals with a comparison of classification methods. At first, these classification methods based on machine learning are described, then a classifier comparison system is designed and implemented. This thesis also describes some classification tasks and datasets on which the designed system will be tested. The evaluation of classification tasks is done according to standard metrics. In this thesis is presented design and implementation of a classifier that is based on the principle of evolutionary algorithms.