Multilingual lexical resources to detect cognates in non-aligned texts

The identification of cognates between two distinct languages has recently start- ed to attract the attention of NLP re- search, but there has been little research into using semantic evidence to detect cognates. The approach presented in this paper aims to detect English-French cog- nates within monolingual texts (texts that are not accompanied by aligned translat- ed equivalents), by integrating word shape similarity approaches with word sense disambiguation techniques in order to account for context. Our implementa- tion is based on BabelNet, a semantic network that incorporates a multilingual encyclopedic dictionary. Our approach is evaluated on two manually annotated da- tasets. The first one shows that across different types of natural text, our method can identify the cognates with an overall accuracy of 80%. The second one, con- sisting of control sentences with semi- cognates acting as either true cognates or false friends, shows that our method can identify 80% of semi-cognates acting as cognates but also identifies 75% of the semi-cognates acting as false friends.

Protecting encrypted cookies from compression side-channel attacks

Compression is desirable for network applications as it saves bandwidth; however, when data is compressed before being encrypted, the amount of compression leaks information about the amount of redundancy in the plaintext. This side channel has led to successful CRIME and BREACH attacks on web traffic protected by the Transport Layer Security (TLS) protocol. The general guidance in light of these attacks has been to disable compression, preserving confidentiality but sacrificing bandwidth. In this paper, we examine two techniques - heuristic separation of secrets and fixed-dictionary compression|for enabling compression while protecting high-value secrets, such as cookies, from attack. We model the security offered by these techniques and report on the amount of compressibility that they can achieve.

Security Analysis of salt||password Hashes

Protection of passwords used to authenticate computer systems and networks is one of the most important application of cryptographic hash functions. Due to the application of precomputed memory look up attacks such as birthday and dictionary attacks on the hash values of passwords to find passwords, it is usually recommended to apply hash function to the combination of both the salt and password, denoted salt||password, to prevent these attacks. In this paper, we present the first security analysis of salt||password hashing application. We show that when hash functions based on the compression functions with easily found fixed points are used to compute the salt||password hashes, these hashes are susceptible to precomputed offline birthday attacks. For example, this attack is applicable to the salt||password hashes computed using the standard hash functions such as MD5, SHA-1, SHA-256 and SHA-512 that are based on the popular Davies-Meyer compression function. This attack exposes a subtle property of this application that although the provision of salt prevents an attacker from finding passwords, salts prefixed to the passwords do not prevent an attacker from doing a precomputed birthday attack to forge an unknown password. In this forgery attack, we demonstrate the possibility of building multiple passwords for an unknown password for the same hash value and salt. Interestingly, password||salt (i.e. salts suffixed to the passwords) hashes computed using Davies-Meyer hash functions are not susceptible to this attack, showing the first security gap between the prefix-salt and suffix-salt methods of hashing passwords.

On the leakage resilience of secure channel establishment

Secure communication channels are typically constructed from an authenticated key exchange (AKE) protocol, which authenticates the communicating parties and establishes shared secret keys, and a secure data transmission layer, which uses the secret keys to encrypt data. We address the partial leakage of communicating parties' long-term secret keys due to various side-channel attacks, and the partial leakage of plaintext due to data compression. Both issues can negatively affect the security of channel establishment and data transmission. In this work, we advance the modelling of security for AKE protocols by considering more granular partial leakage of parties' long-term secrets. We present generic and concrete constructions of two-pass leakage-resilient key exchange protocols that are secure in the proposed security models. We also examine two techniques--heuristic separation of secrets and fixed-dictionary compression--for enabling compression while protecting high-value secrets.

Class-specific sparse codes for representing activities

In this paper we investigate the effectiveness of class specific sparse codes in the context of discriminative action classification. The bag-of-words representation is widely used in activity recognition to encode features, and although it yields state-of-the art performance with several feature descriptors it still suffers from large quantization errors and reduces the overall performance. Recently proposed sparse representation methods have been shown to effectively represent features as a linear combination of an over complete dictionary by minimizing the reconstruction error. In contrast to most of the sparse representation methods which focus on Sparse-Reconstruction based Classification (SRC), this paper focuses on a discriminative classification using a SVM by constructing class-specific sparse codes for motion and appearance separately. Experimental results demonstrates that separate motion and appearance specific sparse coefficients provide the most effective and discriminative representation for each class compared to a single class-specific sparse coefficients.

Liberty

In terms of critical discourse, Liberty contributes to the ongoing aesthetic debate on ‘the sublime.’ Philosopher Immanuel Kant (1724–1804) defined the sublime as a failure of rationality in response to sensory overload: a state where the imagination is suspended, without definitive reference points—a state beyond unequivocal ‘knowing.’ I believe the events of September 11, 2001 eluded our understanding in much the same way, leaving us in a moment of suspension between awe and horror. It was an event that couldn’t be understood in terms of scope or scale. It was a moment of overload, which is so difficult to capture in art. With my work I attempt to rekindle that moment of suspension. Like the events of 9/11, Liberty defies definition. Its form is constantly changing; it is always presenting us with new layers of meaning. Nobody quite had a handle on the events that followed 9/11, because the implications were constantly shifting. In the same way, Liberty cannot be contained or defined at any moment in time. Like the events of 9/11, the full story cannot be told in a snapshot. One of the dictionary definitions for the word ‘sublime’ is the conversion of ‘a solid substance directly into a gas, without there being an intermediate liquid phase’. With this in mind, I would like to present Liberty as a work that is literally ‘sublime.’ But what’s really interesting to me about Liberty is that it presents the sublime on all levels: in its medium, in its subject matter (that moment of suspension), and in its formal (formless) presentation. On every level Liberty is sublime—subverting all tangible reference points and eluding capture entirely. Liberty is based on the Statue of Liberty in New York. However, unlike that statue which has stood in New York since 1886 and can be reasonably expected to stand for millennia, this work takes on diminishing proportions, carved as it is in carbon dioxide, a mysterious, previously unexplored medium—one which smokes, snows and dramatically vanishes into a harmless gas. Like the material this work is carved from, the civil liberties of the free world are diminishing fast, since 9/11 and before. This was my thought when I first conceived this work. Now it’s become evident that Liberty expresses a lot more than just this: it demonstrates the erosion of civil liberties, yes. However, it also presents the intangible, indefinable moments in the days and months that followed 9/11. The sculptural work will last for only a short time, and thereafter will exist only in documentation. During this time, the form is continually changing and self-refining, until it disappears entirely, to be inhaled, metabolised and literally taken to heart by viewers.

Human action recognition from video sequences

This PhD research has proposed new machine learning techniques to improve human action recognition based on local features. Several novel video representation and classification techniques have been proposed to increase the performance with lower computational complexity. The major contributions are the construction of new feature representation techniques, based on advanced machine learning techniques such as multiple instance dictionary learning, Latent Dirichlet Allocation (LDA) and Sparse coding. A Binary-tree based classification technique was also proposed to deal with large amounts of action categories. These techniques are not only improving the classification accuracy with constrained computational resources but are also robust to challenging environmental conditions. These developed techniques can be easily extended to a wide range of video applications to provide near real-time performance.