Modelling word meaning using efficient tensor representations

Models of word meaning, built from a corpus of text, have demonstrated success in emulating human performance on a number of cognitive tasks. Many of these models use geometric representations of words to store semantic associations between words. Often word order information is not captured in these models. The lack of structural information used by these models has been raised as a weakness when performing cognitive tasks. This paper presents an efficient tensor based approach to modelling word meaning that builds on recent attempts to encode word order information, while providing flexible methods for extracting task specific semantic information.

Discovering novel knowledge using granule mining

This paper presents an extended granule mining based methodology, to effectively describe the relationships between granules not only by traditional support and confidence, but by diversity and condition diversity as well. Diversity measures how diverse of a granule associated with the other granules, it provides a kind of novel knowledge in databases. We also provide an algorithm to implement the proposed methodology. The experiments conducted to characterize a real network traffic data collection show that the proposed concepts and algorithm are promising.

Discovering relevant features for effective query formulation

The quality of discovered features in relevance feedback (RF) is the key issue for effective search query. Most existing feedback methods do not carefully address the issue of selecting features for noise reduction. As a result, extracted noisy features can easily contribute to undesirable effectiveness. In this paper, we propose a novel feature extraction method for query formulation. This method first extract term association patterns in RF as knowledge for feature extraction. Negative RF is then used to improve the quality of the discovered knowledge. A novel information filtering (IF) model is developed to evaluate the proposed method. The experimental results conducted on Reuters Corpus Volume 1 and TREC topics confirm that the proposed model achieved encouraging performance compared to state-of-the-art IF models.

An ensemble method for predicting subnuclear localizations from primary protein structures

Background Predicting protein subnuclear localization is a challenging problem. Some previous works based on non-sequence information including Gene Ontology annotations and kernel fusion have respective limitations. The aim of this work is twofold: one is to propose a novel individual feature extraction method; another is to develop an ensemble method to improve prediction performance using comprehensive information represented in the form of high dimensional feature vector obtained by 11 feature extraction methods. Methodology/Principal Findings A novel two-stage multiclass support vector machine is proposed to predict protein subnuclear localizations. It only considers those feature extraction methods based on amino acid classifications and physicochemical properties. In order to speed up our system, an automatic search method for the kernel parameter is used. The prediction performance of our method is evaluated on four datasets: Lei dataset, multi-localization dataset, SNL9 dataset and a new independent dataset. The overall accuracy of prediction for 6 localizations on Lei dataset is 75.2% and that for 9 localizations on SNL9 dataset is 72.1% in the leave-one-out cross validation, 71.7% for the multi-localization dataset and 69.8% for the new independent dataset, respectively. Comparisons with those existing methods show that our method performs better for both single-localization and multi-localization proteins and achieves more balanced sensitivities and specificities on large-size and small-size subcellular localizations. The overall accuracy improvements are 4.0% and 4.7% for single-localization proteins and 6.5% for multi-localization proteins. The reliability and stability of our classification model are further confirmed by permutation analysis. Conclusions It can be concluded that our method is effective and valuable for predicting protein subnuclear localizations. A web server has been designed to implement the proposed method. It is freely available at http://bioinformatics.awowshop.com/snlpr​ed_page.php.

Quantum one-time programs

A one-time program is a hypothetical device by which a user may evaluate a circuit on exactly one input of his choice, before the device self-destructs. One-time programs cannot be achieved by software alone, as any software can be copied and re-run. However, it is known that every circuit can be compiled into a one-time program using a very basic hypothetical hardware device called a one-time memory. At first glance it may seem that quantum information, which cannot be copied, might also allow for one-time programs. But it is not hard to see that this intuition is false: one-time programs for classical or quantum circuits based solely on quantum information do not exist, even with computational assumptions. This observation raises the question, "what assumptions are required to achieve one-time programs for quantum circuits?" Our main result is that any quantum circuit can be compiled into a one-time program assuming only the same basic one-time memory devices used for classical circuits. Moreover, these quantum one-time programs achieve statistical universal composability (UC-security) against any malicious user. Our construction employs methods for computation on authenticated quantum data, and we present a new quantum authentication scheme called the trap scheme for this purpose. As a corollary, we establish UC-security of a recent protocol for delegated quantum computation.

Dynamic robot path planning using an enhanced simulated annealing approach

Evolutionary computation is an effective tool for solving optimization problems. However, its significant computational demand has limited its real-time and on-line applications, especially in embedded systems with limited computing resources, e.g., mobile robots. Heuristic methods such as the genetic algorithm (GA) based approaches have been investigated for robot path planning in dynamic environments. However, research on the simulated annealing (SA) algorithm, another popular evolutionary computation algorithm, for dynamic path planning is still limited mainly due to its high computational demand. An enhanced SA approach, which integrates two additional mathematical operators and initial path selection heuristics into the standard SA, is developed in this work for robot path planning in dynamic environments with both static and dynamic obstacles. It improves the computing performance of the standard SA significantly while giving an optimal or near-optimal robot path solution, making its real-time and on-line applications possible. Using the classic and deterministic Dijkstra algorithm as a benchmark, comprehensive case studies are carried out to demonstrate the performance of the enhanced SA and other SA algorithms in various dynamic path planning scenarios.

An efficient algorithm for the detection of Eden

In this paper, a polynomial time algorithm is presented for solving the Eden problem for graph cellular automata. The algorithm is based on our neighborhood elimination operation which removes local neighborhood configurations which cannot be used in a pre-image of a given configuration. This paper presents a detailed derivation of our algorithm from first principles, and a detailed complexity and accuracy analysis is also given. In the case of time complexity, it is shown that the average case time complexity of the algorithm is \Theta(n^2), and the best and worst cases are \Omega(n) and O(n^3) respectively. This represents a vast improvement in the upper bound over current methods, without compromising average case performance.

Certified pseudonyms colligated with master secret key

A pseudonym provides anonymity by protecting the identity of a legitimate user. A user with a pseudonym can interact with an unknown entity and be confident that his/her identity is secret even if the other entity is dishonest. In this work, we present a system that allows users to create pseudonyms from a trusted master public-secret key pair. The proposed system is based on the intractability of factoring and finding square roots of a quadratic residue modulo a composite number, where the composite number is a product of two large primes. Our proposal is different from previously published pseudonym systems, as in addition to standard notion of protecting privacy of an user, our system offers colligation between seemingly independent pseudonyms. This new property when combined with a trusted platform that stores a master secret key is extremely beneficial to an user as it offers a convenient way to generate a large number of pseudonyms using relatively small storage.

Database relation watermarking resilient against secondary watermarking attacks

There has been tremendous interest in watermarking multimedia content during the past two decades, mainly for proving ownership and detecting tamper. Digital fingerprinting, that deals with identifying malicious user(s), has also received significant attention. While extensive work has been carried out in watermarking of images, other multimedia objects still have enormous research potential. Watermarking database relations is one of the several areas which demand research focus owing to the commercial implications of database theft. Recently, there has been little progress in database watermarking, with most of the watermarking schemes modeled after the irreversible database watermarking scheme proposed by Agrawal and Kiernan. Reversibility is the ability to re-generate the original (unmarked) relation from the watermarked relation using a secret key. As explained in our paper, reversible watermarking schemes provide greater security against secondary watermarking attacks, where an attacker watermarks an already marked relation in an attempt to erase the original watermark. This paper proposes an improvement over the reversible and blind watermarking scheme presented in [5], identifying and eliminating a critical problem with the previous model. Experiments showing that the average watermark detection rate is around 91% even with attacker distorting half of the attributes. The current scheme provides security against secondary watermarking attacks.

Source code watermarking based on function dependency oriented sequencing

In the current market, extensive software development is taking place and the software industry is thriving. Major software giants have stated source code theft as a major threat to revenues. By inserting an identity-establishing watermark in the source code, a company can prove it's ownership over the source code. In this paper, we propose a watermarking scheme for C/C++ source codes by exploiting the language restrictions. If a function calls another function, the latter needs to be defined in the code before the former, unless one uses function pre-declarations. We embed the watermark in the code by imposing an ordering on the mutually independent functions by introducing bogus dependency. Removal of dependency by the attacker to erase the watermark requires extensive manual intervention thereby making the attack infeasible. The scheme is also secure against subtractive and additive attacks. Using our watermarking scheme, an n-bit watermark can be embedded in a program having n independent functions. The scheme is implemented on several sample codes and performance changes are analyzed.

On the effect of topology on cellular automata rule spaces

This thesis presents an empirical study of the effects of topology on cellular automata rule spaces. The classical definition of a cellular automaton is restricted to that of a regular lattice, often with periodic boundary conditions. This definition is extended to allow for arbitrary topologies. The dynamics of cellular automata within the triangular tessellation were analysed when transformed to 2-manifolds of topological genus 0, genus 1 and genus 2. Cellular automata dynamics were analysed from a statistical mechanics perspective. The sample sizes required to obtain accurate entropy calculations were determined by an entropy error analysis which observed the error in the computed entropy against increasing sample sizes. Each cellular automata rule space was sampled repeatedly and the selected cellular automata were simulated over many thousands of trials for each topology. This resulted in an entropy distribution for each rule space. The computed entropy distributions are indicative of the cellular automata dynamical class distribution. Through the comparison of these dynamical class distributions using the E-statistic, it was identified that such topological changes cause these distributions to alter. This is a significant result which implies that both global structure and local dynamics play a important role in defining long term behaviour of cellular automata.