Search

This paper proposes improved algebraic attacks that are effective for lightweight block ciphers. Concretely, we propose a new framework that leverages on algebraic preprocessing as well as modern SAT solvers to perform algebraic cryptanalysis on block ciphers. By combining with chosen plaintext attacks, we show that our framework can be applied to lightweight block ciphers that exhibit a nice differential trail. In particular, we demonstrate our techniques by performing algebraic cryptanalysis on both the Present cipher and the Simon cipher. For the Present cipher, we successfully solved up to 9 rounds with at most 32 key bits fixed and 8 chosen plaintexts. On the other hand, for the Simon cipher, we tested our method on Simon-32/64 and Simon-64/128. For these two versions, our attack can solve up to 13 rounds with only 8 chosen plaintexts by fixing 4 and 6 key bits for Simon-32/64 and Simon-64/128, respectively. Further, by considering a class of weak keys, we can extend our attacks to 16 rounds. As far as we are aware, these are the best algebraic attacks on these ciphers in the literature.

Randomizing the execution of the sequence of operations in an algorithm is one of the most frequently considered solutions to improve the security of cryptographic implementations against sidechannel analysis. Such an algorithm for public-key cryptography was introduced by Tunstall at ACISP, 2009. In his right-to-left m-ary exponentiation algorithm, the radix-m digits of the exponent are treated in somewhat random order. This randomized solution will inhibit attacks that allow operations to be distinguished from one acquisition. In this article, we present a memory-efficient variant of Tunstall's random-order exponentiation algorithm, making it applicable to modular exponentiations in (Z/NZ)* (for instance, the RSA cryptosystem). The proposed algorithm requires only (m+1) memory registers instead of (m+r), where r > m as recommended in Tunstall's algorithm. Namely, the proposed algorithm saves about half the memory registers. Our analysis shows that our algorithm can be used as a supplement in order to defeat statistical side-channel analysis attacks, especially recent collision-correlation power analysis in the horizontal setting. Last but not least, we present a random order binary implementation, which is the first right-to-left binary implementation resisting attacks in the horizontal setting.

Algebraic differential fault attack (ADFA) is an attack in which an attacker combines a differential fault attack and an algebraic technique to break a targeted cipher. In this paper, we present three attacks using three different algebraic techniques combined with a differential fault attack in the bit-flip fault model to break the SIMON block cipher. First, we introduce a new analytic method which is based on a differential trail between the correct and faulty ciphertexts. This method is able to recover the entire master key of any member of the SIMON family by injecting faults into a single round of the cipher. In our second attack, we present a simplified Grobner basis algorithm to solve the faulty system. We show that this method could totally break SIMON ciphers with only 3 to 5 faults injected. Our third attack combines a fault attack with a modern SAT solver. By guessing some key bits and with only a single fault injected at the round $T - 6$T-6, where $T$T is the number of rounds of a SIMON cipher, this combined attack could manage to recover a master key of the cipher. For the last two attacks, we perform experiments to demonstrate the effectiveness of our attacks. These experiments are implemented on personal computers and run in very reasonable timing.

Micro Doppler (m-D) effect is a phenomenon that provides signatures to discriminate different moving objects. Accordingly, this paper presents a novel residual convolutional neural network that can classify different moving targets based on m-D analysis of reflected frequency modulation continuous wave (FMCW) radar signals. The proposed network is optimized through the experiments of varying number of residual blocks. As a result, the proposed network yields the average classification accuracy of \(93.48\%\) with five residual blocks, 64 filters per convolution layer, and the filter size of \(3\times 3\). Moreover, thanks to the residual connection, our network remarkably outperforms two other existing networks in terms of accuracy.

Interpreting the inner workings of neural models is a key step in ensuring the robustness and trustworthiness of the models, but work on neural network interpretability typically faces a trade-off: either the models are too constrained to be very useful, or the solutions found by the models are too complex to interpret. We propose a novel strategy for achieving interpretability that -- in our experiments -- avoids this trade-off. Our approach builds on the success of using probability as the central quantity, such as for instance within the attention mechanism. In our architecture, DoLFIn (Distributions over Latent Features for Interpretability), we do no determine beforehand what each feature represents, and features go altogether into an unordered set. Each feature has an associated probability ranging from 0 to 1, weighing its importance for further processing. We show that, unlike attention and saliency map approaches, this set-up makes it straight-forward to compute the probability with which an input component supports the decision the neural model makes. To demonstrate the usefulness of the approach, we apply DoLFIn to text classification, and show that DoLFIn not only provides interpretable solutions, but even slightly outperforms the classical CNN and BiLSTM text classifiers on the SST2 and AG-news datasets.

This paper demonstrates the performance evaluation of UAV detection based on micro-Doppler radar image data with the proposed inception-residual neural network (IRNN). Accordingly, the network is designed and analyzed by changing network hyper-parameters through experiment with the Real Doppler RAD-DAR (RDRD) dataset that is collected by the practical measurements. Numerical analysis results show that the proposed network with 16 filters yield a good trade-off between accuracy and time-consuming performances. Moreover, the network is taken into account for competing with three other networks. Due to inception-residual structure, the proposed network remarkably outperforms other ones.

Hierarchical Temporal Memory (HTM) is a continuously learning algorithm derived from neuroscience that models spatial and temporal streaming data. It was demonstrated that HTM produces a good performance in predicting unusual patterns or anomaly detection in univariate datasets. In this paper, we deploy the HTM algorithm for the anomaly detection problem in multivariate datasets, which are more common in practical scenarios. We first investigate the implementation of HTM using multi-encoders for multiple variables and analyze its performance in different parameter settings. Then, we introduce a new framework for ensemble learning by using single-encoder HTMs as weak learners. We carried out experiments on public datasets in different dimensions. Our experimental results show that our new approach outperforms the multi-encoder implementation of the HTM algorithm.

The advances of the Internet of Things (IoT) have had a fundamental impact and influence in sharping our rich living experiences. However, since IoT devices are usually resource-constrained, lightweight block ciphers have played a major role in serving as a building block for secure IoT protocols. In CHES 2015, SIMECK, a family of block ciphers, was designed for resource-constrained IoT devices. Since its publication, there have been many analyses on its security. In this paper, under the one bit-flip model, we propose a new efficient fault analysis attack on SIMECK ciphers. Compared to those previously reported attacks, our attack can recover the full master key by injecting faults into only a single round of all SIMECK family members. This property is crucial, as it is infeasible for an attacker to inject faults into different rounds of a SIMECK implementation on IoT devices in the real world. Specifically, our attack is characterized by exercising a deep analysis of differential trail between the correct and faulty immediate ciphertexts. Extensive simulation evaluations are conducted, and the results demonstrate the effectiveness and correctness of our proposed attack., Comment: 8 pages

Linkage Tree Genetic Algorithm (LTGA) is an effective Evolutionary Algorithm (EA) to solve complex problems using the linkage information between problem variables. LTGA performs well in various kinds of single-task optimization and yields promising results in comparison with the canonical genetic algorithm. However, LTGA is an unsuitable method for dealing with multi-task optimization problems. On the other hand, Multifactorial Optimization (MFO) can simultaneously solve independent optimization problems, which are encoded in a unified representation to take advantage of the process of knowledge transfer. In this paper, we introduce Multifactorial Linkage Tree Genetic Algorithm (MF-LTGA) by combining the main features of both LTGA and MFO. MF-LTGA is able to tackle multiple optimization tasks at the same time, each task learns the dependency between problem variables from the shared representation. This knowledge serves to determine the high-quality partial solutions for supporting other tasks in exploring the search space. Moreover, MF-LTGA speeds up convergence because of knowledge transfer of relevant problems. We demonstrate the effectiveness of the proposed algorithm on two benchmark problems: Clustered Shortest-Path Tree Problem and Deceptive Trap Function. In comparison to LTGA and existing methods, MF-LTGA outperforms in quality of the solution or in computation time.

Unsupervised relation extraction (URE) extracts relations between named entities from raw text without manually-labelled data and existing knowledge bases (KBs). URE methods can be categorised into generative and discriminative approaches, which rely either on hand-crafted features or surface form. However, we demonstrate that by using only named entities to induce relation types, we can outperform existing methods on two popular datasets. We conduct a comparison and evaluation of our findings with other URE techniques, to ascertain the important features in URE. We conclude that entity types provide a strong inductive bias for URE., Comment: 8 pages, 1 figure, 2 tables. Accepted in ACL 2020

A multi signature scheme allows a group of signers to produce a joint signature on a common message, which is more compact than a collection of distinct signatures from all signers. Given this signature and the list of signers' public keys, a verifier is able to check if every signer in the group participated in signing. Recently, a multisignature scheme with public key aggregation has drawn a lot of attention due to their applications into the blockchain technology. Such multisignatures provide not only a compact signature, but also a compact aggregated public key, that is both the signature size and the public key size used to verify the correctness of the signature are independent from the number of signers. This is useful for a blockchain because of its duplication over a distributed network, and thus it is required to be as compact as possible. In this paper, we introduce a new multisignature scheme with such a feature. Our scheme is proven secure under the Decisional Diffie-Hellman assumption. In addition, in the presence of rogue key attacks, the security of our scheme is proven in the plain public key model.

The ubiquitous deployment of Internet of Things (IoT) devices enhances connectivity and communication, and benefits almost every aspect of our lives from manufacturing to retail to smart homes. However, low levels of security protection in these devices due to their limited resources open opportunities for malicious users. An IoT forensics system collecting, processing, analyzing and reporting evidence of attack is required to mitigate the IoT security issues. Although such system has been studied over the past decade and solutions such as cloud-based IoT forensic were proposed, limitation still exist. In this paper, leveraging on the blockchain technology, we propose a per-missioned blockchain-based IoT forensics framework to enhance the integrity, authenticity and non-repudiation properties for the collected evidence. We formally define the system architecture, provide framework details, and propose a cryptographic-based approach to mitigate identity privacy concern.

Common-mode voltage is known as an undesirable voltage generated when applying switching techniques for voltage source inverters (VSIs). In electric drive systems based on pulse-width modulation converters, common-mode voltage causes bearing current, which may destroy bearing surface and reduce the mechanical system longlife. This paper presents a novel switching technique for a two-level VSI acts as the rotor side converter in a doubly-fed induction generator system. A simulation model is built in MATLAB/Simulink to verify the proposed algorithm.

Entity linking involves aligning textual mentions of named entities to their corresponding entries in a knowledge base. Entity linking systems often exploit relations between textual mentions in a document (e.g., coreference) to decide if the linking decisions are compatible. Unlike previous approaches, which relied on supervised systems or heuristics to predict these relations, we treat relations as latent variables in our neural entity-linking model. We induce the relations without any supervision while optimizing the entity-linking system in an end-to-end fashion. Our multi-relational model achieves the best reported scores on the standard benchmark (AIDA-CoNLL) and substantially outperforms its relation-agnostic version. Its training also converges much faster, suggesting that the injected structural bias helps to explain regularities in the training data., Comment: ACL 2018

Coreference evaluation metrics are hard to optimize directly as they are non-differentiable functions, not easily decomposable into elementary decisions. Consequently, most approaches optimize objectives only indirectly related to the end goal, resulting in suboptimal performance. Instead, we propose a differentiable relaxation that lends itself to gradient-based optimisation, thus bypassing the need for reinforcement learning or heuristic modification of cross-entropy. We show that by modifying the training objective of a competitive neural coreference system, we obtain a substantial gain in performance. This suggests that our approach can be regarded as a viable alternative to using reinforcement learning or more computationally expensive imitation learning., Comment: 10 pages. CoNLL

Nonstructural 5B (NS5B) polymerase and Nonstructural 3/4A (NS3/4A) protease have proven to be promising targets for the development of anti-HCV (Hepatitis C Virus) agents. The NS5B polymerase is of paramount importance in HCV viral replication; therefore, employing NS5B inhibitors was considered an effective way for the treatment of HCV. Identifying inhibitors against NS3/4A serine protease represents another attractive approach applied in anti-HCV drug discovery, which is evidenced by its crucial role of in the biogenesis of the viral replication activity. In this chapter, many different computational approaches including Quantitative Structure-Activity Relationship (QSAR) and virtual screening in anti-HCV drug discovery were considered and discussed in detail. Virtual Screening (VS) techniques, including ligand-based and structure-based, and QSAR have been utilized for the discovery of NS5B inhibitors. Moreover, using various in silico protocols and workflows, a number of studies have been conducted with an aim of identifying potential NS3/4A blockage agents.

We introduce an LSTM-based method for dynamically integrating several wordprediction experts to obtain a conditional language model which can be good simultaneously at several subtasks. We illustrate this general approach with an application to dialogue where we integrate a neural chat model, good at conversational aspects, with a neural question-answering model, good at retrieving precise information from a knowledge-base, and show how the integration combines the strengths of the independent components. We hope that this focused contribution will attract attention on the benefits of using such mixtures of experts in NLP and dialogue systems specifically.

We present a self-training approach to unsupervised dependency parsing that reuses existing supervised and unsupervised parsing algorithms. Our approach, called `iterated reranking' (IR), starts with dependency trees generated by an unsupervised parser, and iteratively improves these trees using the richer probability models used in supervised parsing that are in turn trained on these trees. Our system achieves 1.8% accuracy higher than the state-of-the-part parser of Spitkovsky et al. (2013) on the WSJ corpus., 11 pages

We propose solutions to enhance the Inside-Outside Recursive Neural Network (IORNN) reranker of Le and Zuidema (2014). Replacing the original softmax function with a hierarchical softmax using a binary tree constructed by combining output of the Brown clustering algorithm and frequency-based Huffman codes, we significantly reduce the reranker’s computational complexity. In addition, enriching contexts used in the reranker by adding subtrees rooted at (ancestors’) cousin nodes, the accuracy is increased.

According to the principle of compositionality, the meaning of a sentence is computed from the meaning of its parts and the way they are syntactically combined. In practice, however, the syntactic structure is computed by automatic parsers which are far-from-perfect and not tuned to the specifics of the task. Current recursive neural network (RNN) approaches for computing sentence meaning therefore run into a number of practical difficulties, including the need to carefully select a parser appropriate for the task, deciding how and to what extent syntactic context modifies the semantic composition function, as well as on how to transform parse trees to conform to the branching settings (typically, binary branching) of the RNN. This paper introduces a new model, the Forest Convolutional Network, that avoids all of these challenges, by taking a parse forest as input, rather than a single tree, and by allowing arbitrary branching factors. We report improvements over the state-of-the-art in sentiment analysis and question classification.

We are proposing an extension of the recursive neural network that makes use of a variant of the long short-term memory architecture. The extension allows information low in parse trees to be stored in a memory register (the `memory cell') and used much later higher up in the parse tree. This provides a solution to the vanishing gradient problem and allows the network to capture long range dependencies. Experimental results show that our composition outperformed the traditional neural-network composition on the Stanford Sentiment Treebank., 10 pages, 7 figures

We propose the first implementation of an infinite-order generative dependency model. The model is based on a new recursive neural network architecture, the Inside-Outside Recursive Neural Network. This architecture allows information to flow not only bottom-up, as in traditional recursive neural networks, but also topdown. This is achieved by computing content as well as context representations for any constituent, and letting these representations interact. Experimental results on the English section of the Universal Dependency Treebank show that the infinite-order model achieves a perplexity seven times lower than the traditional third-order model using counting, and tends to choose more accurate parses in k-best lists. In addition, reranking with this model achieves state-of-the-art unlabelled attachment scores and unlabelled exact match scores.

Ciet et al. (2006) proposed an elegant method for trading inversions for multiplications when computing [2]P+Q from two given points P and Q on elliptic curves of Weierstrass form. Motivated by their work, this paper proposes a fast algorithm for computing [4]P with only one inversion in affine coordinates. Our algorithm that requires 1I + 8S + 8M, is faster than two repeated doublings whenever the cost of one field inversion is more expensive than the cost of four field multiplications plus four field squarings (i.e. I > 4M + 4S). It saves one field multiplication and one field squaring in comparison with the Sakai-Sakurai method (2001). Even better, for special curves that allow "a = 0" (or "b = 0") speedup, we obtain [4]P in affine coordinates using just 1I + 5S + 9M (or 1I + 5S + 6M, respectively).

Timestamping is a cryptographic technique providing us with a proof-of-existence of a digital document at a given time. Combining both digital signature and provable time-stamping guarantees authentication, integrity and non-repudiation of electronic documents. In this paper, we introduce such a service, so called signtiming. Our scheme is based on an ID-based aggregate signature and is secure in the random oracle model.

The aim of timestamping systems is to provide a proof-of-existence of a digital document at a given time. Such systems are important to ensure integrity and non-repudiation of digital data over time. Most of the existing timestamping schemes use the notions of round (a period of time) and round token (a single value aggregating the timestamping requests received during one round). Such schemes have the following drawbacks: (i) Clients who have submitted a timestamping request must wait for the end of the round before receiving their timestamping certificate (ii) TimeStamping Authorities (TSA) based on such schemes are discrete-time systems and provide relative temporal authentication only, i.e. all the documents submitted during the same round are timestamped with the same date and time. (iii) the TSA can tamper timestamps before the round token is published in a widely distributed media. In this paper, we define a new timestamping scheme which overcomes these drawbacks.

Recommender systems provide personalized suggestions about items that users will find interesting. Typically, recommender systems require a user interface that can ``intelligently'' determine the interest of a user and use this information to make suggestions. The common solution, ``explicit ratings'', where users tell the system what they think about a piece of information, is well-understood and fairly precise. However, having to stop to enter explicit ratings can alter normal patterns of browsing and reading. A more ``intelligent'' method is to useimplicit ratings, where a rating is obtained by a method other than obtaining it directly from the user. These implicit interest indicators have obvious advantages, including removing the cost of the user rating, and that every user interaction with the system can contribute to an implicit rating.Current recommender systems mostly do not use implicit ratings, nor is the ability of implicit ratings to predict actual user interest well-understood. This research studies the correlation between various implicit ratings and the explicit rating for a single Web page. A Web browser was developed to record the user's actions (implicit ratings) and the explicit rating of a page. Actions included mouse clicks, mouse movement, scrolling and elapsed time. This browser was used by over 80 people that browsed more than 2500 Web pages.Using the data collected by the browser, the individual implicit ratings and some combinations of implicit ratings were analyzed and compared with the explicit rating. We found that the time spent on a page, the amount of scrolling on a page and the combination of time and scrolling had a strong correlation with explicit interest, while individual scrolling methods and mouse-clicks were ineffective in predicting explicit interest.