Fast and Space-Efficient Construction of AVL Grammars from the LZ77 Parsing.

A grammar compression is a context-free grammar (CFG) deriving a single string deterministically. For an input string of length N over an alphabet of size sigma, the smallest CFG is O(log N)-approximable in the offline setting and O(log N log^* N)-approximable in the online setting. In addition, an information-theoretic lower bound for representing a CFG in Chomsky normal form of n variables is log (n!/n^sigma) + n + o(n) bits. Although there is an online grammar compression algorithm that directly computes the succinct encoding of its output CFG with O(log N log^* N) approximation guarantee, the problem of optimizing its working space has remained open. We propose a fully-online algorithm that requires the fewest bits of working space asymptotically equal to the lower bound in O(N log log n) compression time. In addition we propose several techniques to boost grammar compression and show their efficiency by computational experiments.

Given a string T of length N, the goal of grammar compression is to construct a small context-free grammar generating only T. Among existing grammar compression methods, RePair (recursive paring) [Larsson and Moffat, 1999] is notable for achieving good compression ratios in practice. In this paper, we propose the first RePair algorithm working in compressed space, i.e., potentially o(N) space for highly compressible texts. The key idea is to give a new way to restructure an arbitrary (context-free) grammar S for T into RePair(T) in compressed space and time. We propose an algorithm for RePair(T) running in O(min(N, nm log N)) space and expected O(min(N, nm log N) m) time or O(min(N, nm log N) log log N) time, where n is the size of S and m is the number of variables in RePair(T). We implemented our O(min(N, nm log N) m)-time algorithm and show it can actually run in compressed space. We also present a new approach to reduce the peak memory usage of existing RePair algorithms combining with our algorithms, and show that the new approach outperforms, both in computation time and space, the most space efficient linear-time RePair implementation to date.

Recent advances in random linear systems on finite fields have paved the way for the construction of constant-time data structures representing static functions and minimal perfect hash functions using less space with respect to existing techniques. The main obstruction for any practical application of these results is the cubic-time Gaussian elimination required to solve these linear systems: despite they can be made very small, the computation is still too slow to be feasible. In this paper we describe in detail a number of heuristics and programming techniques to speed up the resolution of these systems by several orders of magnitude, making the overall construction competitive with the standard and widely used MWHC technique, which is based on hypergraph peeling. In particular, we introduce broadword programming techniques for fast equation manipulation and a lazy Gaussian elimination algorithm. We also describe a number of technical improvements to the data structure which further reduce space usage and improve lookup speed. Our implementation of these techniques yields a minimal perfect hash function data structure occupying 2.24 bits per element, compared to 2.68 for MWHC-based ones, and a static function data structure which reduces the multiplicative overhead from 1.23 to 1.03.

The article describes our solution approach for the coach trip with shuttle service problem, a passenger transportation problem introduced in the context of the VeRoLog Solver Challenge 2015, an implementation competition of the EURO Working Group on Vehicle Routing and Logistics Optimization. Our algorithm applies concepts known from Variable Neighborhood Search and Iterated Local Search. On a lower level, we consider the fast construction and modification of tree data structures, as those structures may serve as appropriate representations of (feasible) alternatives. Experiments are carried out, and a comparison to other approaches is given. We also make the source code of our approach (i.e., its computer implementation) available with this article: https://doi.org/10.17632/662mtv6sd8.1. © 2017 Wiley Periodicals, Inc. NETWORKS, Vol. 69(3), 329–345 2017

H.264/AVC is the newest video coding standard with high compression efficiency. Especially it adopts traveling algorithm to find the optimal mode for intra-frame prediction in spatial field, so it is quite complex and costs very long encoding-time. In this paper, the proposed fast mode decision scheme consists of two parts: early intra mode detection, which is to select intra 4 × 4 or intra 16 × 16 mode for each MB before encoding according to smoothness characteristic, and fast intra4 × 4 mode decision, which is to reduce choices of prediction directions from 9 to 3 by SAD of Mode 0 to Mode 2. Experimental results show that without the variety of PSNR basically, the compression time is 32.8% less than the original algorithm, and only increases 3.2% bits-rate on average. This algorithm improves the intra-frame coding efficiency, and is especially applicable for the high quality and complex sequence.

H.264/AVC is the newest video coding standard with high compression efficiency. But because it adopts the traveling algorithm to find the optimal mode for intra-frame prediction in the spatial field, it is quite complex and costs very long time to encode. In this paper, according to the ID-histogram algorithm, it is possible to select Intra 4x4 or Intra 16x16 mode for every macro-block before encode them. If choose the Intra 4 x4 mode, then use the fast prediction algorithm based on the characteristic of the macro- block itself to reduce the choices of the prediction directions from nine to three, thereby it improves the intra-frame coding efficiency. Experimental results show that without the variety of PSNR basically, the compression time is 32.8% less than the original algorithm, and increases 3.2% bits-rate on average. This algorithm is especially applicable for the high quality and complex image.

H.264/AVC is the newest video coding standard with high compression efficiency. Especially it adopts traveling algorithm to find the optimal mode for intra-frame prediction in spatial field, so it is quite complex and costs very long encoding-time. In this paper, the proposed fast mode decision scheme consists of two parts: early intra mode detection, which is to select intra 4×4 or intra 16×16 mode for each MB before encoding according to smoothness characteristic, and fast intra4×4 mode decision, which is to reduce choices of prediction directions from 9 to 3 by SAD of Mode 0 to Mode 2. Experimental results show that without the variety of PSNR basically, the compression time is 32.8% less than the original algorithm, and only increases 3.2% bits-rate on average. This algorithm improves the intra-frame coding efficiency, and is especially applicable for the high quality and complex sequence.

Remote sensing has significantly progressed and are becoming more readily available than ever. There is a growing requirement for the automatic recognition and labelling of these images under various conditions. Many studies have applied deep learning networks like CNN to satellite image recognition. However, the cost of the models is high, which means they are inappropriate when computing resources are limited, for instance, satellites and mobile devices. Moreover, the quality of available satellite images varies, and the ability of the network to adapt to low-quality images is also critical. To solve these problems, this paper uses mobile-net V2 and traditional CNN networks to classify 21 kinds of satellite images from a public dataset and compare the results. Firstly, the dataset is increased from 100 to 500 per class through data augmentation. Secondly, mobile-net V2 is trained and then the performance is evaluated using the test set. Additionally, the quality of the images is reduced to figure out the influence of each model. To verify the effectiveness and accuracy of mobile-net V2, several traditional CNN networks are compared with validation data accuracy, test data accuracy, inference time, peak RAM, and flash usage. The experimental results show that the mobile-net V2 network is a low-cost and well adaptability model with high accuracy for this multiclassification work.

The morphology and distribution of airway tree abnormalities enable diagnosis and disease characterisation across a variety of chronic respiratory conditions. In this regard, airway segmentation plays a critical role in the production of the outline of the entire airway tree to enable estimation of disease extent and severity. Furthermore, the segmentation of a complete airway tree is challenging as the intensity, scale/size and shape of airway segments and their walls change across generations. The existing classical techniques either provide an undersegmented or oversegmented airway tree, and manual intervention is required for optimal airway tree segmentation. The recent development of deep learning methods provides a fully automatic way of segmenting airway trees; however, these methods usually require high GPU memory usage and are difficult to implement in low computational resource environments. Therefore, in this study, we propose a data-centric deep learning technique with big interpolated data, Interpolation-Split, to boost the segmentation performance of the airway tree. The proposed technique utilises interpolation and image split to improve data usefulness and quality. Then, an ensemble learning strategy is implemented to aggregate the segmented airway segments at different scales. In terms of average segmentation performance (dice similarity coefficient, DSC), our method (A) achieves 90.55%, 89.52%, and 85.80%; (B) outperforms the baseline models by 2.89%, 3.86%, and 3.87% on average; and (C) produces maximum segmentation performance gain by 14.11%, 9.28%, and 12.70% for individual cases when (1) nnU-Net with instant normalisation and leaky ReLU; (2) nnU-Net with batch normalisation and ReLU; and (3) modified dilated U-Net are used respectively. Our proposed method outperformed the state-of-the-art airway segmentation approaches. Furthermore, our proposed technique has low RAM and GPU memory usage, and it is GPU memory-efficient and highly flexible, enabling it to be deployed on any 2D deep learning model.

The main purpose of this chapter is to propose a novel boundary element modeling and simulation algorithm for solving fractional bio-thermomechanical problems in anisotropic soft tissues. The governing equations are studied on the basis of the thermal wave model of bio-heat transfer (TWMBT) and Biot’s theory. These governing equations are solved using the boundary element method (BEM), which is a flexible and effective approach since it deals with more complex shapes of soft tissues and does not need the internal domain to be discretized, also, it has low RAM and CPU usage. The transpose-free quasi-minimal residual (TFQMR) solver are implemented with a dual-threshold incomplete LU factorization technique (ILUT) preconditioner to solve the linear systems arising from BEM. Numerical findings are depicted graphically to illustrate the influence of fractional order parameter on the problem variables and confirm the validity, efficiency and accuracy of the proposed BEM technique.

The primary goal of this article is to implement a dual reciprocity boundary element method (DRBEM) to analyze problems of rotating functionally graded anisotropic fiber-reinforced magneto-thermoelastic composites. To solve the governing equations in the half-space deformation model, an implicit–implicit scheme was utilized in conjunction with the DRBEM because of its advantages, such as dealing with more complex shapes of fiber-reinforced composites and not requiring the discretization of the internal domain. So, DRBEM has low RAM and CPU usage. As a result, it is adaptable and effective for dealing with complex fiber-reinforced composite problems. For various generalized magneto-thermoelasticity theories, transient temperature, displacements, and thermal stresses have been computed numerically. The numerical results are represented graphically to demonstrate the effects of functionally graded parameters and rotation on magnetic thermal stresses in the fiber direction. To validate the proposed method, the obtained results were compared to those obtained using the normal mode method, the finite difference method, and the finite element method. The outcomes of these three methods are extremely consistent.

Boundary element modeling and simulation of biothermomechanical behavior in anisotropic laser-induced tissue hyperthermia

We consider fully-online construction of indexing data structures for multiple texts. Let T = {T_1, ..., T_K} be a collection of texts. By fully-online, we mean that a new character can be appended to any text in T at any time. This is a natural generalization of semi-online construction of indexing data structures for multiple texts in which, after a new character is appended to the kth text T_k, then its previous texts T_1, ..., T_k-1 will remain static. Our fully-online scenario arises when we maintain dynamic indexes for multi-sensor data. Let N and sigma denote the total length of texts in T and the alphabet size, respectively. We first show that the algorithm by Blumer et al. [Theoretical Computer Science, 40:31-55, 1985] to construct the directed acyclic word graph (DAWG) for T can readily be extended to our fully-online setting, retaining O(N log sigma)-time and O(N)-space complexities. Then, we give a sophisticated fully-online algorithm which constructs the suffix tree for T in O(N log sigma) time and O(N) space. A key idea of this algorithm is synchronized maintenance of the DAWG and the suffix tree.

Fast scalable construction of ([compressed] static | minimal perfect hash) functions

Humans have the extraordinary ability to learn continually from experience. Not only we can apply previously learned knowledge and skills to new situations, we can also use these as the foundation for later learning. One of the grand goals of Artificial Intelligence (AI) is building an artificial “continual learning” agent that constructs a sophisticated understanding of the world from its own experience through the autonomous incremental development of ever more complex knowledge and skills (Parisi, 2019). However, despite early speculations and few pioneering works (Ring, 1998; Thrun, 1998; Carlson, 2010), very little research and effort has been devoted to address this vision. Current AI systems greatly suffer from the exposure to new data or environments which even slightly differ from the ones for which they have been trained for (Goodfellow, 2013). Moreover, the learning process is usually constrained on fixed datasets within narrow and isolated tasks which may hardly lead to the emergence of more complex and autonomous intelligent behaviors. In essence, continual learning and adaptation capabilities, while more than often thought as fundamental pillars of every intelligent agent, have been mostly left out of the main AI research focus. In this tutorial, we propose to summarize the application of these ideas in light of the more recent advances in machine learning research and in the context of deep architectures for AI (Lomonaco, 2019). Starting from a motivation and a brief history, we link recent Continual Learning advances to previous research endeavours on related topics and we summarize the state-of-the-art in terms of major approaches, benchmarks and key results. In the second part of the tutorial we plan to cover more exploratory studies about Continual Learning with low supervised signals and the relationships with other paradigms such as Unsupervised, Semi-Supervised and Reinforcement Learning. We will also highlight the impact of recent Neuroscience discoveries in the design of original continual learning algorithms as well as their deployment in real-world applications. Finally, we will underline the notion of continual learning as a key technological enabler for Sustainable Machine Learning and its societal impact, as well as recap interesting research questions and directions worth addressing in the future.