Probabilistic evaluation of time series models : a comparison of several approaches

Several methods are examined which allow to produce forecasts for time series in the form of probability assignments. The necessary concepts are presented, addressing questions such as how to assess the performance of a probabilistic forecast. A particular class of models, cluster weighted models (CWMs), is given particular attention. CWMs, originally proposed for deterministic forecasts, can be employed for probabilistic forecasting with little modification. Two examples are presented. The first involves estimating the state of (numerically simulated) dynamical systems from noise corrupted measurements, a problem also known as filtering. There is an optimal solution to this problem, called the optimal filter, to which the considered time series models are compared. (The optimal filter requires the dynamical equations to be known.) In the second example, we aim at forecasting the chaotic oscillations of an experimental bronze spring system. Both examples demonstrate that the considered time series models, and especially the CWMs, provide useful probabilistic information about the underlying dynamical relations. In particular, they provide more than just an approximation to the conditional mean.

Brill Greek, Adobe Arno Pro, Adobe Text/Brevier Greek typefaces

Typeface design: a series of collaborative projects commissioned by Adobe, Inc. and Brill to develop extensive polytonic Greek typefaces. The two Adobe typefaces can be seen as extension of previous research for the Garamond Premier Pro family (2005), and concludes a research theme started in 1998 with work for Adobe’s Minion Pro Greek. These typefaces together define the state of the art for text-intensive Greek typesetting for wide character set texts (from classical texts, to poetry, to essays, to prose). They serve both as exemplar for other developers, and as vehicles for developing the potential of Greek text typography, for example with the parallel inclusion of monotonic and polytonic characters, detailed localised punctuation options, fluid handling of case-conversion issues, and innovative options such as accented small caps (originally requested by bibliographers, and subsequently rolled out to a general user base). The Brill typeface (for the established academic publisher) has an exceptionally wide character set to cover several academic disciplines, and is intended to differentiate sufficiently from its partner Latin typeface, while maintaining a clear texture in both offset and low-resolution print-on-demand reproduction. This work involved substantial amounts of testing and modifying the design, especially of diacritics, to maintain clarity the readability of unfamiliar words. All together these typefaces form a study in how Greek typesetting meets contemporary typographic requirements, while resonating with historically accurate styles, where these are present. Significant research in printing archives helped to identify appropriate styles, as well as originate variants that are coherent stylistically, even when historical equivalents were absent.

CUP Intermediate Greek-English Lexicon

Document design and typeface design: A typographic specification for a new Intermediate Greek-English Lexicon by CUP, accompanied by typefaces modified for the specific typographic requirements of the text. The Lexicon is a substantial (over 1400 pages) publication for HE students and academics intended to complement Liddell-Scott (the standard reference for classical Greek since the 1850s), and has been in preparation for over a decade. The typographic appearance of such works has changed very little since the original editions, largely to the lack of suitable typefaces: early digital proofs of the Lexicon utilised directly digitised versions of historical typefaces, making the entries difficult to navigate, and the document uneven in typographic texture. Close collaboration with the editors of the Lexicon, and discussion of the historical precedents for such documents informed the design at all typographic levels to achieve a highly reader-friendly results that propose a model for this kind of typography. Uniquely for a work of this kind, typeface design decisions were integrated into the wider document design specification. A rethinking of the complex typography for Greek and English based on historical editions as well as equivalent bilingual reference works at this level (from OUP, CUP, Brill, Mondadori, and other publishers) led a redefinition of multi-script typeface pairing for the specific context, taking into account recent developments in typeface design. Specifically, the relevant weighting of elements within each entry were redefined, as well as the typographic texture of type styles across the two scripts. In details, Greek typefaces were modified to emphasise clarity and readability, particularly of diacritics, at very small sizes. The relative weights of typefaces typeset side-by-side were fine-tuned so that the visual hierarchy of the entires was unambiguous despite the dense typesetting.

A time-series method to identify and correct range sidelobes in meteorological radar data

The use of pulse compression techniques to improve the sensitivity of meteorological radars has become increasingly common in recent years. An unavoidable side-effect of such techniques is the formation of ‘range sidelobes’ which lead to spreading of information across several range gates. These artefacts are particularly troublesome in regions where there is a sharp gradient in the power backscattered to the antenna as a function of range. In this article we present a simple method for identifying and correcting range sidelobe artefacts. We make use of the fact that meteorological targets produce an echo which fluctuates at random, and that this echo, like a fingerprint, is unique to each range gate. By cross-correlating the echo time series from pairs of gates therefore we can identify whether information from one gate has spread into another, and hence flag regions of contamination. In addition we show that the correlation coefficients contain quantitative information about the fraction of power leaked from one range gate to another, and we propose a simple algorithm to correct the corrupted reflectivity profile.

Reliability comparison of transmit/receive diversity and error control coding in low-power medium access control protocols

Low-power medium access control (MAC) protocols used for communication of energy constraint wireless embedded devices do not cope well with situations where transmission channels are highly erroneous. Existing MAC protocols discard corrupted messages which lead to costly retransmissions. To improve transmission performance, it is possible to include an error correction scheme and transmit/receive diversity. It is possible to add redundant information to transmitted packets in order to recover data from corrupted packets. It is also possible to make use of transmit/receive diversity via multiple antennas to improve error resiliency of transmissions. Both schemes may be used in conjunction to further improve the performance. In this study, the authors show how an error correction scheme and transmit/receive diversity can be integrated in low-power MAC protocols. Furthermore, the authors investigate the achievable performance gains of both methods. This is important as both methods have associated costs (processing requirements; additional antennas and power) and for a given communication situation it must be decided which methods should be employed. The authors’ results show that, in many practical situations, error control coding outperforms transmission diversity; however, if very high reliability is required, it is useful to employ both schemes together.

Tensor LRR and sparse coding-based subspace clustering

Subspace clustering groups a set of samples from a union of several linear subspaces into clusters, so that the samples in the same cluster are drawn from the same linear subspace. In the majority of the existing work on subspace clustering, clusters are built based on feature information, while sample correlations in their original spatial structure are simply ignored. Besides, original high-dimensional feature vector contains noisy/redundant information, and the time complexity grows exponentially with the number of dimensions. To address these issues, we propose a tensor low-rank representation (TLRR) and sparse coding-based (TLRRSC) subspace clustering method by simultaneously considering feature information and spatial structures. TLRR seeks the lowest rank representation over original spatial structures along all spatial directions. Sparse coding learns a dictionary along feature spaces, so that each sample can be represented by a few atoms of the learned dictionary. The affinity matrix used for spectral clustering is built from the joint similarities in both spatial and feature spaces. TLRRSC can well capture the global structure and inherent feature information of data, and provide a robust subspace segmentation from corrupted data. Experimental results on both synthetic and real-world data sets show that TLRRSC outperforms several established state-of-the-art methods.

A game theoretical approach to the evolution of structured communication codes

Structured meaning-signal mappings, i.e., mappings that preserve neighborhood relationships by associating similar signals with similar meanings, are advantageous in an environment where signals are corrupted by noise and sub-optimal meaning inferences are rewarded as well. The evolution of these mappings, however, cannot be explained within a traditional language evolutionary game scenario in which individuals meet randomly because the evolutionary dynamics is trapped in local maxima that do not reflect the structure of the meaning and signal spaces. Here we use a simple game theoretical model to show analytically that when individuals adopting the same communication code meet more frequently than individuals using different codes-a result of the spatial organization of the population-then advantageous linguistic innovations can spread and take over the population. In addition, we report results of simulations in which an individual can communicate only with its K nearest neighbors and show that the probability that the lineage of a mutant that uses a more efficient communication code becomes fixed decreases exponentially with increasing K. These findings support the mother tongue hypothesis that human language evolved as a communication system used among kin, especially between mothers and offspring.

Prosodic Morphology : Gender in Arabic Perfect Active and Passive 3rd Person Singular Verbs

Prosodic /template Morphology, that "draws heavily on the theoretical apparatus and formalisms of the generative phonology model known as autosegmental phonology" (Katamba, F. 1993: 154), is the best analysis that can handle Arabic morphology. Verbs in Arabic are represented on three independent tiers: root tier, the skeletal tier and the vocalic melody tier (Katamba, F. 1993). Vowel morphemes, which are represented by diacritics, are inserted within the consonant morphemes, which are represented by primary symbols, to form words. The morpheme tier hypothesis paves the way to understand the nonconcatenative Arabic morphology. This paper analyzes gender in perfect active and passive 3rd person singular verbs on the basis of PM. The focus of the analysis shall be drawn heavily on the most common Arabic verbs; triconsonantal verbs, with brief introduction of the less common verbs; quadriconsonantal perfect active and passive masculine and feminine 3rd person singular verbs. I shall, too, cast the light on some vowel changes that some verbs undergo when voice changes.

Sounding alien, touching the future : beyond the sonorous limit in science fiction film

In this paper I examine two particular aspects of sounding science fiction film: first, the ulterior, Othering sounds of the alien, whether it is creature, object, technology or environment; and second, the soundscape that accompanies or underscores the type of space travel that crosses temporal and spatial thresholds. In both instances of sounding science fiction film I suggest that human limits are reached and breached, leading to a deterritorialization of the self and a hearing that touches the future which is a moment of pure becoming. I focus on the womanly sonority of the alien to suggest that patriarchal and heterosexist sound devices can be ultimately corrupted. In the analysis of sounding space travel I suggest that science film can create a series of moments in which one experiences the double sublime. This spectacular rendering of a liquid chaos enables the viewer to experience the logic of sensation beyond bodily integrity. In this paper my over-arching position is one that hears in science fiction film the profound potential of a radical alterity that exists beyond the sonorous limit.

Improved subspace clustering via exploitation of spatial constraints

We present a novel approach to improving subspace clustering by exploiting the spatial constraints. The new method encourages the sparse solution to be consistent with the spatial geometry of the tracked points, by embedding weights into the sparse formulation. By doing so, we are able to correct sparse representations in a principled manner without introducing much additional computational cost. We discuss alternative ways to treat the missing and corrupted data using the latest theory in robust lasso regression and suggest numerical algorithms so solve the proposed formulation. The experiments on the benchmark Johns Hopkins 155 dataset demonstrate that exploiting spatial constraints significantly improves motion segmentation.

Bunker-de-bunk 2012

Image reduction operators based on non-monotonic averaging functions

Image reduction is a crucial task in image processing, underpinning many practical applications. This work proposes novel image reduction operators based on non-monotonic averaging aggregation functions. The technique of penalty function minimisation is used to derive a novel mode-like estimator capable of identifying the most appropriate pixel value for representing a subset of the original image. Performance of this aggregation function and several traditional robust estimators of location are objectively assessed by applying image reduction within a facial recognition task. The FERET evaluation protocol is applied to confirm that these non-monotonic functions are able to sustain task performance compared to recognition using nonreduced images, as well as significantly improve performance on query images corrupted by noise. These results extend the state of the art in image reduction based on aggregation functions and provide a basis for efficiency and accuracy improvements in practical computer vision applications.

Transfer learning using the online Fuzzy Min-Max neural network

In this paper, we present an empirical analysis on transfer learning using the Fuzzy Min–Max (FMM) neural network with an online learning strategy. Three transfer learning benchmark data sets, i.e., 20 Newsgroups, WiFi Time, and Botswana, are used for evaluation. In addition, the data samples are corrupted with white Gaussian noise up to 50 %, in order to assess the robustness of the online FMM network in handling noisy transfer learning tasks. The results are analyzed and compared with those from other methods. The outcomes indicate that the online FMM network is effective for undertaking transfer learning tasks in noisy environments.

Fuzzy measures of pixel cluster compactness

Pixel-scale fine details are often lost during image processing tasks such as image reduction and filtering. Block or region based algorithms typically rely on averaging functions to implement the required operation and traditional function choices struggle to preserve small, spatially cohesive clusters of pixels which may be corrupted by noise. This article proposes the construction of fuzzy measures of cluster compactness to account for the spatial organisation of pixels. We present two construction methods (minimum spannning trees and fuzzy measure decomposition) to generate measures with specific properties: monotonicity with respect to cluster size; invariance with respect to translation, reflection and rotation; and, discrimination between pixel sets of fixed cardinality with different spatial arrangements. We apply these measures within a non-monotonic mode-like averaging function used for image reduction and we show that this new function preserves pixel-scale structures better than existing monotonie averages.

On the weak monotonicity of Gini means and other mixture functions

Weak monotonicity was recently proposed as a relaxation of the monotonicity condition for averaging aggregation, and weakly monotone functions were shown to have desirable properties when averaging data corrupted with outliers or noise. We extended the study of weakly monotone averages by analyzing their ϕ-transforms, and we established weak monotonicity of several classes of averaging functions, in particular Gini means and mixture operators. Mixture operators with Gaussian weighting functions were shown to be weakly monotone for a broad range of their parameters. This study assists in identifying averaging functions suitable for data analysis and image processing tasks in the presence of outliers.