Search

In the last decade, artificial intelligence and machine learning algorithms have been more and more established for the screening and detection of diseases and pathologies, as well as for describing interactions between measures where classical methods are too complex or fail. The purpose of this paper is to model the measured postoperative position of an intraocular lens implant after cataract surgery, based on preoperatively assessed biometric effect sizes using techniques of machine learning.In this study, we enrolled 249 eyes of patients who underwent elective cataract surgery at Augenklinik Castrop-Rauxel. Eyes were measured preoperatively with the IOLMaster 700 (Carl Zeiss Meditec), as well as preoperatively and postoperatively with the Casia 2 OCT (Tomey). Based on preoperative effect sizes axial length, corneal thickness, internal anterior chamber depth, thickness of the crystalline lens, mean corneal radius and corneal diameter a selection of 17 machine learning algorithms were tested for prediction performance for calculation of internal anterior chamber depth (AQD_post) and axial position of equatorial plane of the lens in the pseudophakic eye (LEQ_post).The 17 machine learning algorithms (out of 4 families) varied in root mean squared/mean absolute prediction error between 0.187/0.139 mm and 0.255/0.204 mm (AQD_post) and 0.183/0.135 mm and 0.253/0.206 mm (LEQ_post), using 5-fold cross validation techniques. The Gaussian Process Regression Model using an exponential kernel showed the best performance in terms of root mean squared error for prediction of AQDpost and LEQpost. If the entire dataset is used (without splitting for training and validation data), comparison of a simple multivariate linear regression model vs. the algorithm with the best performance showed a root mean squared prediction error for AQD_post/LEQ_post with 0.188/0.187 mm vs. the best performance Gaussian Process Regression Model with 0.166/0.159 mm.In this paper we wanted to show the principles of supervised machine learning applied to prediction of the measured physical postoperative axial position of the intraocular lenses. Based on our limited data pool and the algorithms used in our setting, the benefit of machine learning algorithms seems to be limited compared to a standard multivariate regression model.In den vergangenen Jahren wurden zunehmend Systeme der künstlichen Intelligenz in der Medizin etabliert, die Pathologien oder Erkrankungen erkennen und schwer zu erfassende Zusammenhänge zwischen Eingangsgrößen beschreiben sollen. Ziel der Studie ist es, anhand von biometrischen Messgrößen vor der Kataraktoperation die physikalische Position der Intraokularlinse nach der Operation mit Verfahren des maschinellen Lernens vorherzusagen.249 Augen von Patienten, die sich zur elektiven Kataraktoperation an der Augenklinik Castrop-Rauxel vorstellten, wurden mit dem IOLMaster 700 (Carl Zeiss Meditec) und vor sowie nach der Operation mit dem Casia 2 (Tomey) untersucht. Aus den Effektgrößen Augenlänge, Hornhautdicke, interne Vorderkammertiefe, Linsendicke, mittlerer Hornhautradius und Hornhautdurchmesser wurde eine Auswahl von 17 Algorithmen des Maschinenlernens auf ihre Vorhersagequalität hin getestet für die Bestimmung der postoperativen internen Vorderkammertiefe (AQD_post) und der Lage des Linsenäquators relativ zum Hornhautscheitel (LEQ_post).Die 17 Modelle (4 Algorithmusfamilien) variierten im mittleren quadratischen/mittleren absoluten Vorhersagefehler zwischen 0,187/0,139 mm und 0,255/0,204 mm (AQD_post) und 0,183/0,135 mm und 0,253/0,206 mm (LEQ_post) bei einer Kreuzvalidierung (5-fold cross validation). Der Algorithmus mit der besten Performance war bei beiden Vorhersagen ein Gaussian Process Regression Model mit einem exponentiellen Kernel. Prüft man am gesamten Datensatz (ohne Aufteilung in Trainings- und Validierungsdaten) ein einfaches multiples lineares Regressionsmodell gegen den Algorithmus mit der besten Performance, so ergibt sich bei der Vorhersage von AQD_post/LEQ_post ein mittlerer quadratischer Vorhersagefehler von 0,188/0,187 mm gegenüber dem Gaussian Process Regression Model mit 0,166/0,159 mm.In der Arbeit soll das Prinzip des überwachten Maschinenlernens in der Anwendung auf die Nachbildung der physikalisch gemessenen postoperativen axialen Position von Intraokularlinsen gezeigt werden. Bei den vorliegenden Daten ist der Gewinn durch die Verwendung von Algorithmen des maschinellen Lernens (bei den hier verwendeten Algorithmen) gegenüber einem multiplen linearen Regressionsansatz gering.

The present paper is a closer examination of that version of so-called "truth-theoretic semantics" which is developed by Larson and Segal (1995). Generally, the claim of truth-functional semantics is that the semantic meta-theory of an object language is exhausted by an adequate theory of truth for this language. Here the measure of adequacy is Tarski's celebrated "criterion T", which requires for each object language sentence A a sentence (of a special form) stating the truth-conditions for A. Larson and Segal combine the truth-theoretic view on semantics with a modular approach to linguistics. They stipulate a semantic module in which the truth-conditions for the object language sentences are derived. Since the "logic module" is, according to Larson and Segal, separated from the semantic module, these derivations can only make quite limited use of logical means. In the present paper I shall demonstrate that even in one of the most primitive sample theories which Larson and Segal develop in their monography, it is necessary to use much more of logic than they are ready to admit. Finally, I will set up and comment on a general formulation of the problem involved here, namely: how much logic is necessary in an adequate semantics.

Summary Individual-tree growth models are the new standard for modeling growth and yield. Their main purpose is to simulate future forest management scenarios but they can also be used to predict wood quality, rockfall protection or habitat quality. Individual tree growth models may consist of different models but core models are diameter increment, height increment, crown ratio (often used as a predictor for increment) and mortality. The model differentiation is based on how these four models include tree age (size), competition and site. Four common growth simulators in Central Europe are BWIN, Moses, Prognaus and Silva. These four models are commonly deployed to simulate 30 years of growth, but a prospective application is the simulation of a whole rotation period (80–150 years). It is therefore crucial to understand the possibilities and limitations of these models by evaluating them. This review paper summarizes the statistical and emergent properties’ evaluation results for these models. Statistical evaluations focus on individual models of a simulator, whereas the evaluation of emergent properties evaluates the entire simulator, by testing if the models conform to known principles of stand growth. Further, the meaning of these evaluation results for the development and improvement of individual-tree growth models is discussed.

Many authors have hoped to understand the indicative conditional construction in everyday language by means of what are usually called conditional probabilities. Other authors have hoped to make sense of conditional probabilities in terms of the absolute probabilities of conditional statements. Although all such hopes were disappointed by the triviality theorems of Lewis (1976), there have been copious subsequent attempts both to rescue CCCP (the conditional construal of conditional probability) and to extend and to intensify the arguments against it. In this paper it will be shown that triviality is avoidable if the probability function is replaced by an alternative generalization of the deducibility relation, the measure of deductive dependence of Miller and Popper (1986). It will be suggested further that this alternative way of orchestrating conditionals is nicely in harmony with the test proposed in Ramsey (1931), and also with the idea that it is not the truth value of a conditionalstatement that is of primary concern but its assertability or acceptability.

In this paper, we present a method for regularizing noisy 3D reconstructions, which is especially well suited for scenes containing planar structures like buildings. At horizontal structures, the input model is divided into slices and for each slice, an inside/outside labeling is computed. With the outlines of each slice labeling, we create an irregularly shaped volumetric cell decomposition of the whole scene. Then, an optimized inside/outside labeling of these cells is computed by solving an energy minimization problem. For the cell labeling optimization we introduce a novel smoothness term, where lines in the images are used to improve the regularization result. We show that our approach can take arbitrary dense meshed point clouds as input and delivers well regularized building models, which can be textured afterwards.

The genitive singular of the weak masculine nouns appears in three non-standard variations in addition to the standard suffix -(e)n: one variation is case augmenting (des Bär-en-s), one is stem affixing (des Polizist-s) and one lacks a case suffix (des Patient-Ø). Linguists as well as language critics have focussed mainly on the stem affixing genitive while more or less ignor¬ing the two other non-standard variations. This paper presents a corpus-based investigation regarding frequency and contexts of all three variations, also considering the diachronic back¬ground and the implications the results might have for predicting the further development of the declension class of the weak masculine nouns.

In the present paper, on the one hand, an analysis of agreement and number markers allows us to understand the disintegrated nature of the nominal classification system in the Dadjiriwale language. It turnes out that these markers show a rather phonological than grammatical agreement: While the singular is marked by a zero morpheme, it is difficult to identify "binary gender markers" or "ordered pairs". Rather, the notions of gender and noun classification become intertwined in the language. On the other hand, -sι and -ny ὶ are the plural suffixes of the noun class [+human], while the suffixes -jé and -bhó are those of the class [–human]; but furthermore, they denote divisible, and indivisible things respectively. Additionally, and analogous to the suffix-bhó, the markers -pέlώ and -kl ё́ divide objects into the group of round objects and the group of elongated objects. Finally, it is shown that the relationship established between the notion of noun classes and the notion of deixis is far from being synonymous or opposing, and that it uncovers the semantic values of the modal suffixes -sι and -nyὶ in the context of utterance: In speech, they evoke exophoric and homophoric deictic values and recall instances of interlocution and delocution.

ON SYSTEMATIC CHARACTER OF THE SEMANTIC STRUCTURE OF LITHUANIAN ZOOMORPHISMSSummaryIn the paper an attempt has been made to establish three types of systematic relations, namely, hierarchical, associative and associative-hierarchical, in the semantic structure of Lithuanian zoomorphisms, or animal terms in figurative use.The above relations exist in the meanings of one and the same zoomorphism as well as among all the zoomorphisms in question. They can be determined by the components of the semantic struc­ture under discussion, i. e. by the content plane and the forms of realization of the content plane in speech.

Idiosyncrasies in morphological agreement inside and outside phrases with numerals of the pięć ‘five’-type as a consequences of syntactic agreement failures The paper discusses some problems inherent in the hypothesis on which numerals are heads that have a valued feature of case in nominative and accusative contexts. In the analysis offered here numerals and quantifiers are phrasal adjuncts in the extended projection of the noun whose head is a null functional category. 3 person, singular number and neuter gender verb morphology as well as alternating patterns of the morphology of pre-numeral and predicative adjectives reflect attempted but (partly) failed syntactic agreement.

Simple color, intensity representations of polarimetric synthetic aperture radar (PolSAR) images fail to show the physical characteristics of the recorded ground objects, so several coherent and incoherent decomposition theorems have been proposed in the state-of-the-art literature. All these decompositions assume the fact that any scattering mechanism can be represented as the sum of some simpler, canonical" scattering mechanisms. Following the same assumption, in this paper we employ the independent component analysis (ICA) for PolSAR images representation. Since ICA is a method used for blind sources separation, we expect that the derived ICA channels represent as well as possible certain types of scattering mechanisms present in the image. ICA decomposition is validated against the coherent Pauli and the incoherent H/a/α decompositions. The normalized compression distance (NCD) is used as a measure of quality of decompositions. Experiments are made on a SLC L-band F-SAR image over Kaufbeuren airfield, Germany..

English neoclassical compounds rely on a distinct vocabulary stock and present morphological features which raise a number of theoretical questions. Generalisations about neoclassical compounds are also problematic because the output is by no means homogeneous, that is, defining features of neoclassical compounds sometimes co-exist with features that are not prototypical of these formations. The paper looks at neoclassical compounds with a view to exploring patterns of morphological behaviour and development in this class of compounds. The approach is both synchronic and diachronic: it researches whether the morphological behaviour of recently formed compounds is different from that of earlier compounds and, if so, in which respects. This is assessed on data from the BNC with respect to some of the features that are cited in the literature as defining properties of neoclassical compounds, specifically, their internal configuration, the occurrence or not of a linking vowel, and their productivity.

This paper deals with the semantic structures of the Event -ízo derivatives in Modern Greek, appearing in the syntactic frames NPi __ NP and NP __ . The present analysis incorporates a version of Ray Jackendoff's conceptual semantics (1983, 1990, 1992). Special attention is paid to the semantic under-determination of word-formation rules. Semantic fields, conceptual functions, formation rules, and mechanisms/rules involved in -ízo derivation are presented. A principled account of various ambiguous structures is also provided. In the last part, the conceptual structures in -ízo derivation are finally laid down and the question of keeping these structures minimal while extending the semantic fields is once more addressed.

This paper introduces a novel approach for estimating the numbers of endmembers in hyperspectral imagery. It exploits the geometrical properties of the noise hypersphere and considers the signal as outlier of the noise hypersphere. The proposed method, called outlier detection method (ODM), is automatic and non-parametric. In a principal component space, noise is spherically symmetric in all directions and lies on the surface of a hypersphere with a constant radius. Reversely, signal radiuses are much larger that noise radius and vary in all directions, thus signal lies in a hyperellipsoid. The proposed method involves three steps: 1) noise estimation; 2) minimum noise fraction transformation; and 3) outlier detection using inter quartile range. Estimation of the number of endmembers is accomplished by the estimation of the number of noise hypersphere outliers using a robust outlier detection method. The ODM was evaluated using simulated and real hyperspectral data, and it was also compared with well-known methods for estimating the number of endmembers. Evaluation of the method showed that the method produces robust and satisfactory results, and outperforms in relation to its competitors.

In this paper I focus on scope phenomena connected with semimodal (and modal) verbs and mainly on the syntactic behaviour of these groups of verbs. One important question is: why can semimodal verbs (and modal verbs in epistemic use) not have perfect and future tense forms? Taking among other things Reichenbach's tense system as a starting point I try to point out that the interpretation of a semimodal or a modal in epistemic use is problematic 1. if there is more than one reference time/if the reference time is indefinite or 2. if the verb in question stands together with an auxiliary of future which has a certain modal meaning itself. The comparison of the treatment of these phenomena in the framework of the Semantic Syntax with a non transformational approach (fragment of a categorial grammar) shows, that some important transformational rules and principles easily and economically can be represented in a non transformational grammar. The transformational approach needs rules like RAISING and LOWERING (or at least one of the two, and in addition to this a rather extended set of rules) for the generation of sentences, while in the categorial system we need only two reduction laws. It has to be investigated whether and to what extent the formation and transformation rules in a transformational grammar on the one hand and the dominance / linear precedence rules together with the lexical entries on the other hand are equivalent.

In this paper we give a modular account of the grammar of additive particles. In doing this we take issue with the standard descriptions of focus particles, which are based on just one possible pattern: the particle preceding the main stressed constituent it relates to (its RC). Additive particles, however, occur in a second, equally unmarked pattern: the RC preceding the main stressed particle. Former accounts do not only miss this complementary distribution as to position and stress pattern relative to the RC, but, as we demonstrate in detail, they misrepresent the relation between syntax, semantics and focus structure of these (and similar) particles in general. Using German auch as our prime example, we argue in particular (i) that there is just one auch underlying the ± stressed variants, and that the complementary distribution cannot be explained by a movement analysis; (ii) that the set of alternatives the auch proposition p and some contextually given proposition q induced by auch belong to, is not supplied by the focus structure of p but by comparing p and q; (iii) that the syntactic scope of auch is crucial for its semantics in that the adding operation applies to the material it contains, no matter whether it is the RC or predicative material common to p, q; (iv) that the complementary distribution of ± stressed auch follows from the modular interaction of the syntax and semantics of auch with focus structure; (v) that auch gives rise to two utterance meanings, 'in addition/furthermore' and 'likewise', directly correlating with whether or not the scope of auch contains RC material. What we argue, in short, is that so-called 'focus particles' are in reality 'scope particles'