Margin Features Research Articles

Large Margin Feature Weighting Method via Linear Programming

The problem of feature selection is a difficult combinatorial task in machine learning and of high practical relevance. In this paper, we consider feature selection method for multimodally distributed data, and present a large margin feature weighting method for k-nearest neighbor (kNN) classifiers. The method learns the feature weighting factors by minimizing a cost function, which aims at separating different classes by large local margins and pulling closer together points from the same class, based on using as few features as possible. The consequent optimization problem can be efficiently solved by linear programming. Finally, the proposed approach is assessed through a series of experiments with UCI and microarray data sets, as well as a more specific and challenging task, namely, radar high-resolution range profiles (HRRP) automatic target recognition (ATR). The experimental results demonstrate the effectiveness of the proposed algorithms.

Second-order estimating equations for the analysis of clustered current status data

With clustered event time data, interest most often lies in marginal features such as quantiles or probabilities from the marginal event time distribution or covariate effects on marginal hazard functions. Copula models offer a convenient framework for modeling. We present methods of estimating the baseline marginal distributions, covariate effects, and association parameters for clustered current status data based on second-order generalized estimating equations. We examine the efficiency gains realized from using second-order estimating equations compared with first-order equations, issues of copula misspecification, and apply the methods to motivating studies including one on the incidence of joint damage in patients with psoriatic arthritis.

Application of pure brilliance imaging in diagnosis of benign and malignant breast neoplasms

Objective To evaluate the application of pure brilliance imaging(PB1) in diagnosis of benign and malignant breast neoplasms. Methods Ninty-seven breast neoplasms were respectively examined by high frequency ultrasound and PBI before operation,then correlated with pathological results. Results The accuracy of high frequency ultrasound for breast neoplasms was 73.2% with the sensitivity 77.8% ,specificity 69.2%. The detectable rate of spiculate margin, halo,calcification were much higher in PBI than those in high frequency ultrasound(P<0.05). And the accuracy,sensitivity and specificity of PBI were 86.6%, 91.1%, 82.6%. Conclusions PBI can significantly increase the detectable rate of ultrasonographie marginal features of breast neoplasms which were also helpful to the differential diagnosis. Key words: Ultrasonography; Breast disease

The influence of Congo River discharges in the surface and deep layers of the Gulf of Guinea

The main feature of the Congo-Angola margin in the Gulf of Guinea is the Congo (ex-Zaire) deep-sea fan composed of a submarine canyon directly connected to the Congo River, a channel and a [sediment] lobe area. During the multi-disciplinary programme called BIOZAIRE conducted by Ifremer from 2000 to 2005, two CTD-O2 sections with discrete water column samples were performed (BIOZAIRE3 cruise: 2003–2004) to study the influence of the Congo River discharges, both in the surface layer and in the deep and near-bottom layers. The surface layer water is greatly diluted with river water that has a heavy particle load. The deep layer is affected by episodic turbidity currents that flow in the deep Congo channel and reach deep areas far from the coast. Previous studies revealed deep anomalies in oxygen (deficit) and nutrient (excess) concentrations at not, vert, similar4000 m depth and assumed that they resulted from mineralisation of the particulate organic matter from the Congo River. The BIOZAIRE3 sections were designed to explore these phenomena in more detail near the Congo channel. Oxygen and nutrients were measured as well as additional parameters, including stable isotopes of oxygen and carbon, dissolved inorganic carbon and pH. For the surface layer, the effect of the Congo River was studied with reference to salinity. Deviations from the theoretical dilution of various inorganic solutes suggested the occurrence of mineralisation and consumption processes. For the deep layer, the network of CTD-O2 stations gave a more detailed description of the deep anomalies than in previous studies. From the east–west section, anomalies appeared on the bottom at 4000 m depth and became slightly shallower when they spread to the west. They were also present north and south on the bottom along the 4000 m isobath. In these deep waters, the decrease in the δ13C values of dissolved inorganic carbon confirmed that the mineralisation of organic matter plays a role in generating these anomalies. The location of the origin of this deep anomaly is debated. Here, arguments are given in favour of mineralisation of the particulate organic matter input that overflows from the Congo channel at not, vert, similar4000 m depth during turbidity current events. Other authors suggest that this input comes from downslope particle transport. Anomalies of the same origin, but weaker, also occurred deeper on the Congo lobe, where the Congo channel ends, but with a significant pH decrease on the bottom which was not seen at 4000 m depth.

The Alchemy of Accomplishing Medicine (sman sgrub): Situating the Yuthok Heart Essence (G.yu thog snying thig) in Literature and History

This essay examines historical and contemporary connections between Buddhist and medical traditions through a study of the Accomplishing Medicine (sman sgrub) practice and the Yuthok Heart Essence (G.yu thog snying thig) anthology. Accomplishing Medicine is an esoteric Buddhist yogic and contemplative exercise focused on several levels of “alchemical” transformation. The article will trace the acquisition of this practice from India by Tibetan medical figures and its assimilation into medical practice. It will propose that this alchemical practice forms the central nexus of connection between Tibetan medicine and the Buddhist Nyingma tradition, and that this little-studied link is not a marginal feature of Tibetan medicine but rather one that has had a significant shaping factor on each tradition throughout history.

Depth of focus enhancement for sub-&lt;inline-formula&gt;&lt;math display="inline" overflow="scroll"&gt;&lt;mrow&gt;&lt;mn&gt;110&lt;/mn&gt;&lt;mtext&gt;-&lt;/mtext&gt;&lt;mi&gt;nm&lt;/mi&gt;&lt;/mrow&gt;&lt;/math&gt;&lt;/inline-formula&gt; technology by using KrF double-exposure lithography

Traditional double-exposure lithography (DEL) or double-patterning lithography (DPL) methodologies stem most from the resolution enhancement standpoint. A single mask with high feature densities is split into two exposure steps, each with lower feature densities that can be easily resolved. The DEL is proposed as the process window enhancement technology for sub-110-nm technology. Features with sparse pitches are printed by a first step of dense pitch exposures and a second exposure with dummy features removed. The pattern decomposition strategy described is similar to that of subresolution assisting features (SRAF). So it is compatible with the traditional rule-based SRAF implementation methodology. By comparing the depth of focus (DOF) of the 110-nm lithography process between the single exposure and the double exposure, it is found that the DOF for marginal features is extended by using double-exposure methodology, and thus extends the capability of KrF exposure tools. Furthermore, the link between the overlay performance and the overlap of the second exposure's trim slots over the first exposure is studied. The results show that the overlay control is within the KrF scanner capability. As a further study, the proposed double-exposure methodology for the 90-nm lithography process is evaluated.

Two-Dimensional Maximum Margin Feature Extraction for Face Recognition

On face recognition, most previous works on dimensionality reduction and classification would first transform the input image into 1-D vector, which ignores the underlying data structure and often leads to the small sample size problem. More recently, 2-D discriminant analysis has become an interesting technique which can overcome the aforementioned drawbacks. However, 2-D methods extract features based on the rows or the columns of all images, so it is possible that the features using 2-D methods still contain some redundant information. In addition, most existing 2-D methods cannot provide an automatic strategy to choose discriminant vectors. In this paper, we study the combination of 2-D discriminant analysis and 1-D discriminant analysis and propose a two-stage framework: " (2D)(2)MMC + LDA." Because the extracted features based on maximal margin criterion (MMC) is robust, stable, and efficient, in the first stage, a 2-D two-directional feature extraction technique, (2D)(2)MMC , is presented. In the second stage, the linear discriminant analysis (LDA) step is performed in the (2D)(2)MMC subspace. Experiments with Feret, Olivetti and Oracle Research Laboratory, and Carnegie Mellon University Pose, Illumination, and Expression databases are conducted to evaluate our method in terms of classification accuracy and robustness.

Robust Estimation of Mean Functions and Treatment Effects for Recurrent Events Under Event-Dependent Censoring and Termination: Application to Skeletal Complications in Cancer Metastatic to Bone

In clinical trials featuring recurrent clinical events, the definition and estimation of treatment effects involves a number of interesting issues, especially when loss to follow-up may be event-related and when terminal events such as death preclude the occurrence of further events. This paper discusses a clinical trial of breast cancer patients with bone metastases where the recurrent events are skeletal complications, and where patients may die during the trial. We argue that treatment effects should be based on marginal rate and mean functions. When recurrent event data are subject to event-dependent censoring, however, ordinary marginal methods may yield inconsistent estimates. Incorporating correctly specified inverse probability of censoring weights into analyses can protect against dependent censoring and yield consistent estimates of marginal features. An alternative approach is to obtain estimates of rate and mean functions from models that involve some conditioning to render censoring conditionally independent. We consider three methods of estimating mean functions of recurrent event processes and examine the bias and efficiency of unweighted and inverse probability weighted versions of the methods with and without a terminating event. We compare the methods via simulation and use them to analyse the data from the breast cancer trial.

Abstract B58: Assessment of models for combining colposcopy features for cervical cancer diagnosis

Abstract B58 Despite the introduction of vaccines to prevent two types of high-risk oncogenic strains of the human papillomavirus (HPV), screening and diagnosis will still be important for the prevention of cervical cancer. Colposcopy is used to visualize the cervix to identify abnormal lesions for biopsy, which in turn determines disease severity and management. Recent studies have shown that colposcopy is not as sensitive as once believed, and colposcopy is a critical step in the prevention of cervical cancer. The HPV vaccine will likely result in a decline in the number of cases referred to colposcopy due to HPV 16 and 18. HPV 16 or 18 infection confers a high risk of precancer or cancer, and may represent cases that are easier to diagnosis. A reduction in the number of these cases may result in colposcopy becoming more challenging, further highlighting the need to optimize colposcopy. During colposcopic interpretation, physicians use the features of color, margins, mosacisim, and punctation, to determine a diagnosis for a patient and which areas to biopsy. How the features are combined to make these decisions significantly impacts the performance of colposcopy and biopsy as a diagnostic tool for cervical cancer. With the changing population referred to colposcopy, it becomes even more important to understand how to best utilize colposcopic features and colposcopy as a diagnostic tool. While we cannot predict the population that will be referred to colposcopy after widespread introduction of the HPV vaccine, patients diagnosed with atypical squamous cells of undetermined significance (ASCUS) and low-grade squamous intraepithelial (LSIL) Pap smears represent challenging populations for colposcopic diagnosis. We analyzed the performance of colposcopy using image annotations from 21 colposcopists evaluating digitized Cervigrams, which are magnified images of the cervix, from a sub-study of the ASCUS LSIL Triage Study (ALTS). We randomly selected one-third of the patients enrolled in the sub-study, stratified by final pathology and HPV status, for this evaluation. Computerized models for combining colposcopic features may improve the performance of colposcopists. We explore the performance of different models for combining colposcopic features, including a summed index score similar to the Reid index and logistic regression models, in comparison to the gold standard of diagnosis from pathology, to determine if better methods are available for combining colposcopic features than the colposcopic diagnosis provided by the colposcopist. The colposcopic diagnosis is the implicit aggregation of features performed by the colposcopist. To assess the performance of the various models, we use prediction probability as the figure of merit. Prediction probability is a generalization of the area under the receiver operating characteristic curve from binary to multi-level disease outcomes, allowing us to assess outcomes of cervical intraepithelial neoplasia 0, 1, and 2+. There is variability in performance among colposcopists, with a handful performing better than either the summed index score or logistic regression models. The driving forces of the variability among colposcopists need to be further elucidated to determine if there is additional information that should be used in colposcopy to improve the overall performance of colposcopists in general practice. Citation Information: Cancer Prev Res 2008;1(7 Suppl):B58.

High-resolution computed tomographic findings of small peripherally located squamous cell carcinoma

With the spread of high-resolution computed tomography (HRCT) screening for lung cancer, we are increasingly faced with the need to determine whether certain small lesions are benign or malignant. The features of small adenocarcinomas have been clarified but not those of squamous cell carcinoma. The objective of our study was therefore to clarify the HRCT findings of peripherally located squamous cell carcinomas less than 2 cm in maximum dimension. Subjects consisted of 27 consecutive pathologically proven cases of peripherally located squamous cell carcinoma that were less than 2 cm in maximum dimension. HRCT findings of all 27 cases were analyzed retrospectively and independently by three radiologists who were unaware of the pathological diagnosis, and decisions were reached by consensus with special attention to 10 review points. Internal characteristic features included calcification, cavity formation, and air bronchogram. Tumor margin features included spiculation, notching, irregularity, and ground-glass opacity. Surrounding structural features consisted of pleural indentation, pulmonary emphysema, and satellite lesions. The presence of irregularity (70.4%), surrounding pulmonary emphysema (70.4%), and pleural indentation (51.9%) was observed frequently. No mass was accompanied by calcification. HRCT images of peripherally located squamous cell carcinoma suggested that the demonstration of irregularity, surrounding pulmonary emphysema, pleural indentation, and absence of calcification may contribute to the accurate CT diagnosis of small peripheral squamous cell carcinoma.

The Acquisition of Past Participle Agreement in Québec French L1

Much developmental work has been devoted to the acquisition of object clitics in French. There is a consensus that in early grammar, children omit object clitics in contexts where an adult would not. Several analyses have been put forth, among which, one proposing a close link between the omission of object clitics and the presence of past participle agreement. In this short article we address this hypothesis by examining the acquisition of past participle agreement through two sentence-preference tasks administered to children (3- to 5-year-olds divided into three age groups) and adults from the Montréal area. The results of the experiments show that past participle agreement is a marginal feature in early grammar and therefore too weak to be considered as controlling the realization of the direct object clitics in child production. Children and adults display different behavior with respect to past participle agreement (weak and optional) versus subject-verb agreement (consistently strong and obligatory). This result points toward an analysis of past participle agreement in terms of language-specific optional verb movement.

Bone Tumors and Tumorlike Conditions: Analysis with Conventional Radiography

The approach to the radiographic diagnosis of bone tumors consists of analyzing the lesion in an organized fashion, with attention to the specific radiographic features of tumor location, margins, and zone of transition; periosteal reaction; mineralization; size and number of lesions; and presence of a soft-tissue component. Patient age is also an important clinical factor in the diagnosis of bone tumors, because various lesions have predilections for specific age groups.

パルスレーザー堆積法により作製したモリブデン酸化物光記録膜の光記録特性

Approximately 30 nm thick MoO3 films were deposited on glass or polycarbonate substrates by Pulsed Laser Deposition Method using ArF excimer laser (λ=193 nm). For films deposited on glass substrates, transmittance change between the as-deposited and the annealed (340°C for 10 min) states was about 40% at the wavelength of 405 nm. This increase in transmittance was presumably caused by a transformation of oxygen-deficient as-deposited state into oxygen-sufficient annealed state through an annealing-induced reaction involving oxygen absorption, which was confirmed by XRD spectra. The value of CNR for 30 nm thick film grown on polycarbonates was about 50 dB at the write power of 3.0~11.0 mW. The value of CNR for 30 nm thick film with the protection layer of 30 nm was 50 dB at the peak power of 9.0~17.0 mW. These results give a characteristic feature of the wide write-power margin.

Elaeocarpus sedentarius sp. nov. (Elaeocarpaceae)—morphometric analysis of a new, rare species from eastern Australia

To assess the status of a putative new species of Elaeocarpus L. (Elaeocarpaceae) from north-eastern New South Wales (NSW), with respect to the morphologically similar E. blepharoceras Schltr. from New Guinea, we undertook morphometric analysis of 11 vegetative attributes measured on 11 specimens of the putative new species and eight of E. blepharoceras. Cluster analysis (flexible UPGMA) and ordination (PCC) separates highland specimens of E. blepharoceras from the NSW material plus lowland E. blepharoceras. Furthermore, the ordination shows some separation of the NSW material into Koonyum Range and Nightcap Range groups. Although it is not clearly differentiated from lowland E. blepharoceras on morphometric analysis, description of the NSW material as E. sedentarius D.J.Maynard &amp; Crayn is justified by (1) additional features such as the anther awns (short and sparsely hairy in E. sedentarius and much longer and densely bristly in E. blepharoceras), variation in the number of locules (3(–4) in E. sedentarius and (2–)3 in E. blepharoceras), leaf margin features (short setae terminating veins on leaf margins of E. blepharoceras, lacking in E. sedentarius) and (2) geographic isolation (&gt;2000 km) is likely to prevent gene flow. The distribution, relationships and conservation status of E. sedentarius are discussed.

Recursive Support Vector Machines for Dimensionality Reduction

The usual dimensionality reduction technique in supervised learning is mainly based on linear discriminant analysis (LDA), but it suffers from singularity or undersampled problems. On the other hand, a regular support vector machine (SVM) separates the data only in terms of one single direction of maximum margin, and the classification accuracy may be not good enough. In this letter, a recursive SVM (RSVM) is presented, in which several orthogonal directions that best separate the data with the maximum margin are obtained. Theoretical analysis shows that a completely orthogonal basis can be derived in feature subspace spanned by the training samples and the margin is decreasing along the recursive components in linearly separable cases. As a result, a new dimensionality reduction technique based on multilevel maximum margin components and then a classifier with high accuracy are achieved. Experiments in synthetic and several real data sets show that RSVM using multilevel maximum margin features can do efficient dimensionality reduction and outperform regular SVM in binary classification problems.

Age Relationships of Laurentide and Montane Glaciations, Mackenzie Mountains, Northwest Territories

The Mackenzie Mountains were glaciated repeatedly by large valley glaciers that emanated from the Backbone Ranges, and by much smaller valley glaciers that emanated from peaks in the Canyon Ranges. During the Late Wisconsinan the Laurentide Ice Sheet reached its all-time maximum position. The ice sheet pressed against the Canyon Ranges and moved up major valleys causing the diversion of mountain waters and organizing a complex meltwater system that drained across mountain interfluve areas towards the northwest. Two ages of moraines deposited by montane glaciers occur widely in the Mackenzie Mountains. Near the mountain front certain of the older moraines have been truncated by the Laurentide Ice Sheet, and others have been incised by meltwater streams emanating from the Laurentide ice margin, indicating that these older moraines predate the maximum Laurentide advance. Locally, certain of the younger montane moraines breach moraines and other ice marginal features of the Laurentide maximum, indicating that the younger montane glaciation post-dated the Laurentide maximum. Some large montane glaciers extended out from the mountains to merge with the retreating Laurentide Ice Sheet. There are several localities that display the age relationships between montane and Laurentide glaciations such as Dark Rock Creek, Durkan-Lukas Valley, Little Bear River and Katherine Creek. The older of the local montane glaciations is correlated tentatively with Reid Glaciation (lllinoian?) of central Yukon, and the younger with the Late Wisconsinan McConnell Glaciation. The Laurentide Glaciation is correlated with Hungry Creek Glaciation of Bonnet Plume Depression, which probably culminated about 30,000 years ago or somewhat later.

Revisiting problem of chilled margins associated with marginal reversals in mafic–ultramafic intrusive bodies

Numerous observations on mafic–ultramafic layered intrusions, sills and dykes show that chilled margins always develop as an integral part of their marginal reversals and possess the following features: (a) they are commonly much more evolved or primitive than bulk intrusion compositions, (b) evolved chilled margins are composed of the low temperature cotectic assemblages of relevant magmatic systems and (c) tend to be compositionally similar in intrusions formed from different parental magmas, (d) fine-grained chilled margins are notably absent in many intrusions, with contact rocks being represented by medium- to coarse-grained cumulates. The anomalous features of chilled margins can be partly attributed to contamination, intratelluric inhomogeneity of magma, changes in composition of intruding magma, loss of magma from the chamber, supercooling, etc. A major process still remains, however, illusive, but appears to be universally operating along the cooling margins of magmatic bodies in a liquid state, being gravity-independent and temperature gradient-driven. We recognize this not yet specified process as Soret fractionation and explain the above observations in the following way. Primary chilled margins do not commonly survive because of intensive remelting by heat flux from the interior of the chamber. The subsequently formed “secondary chilled margins” represent cumulates that crystallized from liquids produced by temperature gradient-driven Soret fractionation. At high temperature gradients the process tends to produce similar cotectic liquids crystallizing gabbronorite (or gabbro) from all parental magmas of a given magmatic system, resulting in compositionally similar “secondary chilled margins” that are more evolved than bulk compositions. At low temperature gradients the process produces liquids that are only slightly more fractionated than the parental magma and form “secondary chilled margins” that are more primitive than bulk compositions. This interpretation suggests that, apart from the rare cases of chilled margins that survived remelting, they should not be used as monitors for parental magma compositions of intrusive bodies, even if all conventional complicating factors were not operative.

Mega-pockmarks and linear pockmark trains on the West African continental margin

Seabed pockmarks, the manifestation of the natural process of fluid escape at the seabed, are a widespread feature of the equatorial West African continental margin. Pockmarks occur singly, in small groups, in large random fields and in organized arrays or ‘pockmark trains’. Pockmark trains are associated with areas of steeper seabed gradient and evolve though time to form deep gullies. Pockmark gullies may exceed 1 km in width and extend for 10–20 km down slope, and form through the interaction of slope failure and fluid escape processes. Gullies maintain a rugose internal geometry throughout their development and do not represent sediment transport pathways to the deep basin. The geological processes that form seabed pockmarks and pockmark gullies are active today and these features may represent a hazard to subsea infrastructure.

Breakup of the Newfoundland–Iberia rift

Abstract The Newfoundland–Iberia rift is considered to be a type example of a non-volcanic rift. Key features of the conjugate margins are transition zones (TZs) that lie between clearly continental crust and presumed normal (Penrose-type) oceanic crust that appears up to 150–180 km farther seaward. Basement ridges drilled in the Iberia TZ consist of exhumed, serpentinized peridotite of continental affinity, consistent with seismic refraction studies. Although the boundaries between continental crust and the TZs can be defined with relative confidence, there are major questions about the position and nature of the change from rifting to normal sea-floor spreading at the seaward edges of the TZs. Notably, drilling of presumed oceanic crust in the young M-series anomalies (&lt;M5) has recovered serpentinized peridotite, and this basement experienced major extension up to approximately 15 million years after it was emplaced. In addition, existing interpretations place the ‘breakup unconformity’ (normally associated with the separation of continental crust and simultaneous formation of oceanic crust) near the Aptian–Albian boundary, which is also some 15 million years younger than the oldest proposed oceanic crust (anomaly M5–M3) in the rift. To investigate and potentially resolve these conflicts, we analysed the tectonic history and deep (pre-Cenomanian) stratigraphy of the rift using seismic reflection profiles and drilling results. Rifting occurred in two main phases (Late Triassic–earliest Jurassic and Late Jurassic–Early Cretaceous). The first phase formed continental rift basins without significant thinning of continental crust. The second phase led to continental breakup, with extension concentrated in three episodes that culminated near the end of Berriasian, Hauterivian and Aptian time. The first two episodes appear to correlate with separation of continental crust in the southern and northern parts of the rift, respectively, suggesting that the rift opened from south to north in a two-step process. The third episode persisted through Barremian and Aptian time. We suggest that during this period there was continued exhumation of subcontinental mantle lithosphere at the plate boundary, and that elevated in-plane tensile stress throughout the rift caused intraplate extension, primarily within the exhumed mantle. This rifting may have been interrupted for a time during the Barremian when melt was introduced from the southern edge of the rift by plume magmatism that formed the Southeast Newfoundland Ridge and J Anomaly Ridge, and the conjugate Madeira–Tore Rise. We propose that the rising asthenosphere breached the subcontinental mantle lithosphere in latest Aptian–earliest Albian time, initiating sea-floor spreading. This resulted in relaxation of in-plane tensile stress (i.e. a pulse of relative compression) that caused internal plate deformation and enhanced mass wasting. This ‘Aptian event’ produced a strong, rift-wide reflection that is unconformably onlapped by post-rift sediments that were deposited as a stable sea-floor-spreading regime was established. Although previously considered to be a breakup unconformity associated with separation of continental crust, the event instead marks the final separation of the subcontinental mantle lithosphere. Our analysis indicates that interpretation of tectonic events in a non-volcanic rift must consider the rheology of the full thickness of the continental lithosphere, in addition to spatial and temporal changes in extension that may occur from segment to segment along the rift.

Heat transfer in volcano–ice interactions on Mars: synthesis of environments and implications for processes and landforms

AbstractWe review new advances in volcano–ice interactions on Mars and focus additional attention on (1) recent analyses of the mechanisms of penetration of the cryosphere by dikes and sills; (2) documentation of the glacial origin of huge fan-shaped deposits on the northwest margins of the Tharis Montes and evidence for abundant volcano–ice interactions during the later Amazonian period of volcanic edifice construction and (3) the circumpolar Hesperian-aged Dorsa Argentea Formation, interpreted as an ice sheet and displaying marginal features (channels, lakes and eskers) indicative of significant melting and interior features interpreted to be due to volcano–ice interactions (e.g. subglacial volcanic edifices, pits, basins, channels and eskers). In this context, we describe and analyse several stages and types of volcano–ice interactions: (1) magmatic interactions with ice-rich parts of the cryosphere; (2) subglacial volcanism represented by intrusion under and into the ice and formation of dikes and moberg-like ridges, intrusion of sills at the glacier–volcano substrate interface and their evolution into subglacial lava flows, formation of subglacial edifices, marginal melting and channels; (3) synglacial (ice contact) volcanism represented by flows banking up against glacier margins, chilling and forming remnant ridges and (4) post-glacial volcanism and interactions with ice deposits.