scispace - formally typeset
Search or ask a question
Institution

University of Ljubljana

EducationLjubljana, Slovenia
About: University of Ljubljana is a education organization based out in Ljubljana, Slovenia. It is known for research contribution in the topics: Population & Liquid crystal. The organization has 17210 authors who have published 47013 publications receiving 1082684 citations. The organization is also known as: Univerza v Ljubljani.


Papers
More filters
Journal ArticleDOI
TL;DR: The requirements are the basis of a new evaluation methodology that aims at a simple and easily interpretable tracker comparison and a fully-annotated dataset with per-frame annotations with several visual attributes, which is the largest benchmark to date.
Abstract: This paper addresses the problem of single-target tracker performance evaluation. We consider the performance measures, the dataset and the evaluation system to be the most important components of tracker evaluation and propose requirements for each of them. The requirements are the basis of a new evaluation methodology that aims at a simple and easily interpretable tracker comparison. The ranking-based methodology addresses tracker equivalence in terms of statistical significance and practical differences. A fully-annotated dataset with per-frame annotations with several visual attributes is introduced. The diversity of its visual properties is maximized in a novel way by clustering a large number of videos according to their visual attributes. This makes it the most sophistically constructed and annotated dataset to date. A multi-platform evaluation system allowing easy integration of third-party trackers is presented as well. The proposed evaluation methodology was tested on the VOT2014 challenge on the new dataset and 38 trackers, making it the largest benchmark to date. Most of the tested trackers are indeed state-of-the-art since they outperform the standard baselines, resulting in a highly-challenging benchmark. An exhaustive analysis of the dataset from the perspective of tracking difficulty is carried out. To facilitate tracker comparison a new performance visualization technique is proposed.

596 citations

Journal ArticleDOI
TL;DR: In the high-dimensional setting only k-NN classifiers based on the Euclidean distance seem to benefit substantially from the use of SMOTE, provided that variable selection is performed before using SMOTE; the benefit is larger if more neighbors are used.
Abstract: Classification using class-imbalanced data is biased in favor of the majority class. The bias is even larger for high-dimensional data, where the number of variables greatly exceeds the number of samples. The problem can be attenuated by undersampling or oversampling, which produce class-balanced data. Generally undersampling is helpful, while random oversampling is not. Synthetic Minority Oversampling TEchnique (SMOTE) is a very popular oversampling method that was proposed to improve random oversampling but its behavior on high-dimensional data has not been thoroughly investigated. In this paper we investigate the properties of SMOTE from a theoretical and empirical point of view, using simulated and real high-dimensional data. While in most cases SMOTE seems beneficial with low-dimensional data, it does not attenuate the bias towards the classification in the majority class for most classifiers when data are high-dimensional, and it is less effective than random undersampling. SMOTE is beneficial for k-NN classifiers for high-dimensional data if the number of variables is reduced performing some type of variable selection; we explain why, otherwise, the k-NN classification is biased towards the minority class. Furthermore, we show that on high-dimensional data SMOTE does not change the class-specific mean values while it decreases the data variability and it introduces correlation between samples. We explain how our findings impact the class-prediction for high-dimensional data. In practice, in the high-dimensional setting only k-NN classifiers based on the Euclidean distance seem to benefit substantially from the use of SMOTE, provided that variable selection is performed before using SMOTE; the benefit is larger if more neighbors are used. SMOTE for k-NN without variable selection should not be used, because it strongly biases the classification towards the minority class.

588 citations

Journal ArticleDOI
Renu R. Bahadoer1, Esmée A Dijkstra2, Boudewijn van Etten2, Corrie A.M. Marijnen3, Corrie A.M. Marijnen1, Hein Putter1, Elma Meershoek-Klein Kranenbarg1, Annet G H Roodvoets1, Iris D. Nagtegaal4, Regina G. H. Beets-Tan3, Lennart Blomqvist5, Tone Fokstuen5, Albert J. ten Tije, Jaume Capdevila6, Mathijs P. Hendriks, Ibrahim Edhemovic7, Andrés Cervantes8, Per Nilsson5, Bengt Glimelius9, Cornelis J.H. van de Velde1, Geke A. P. Hospers2, L. Østergaard, F. Svendsen Jensen, P. Pfeiffer, K.E.J. Jensen, M.P. Hendriks, W.H. Schreurs, H.P. Knol, J.J. van der Vliet, J.B. Tuynman, A.M.E. Bruynzeel, E.D. Kerver, S. Festen, M E van Leerdam, G.L. Beets, L.G.H. Dewit, C.J.A. Punt, Pieter J. Tanis, E.D. Geijsen, P. Nieboer, W.A. Bleeker, A.J. Ten Tije, R.M.P.H. Crolla, A.C.M. van de Luijtgaarden, J.W.T. Dekker, J.M. Immink, F.J.F. Jeurissen, A.W.K.S. Marinelli, H.M. Ceha, T.C. Stam, P. Quarles an Ufford, W.H. Steup, A.L.T. Imholz, R.J.I. Bosker, J.H.M. Bekker, G.J. Creemers, G.A.P. Nieuwenhuijzen, H. van den Berg, W.M. van der Deure, R.F. Schmitz, J.M. van Rooijen, A.F.T. Olieman, A.C.M. van den Bergh, Derk Jan A. de Groot, Klaas Havenga, Jannet C. Beukema, J. de Boer, P.H.J.M. Veldman, E.J.M. Siemerink, J.W.P. Vanstiphout, B. de Valk, Q.A.J. Eijsbouts, M.B. Polée, C. Hoff, A. Slot, H.W. Kapiteijn, K.C.M.J. Peeters, F.P. Peters, P.A. Nijenhuis, S.A. Radema, H. de Wilt, P. Braam, G.J. Veldhuis, D. Hess, T. Rozema, O. Reerink, D. Ten Bokkel Huinink, A. Pronk, Janet R. Vos, M. Tascilar, G.A. Patijn, C. Kersten, O. Mjåland, M. Grønlie Guren, A.N. Nesbakken, J. Benedik, I. Edhemovic7, V. Velenik, J. Capdevila6, E. Espin, R. Salazar, S. Biondo, V. Pachón, J. die Trill, J. Aparicio, E. Garcia Granero, M.J. Safont, J.C. Bernal, A. Cervantes8, A. Espí Macías, L. Malmberg, G. Svaninger, H. Hörberg, G. Dafnis, A. Berglund, L. Österlund, K. Kovacs, J. Hol, S. Ottosson, G. Carlsson, C. Bratthäll, J. Assarsson, B.L. Lödén, P. Hede, I. Verbiené, O. Hallböök, A. Johnsson, M.L. Lydrup, K. Villmann, P. Matthiessen, J.H. Svensson, J. Haux, S. Skullman, T. Fokstuen5, Torbjörn Holm, P. Flygare, M. Walldén, B. Lindh, O. Lundberg, C. Radu, L. Påhlman, A. Piwowar, K. Smedh, U. Palenius, S. Jangmalm, P. Parinkh, H. Kim, M.L. Silviera 
TL;DR: The Rectal cancer And Preoperative Induction therapy followed by Dedicated Operation (RAPIDO) trial aimed to reduce distant metastases without compromising locoregional control.
Abstract: Summary Background Systemic relapses remain a major problem in locally advanced rectal cancer. Using short-course radiotherapy followed by chemotherapy and delayed surgery, the Rectal cancer And Preoperative Induction therapy followed by Dedicated Operation (RAPIDO) trial aimed to reduce distant metastases without compromising locoregional control. Methods In this multicentre, open-label, randomised, controlled, phase 3 trial, participants were recruited from 54 centres in the Netherlands, Sweden, Spain, Slovenia, Denmark, Norway, and the USA. Patients were eligible if they were aged 18 years or older, with an Eastern Cooperative Oncology Group (ECOG) performance status of 0–1, had a biopsy-proven, newly diagnosed, primary, locally advanced rectal adenocarcinoma, which was classified as high risk on pelvic MRI (with at least one of the following criteria: clinical tumour [cT] stage cT4a or cT4b, extramural vascular invasion, clinical nodal [cN] stage cN2, involved mesorectal fascia, or enlarged lateral lymph nodes), were mentally and physically fit for chemotherapy, and could be assessed for staging within 5 weeks before randomisation. Eligible participants were randomly assigned (1:1), using a management system with a randomly varying block design (each block size randomly chosen to contain two to four allocations), stratified by centre, ECOG performance status, cT stage, and cN stage, to either the experimental or standard of care group. All investigators remained masked for the primary endpoint until a prespecified number of events was reached. Patients allocated to the experimental treatment group received short-course radiotherapy (5 × 5 Gy over a maximum of 8 days) followed by six cycles of CAPOX chemotherapy (capecitabine 1000 mg/m2 orally twice daily on days 1–14, oxaliplatin 130 mg/m2 intravenously on day 1, and a chemotherapy-free interval between days 15–21) or nine cycles of FOLFOX4 (oxaliplatin 85 mg/m2 intravenously on day 1, leucovorin [folinic acid] 200 mg/m2 intravenously on days 1 and 2, followed by bolus fluorouracil 400 mg/m2 intravenously and fluorouracil 600 mg/m2 intravenously for 22 h on days 1 and 2, and a chemotherapy-free interval between days 3–14) followed by total mesorectal excision. Choice of CAPOX or FOLFOX4 was per physician discretion or hospital policy. Patients allocated to the standard of care group received 28 daily fractions of 1·8 Gy up to 50·4 Gy or 25 fractions of 2·0 Gy up to 50·0 Gy (per physician discretion or hospital policy), with concomitant twice-daily oral capecitabine 825 mg/m2 followed by total mesorectal excision and, if stipulated by hospital policy, adjuvant chemotherapy with eight cycles of CAPOX or 12 cycles of FOLFOX4. The primary endpoint was 3-year disease-related treatment failure, defined as the first occurrence of locoregional failure, distant metastasis, new primary colorectal tumour, or treatment-related death, assessed in the intention-to-treat population. Safety was assessed by intention to treat. This study is registered with the EudraCT, 2010-023957-12, and ClinicalTrials.gov , NCT01558921 , and is now complete. Findings Between June 21, 2011, and June 2, 2016, 920 patients were enrolled and randomly assigned to a treatment, of whom 912 were eligible (462 in the experimental group; 450 in the standard of care group). Median follow-up was 4·6 years (IQR 3·5–5·5). At 3 years after randomisation, the cumulative probability of disease-related treatment failure was 23·7% (95% CI 19·8–27·6) in the experimental group versus 30·4% (26·1–34·6) in the standard of care group (hazard ratio 0·75, 95% CI 0·60–0·95; p=0·019). The most common grade 3 or higher adverse event during preoperative therapy in both groups was diarrhoea (81 [18%] of 460 patients in the experimental group and 41 [9%] of 441 in the standard of care group) and neurological toxicity during adjuvant chemotherapy in the standard of care group (16 [9%] of 187 patients). Serious adverse events occurred in 177 (38%) of 460 participants in the experimental group and, in the standard of care group, in 87 (34%) of 254 patients without adjuvant chemotherapy and in 64 (34%) of 187 with adjuvant chemotherapy. Treatment-related deaths occurred in four participants in the experimental group (one cardiac arrest, one pulmonary embolism, two infectious complications) and in four participants in the standard of care group (one pulmonary embolism, one neutropenic sepsis, one aspiration, one suicide due to severe depression). Interpretation The observed decreased probability of disease-related treatment failure in the experimental group is probably indicative of the increased efficacy of preoperative chemotherapy as opposed to adjuvant chemotherapy in this setting. Therefore, the experimental treatment can be considered as a new standard of care in high-risk locally advanced rectal cancer. Funding Dutch Cancer Foundation, Swedish Cancer Society, Spanish Ministry of Economy and Competitiveness, and Spanish Clinical Research Network.

586 citations

Journal ArticleDOI
Federica Spoto1, Federica Spoto2, Paolo Tanga2, Francois Mignard2  +498 moreInstitutions (86)
TL;DR: In this paper, the authors describe the processing of the Gaia DR2 data, and describe the criteria used to select the sample published in Gaia DR 2, and explore the data set to assess its quality.
Abstract: Context. The Gaia spacecraft of the European Space Agency (ESA) has been securing observations of solar system objects (SSOs) since the beginning of its operations. Data Release 2 (DR2) contains the observations of a selected sample of 14,099 SSOs. These asteroids have been already identified and have been numbered by the Minor Planet Center repository. Positions are provided for each Gaia observation at CCD level. As additional information, complementary to astrometry, the apparent brightness of SSOs in the unfiltered G band is also provided for selected observations.Aims. We explain the processing of SSO data, and describe the criteria we used to select the sample published in Gaia DR2. We then explore the data set to assess its quality.Methods. To exploit the main data product for the solar system in Gaia DR2, which is the epoch astrometry of asteroids, it is necessary to take into account the unusual properties of the uncertainty, as the position information is nearly one-dimensional. When this aspect is handled appropriately, an orbit fit can be obtained with post-fit residuals that are overall consistent with the a-priori error model that was used to define individual values of the astrometric uncertainty. The role of both random and systematic errors is described. The distribution of residuals allowed us to identify possible contaminants in the data set (such as stars). Photometry in the G band was compared to computed values from reference asteroid shapes and to the flux registered at the corresponding epochs by the red and blue photometers (RP and BP).Results. The overall astrometric performance is close to the expectations, with an optimal range of brightness G ~ 12 − 17. In this range, the typical transit-level accuracy is well below 1 mas. For fainter asteroids, the growing photon noise deteriorates the performance. Asteroids brighter than G ~ 12 are affected by a lower performance of the processing of their signals. The dramatic improvement brought by Gaia DR2 astrometry of SSOs is demonstrated by comparisons to the archive data and by preliminary tests on the detection of subtle non-gravitational effects.

584 citations

Journal ArticleDOI
TL;DR: The data obtained demonstrate the heavy metal-dependent expression of different AMF genes in the intra- and extraradical mycelium in plants and fungi under heavy metal stress.

583 citations


Authors

Showing all 17388 results

NameH-indexPapersCitations
David Miller2032573204840
Hyun-Chul Kim1764076183227
James M. Tour14385991364
Carmen García139150396925
Bernt Schiele13056870032
Vladimir Cindro129115782000
Teresa Barillari12998478782
Sven Menke129112182034
Horst Oberlack12998580069
Hubert Kroha129112680746
Peter Schacht129103080092
Siegfried Bethke1291266103520
Igor Mandić128106579498
Stefan Kluth128126184534
Andrej Gorišek12895167830
Network Information
Related Institutions (5)
Ghent University
111K papers, 3.7M citations

91% related

National Research Council
76K papers, 2.4M citations

90% related

École Polytechnique Fédérale de Lausanne
98.2K papers, 4.3M citations

90% related

Royal Institute of Technology
68.4K papers, 1.9M citations

90% related

University of Padua
114.8K papers, 3.6M citations

90% related

Performance
Metrics
No. of papers from the Institution in previous years
YearPapers
202390
2022331
20213,149
20203,110
20192,780
20182,479