no code implementations • EMNLP (NLP+CSS) 2020 • Katja Geertruida Schmahl, Tom Julian Viering, Stavros Makrodimitris, Arman Naseri Jahfari, David Tax, Marco Loog
These developments provide additional understanding of what can be done to make Wikipedia more gender neutral and how important time of writing can be when considering biases in word embeddings trained from Wikipedia or from other text corpora.
no code implementations • 25 Nov 2022 • Marco Loog, Tom Viering
Plotting a learner's generalization performance against the training set size results in a so-called learning curve.
no code implementations • 30 Oct 2022 • Yuko Kato, David M. J. Tax, Marco Loog
Estimating uncertainty of machine learning models is essential to assess the quality of the predictions that these models provide.
no code implementations • NeurIPS 2021 • Rolf A. N. Starre, Marco Loog, Elena Congeduti, Frans A. Oliehoek
This result makes it possible to extend the guarantees of existing MBRL algorithms to the setting with abstraction.
Model-based Reinforcement Learning reinforcement-learning +1
1 code implementation • 13 Jun 2022 • Gijs van Tulder, Marco Loog
This paper explains why sample reusability is of practical interest, why reusability can be a problem, how reusability could be improved by importance-weighted active learning, and which obstacles to universal reusability remain.
no code implementations • 1 Jun 2022 • Chirag Raman, Hayley Hung, Marco Loog
In this work, we introduce and address the following question: given a probabilistic forecasting model how can we identify observed windows that the model considers salient when making its forecasts?
no code implementations • CVPR 2022 • Ziqi Wang, Marco Loog
We illustrate the detrimental effect, such as overconfident decisions, that exponential behavior can have in methods like classical LDA and logistic regression.
1 code implementation • NeurIPS 2021 • Chirag Raman, Hayley Hung, Marco Loog
In this way, an SP model learns to adapt its forecasts to the unique dynamics of the interacting partners, generalizing to unseen groups in a data-efficient manner.
1 code implementation • 7 Jun 2021 • Silvia L. Pintea, Nergis Tomen, Stanley F. Goes, Marco Loog, Jan C. van Gemert
We use scale-space theory to obtain a self-similar parametrization of filters and make use of the N-Jet: a truncated Taylor series to approximate a filter by a learned combination of Gaussian derivative filters.
1 code implementation • 19 Mar 2021 • Tom Viering, Marco Loog
This important tool can be used for model selection, to predict the effect of more training data, and to reduce the computational complexity of model training and hyperparameter tuning.
no code implementations • 3 Feb 2021 • Marco Loog
Importance weighting is widely applicable in machine learning in general and in techniques dealing with data covariate shift problems in particular.
1 code implementation • 1 Dec 2020 • Burak Yildiz, Hayley Hung, Jesse H. Krijthe, Cynthia C. S. Liem, Marco Loog, Gosia Migut, Frans Oliehoek, Annibale Panichella, Przemyslaw Pawelczak, Stjepan Picek, Mathijs de Weerdt, Jan van Gemert
We present ReproducedPapers. org: an open online repository for teaching and structuring machine learning reproducibility.
no code implementations • 15 Oct 2020 • Ziqi Wang, Marco Loog, Jan van Gemert
In this work, we define DIRs employed by existing works in probabilistic terms and show that by learning DIRs, overly strict requirements are imposed concerning the invariance.
1 code implementation • 13 Aug 2020 • Kanav Anand, Ziqi Wang, Marco Loog, Jan van Gemert
Our study investigates the subjective human factor in comparisons of state of the art results and scientific reproducibility in deep learning.
no code implementations • 7 Apr 2020 • Marco Loog, Tom Viering, Alexander Mey, Jesse H. Krijthe, David M. J. Tax
In their thought-provoking paper [1], Belkin et al. illustrate and discuss the shape of risk curves in the context of modern high-complexity learners.
no code implementations • 25 Nov 2019 • Tom J. Viering, Alexander Mey, Marco Loog
Learning performance can show non-monotonic behavior.
no code implementations • 30 Aug 2019 • Alexander Mey, Marco Loog
Our main contribution is to present a way to derive finite sample L1-convergence rates of this estimator for different surrogate loss functions.
no code implementations • 26 Aug 2019 • Alexander Mey, Marco Loog
In this review we gather results about the possible gains one can achieve when using semi-supervised learning as well as results about the limits of such methods.
no code implementations • 25 Jul 2019 • Tom Viering, Ziqi Wang, Marco Loog, Elmar Eisemann
This illustrates that GradCAM cannot explain the decision of every CNN and provides a proof of concept showing that it is possible to obfuscate the inner workings of a CNN.
1 code implementation • NeurIPS 2019 • Marco Loog, Tom Viering, Alexander Mey
Plotting a learner's average performance against the number of training samples results in a learning curve.
no code implementations • 14 Jun 2019 • Alexander Mey, Tom Viering, Marco Loog
Here, we derive sample complexity bounds based on pseudo-dimension for models that add a convex data dependent regularization term to a supervised learning process, as is in particular done in Manifold regularization.
1 code implementation • 28 May 2019 • Julius von Kügelgen, Alexander Mey, Marco Loog, Bernhard Schölkopf
While the success of semi-supervised learning (SSL) is still not fully understood, Sch\"olkopf et al. (2012) have established a link to the principle of independent causal mechanisms.
1 code implementation • 16 Jan 2019 • Wouter M. Kouw, Marco Loog
Feature-based methods revolve around on mapping, projecting and representing features such that a source classifier performs well on the target domain and inference-based methods incorporate adaptation into the parameter estimation procedure, for instance through constraints on the optimization procedure.
no code implementations • 31 Dec 2018 • Wouter M. Kouw, Marco Loog
Domain adaptation and transfer learning are sub-fields within machine learning that are concerned with accounting for these types of changes.
1 code implementation • 17 Oct 2018 • Wouter M. Kouw, Marco Loog, Wilbert Bartels, Adriënne M. Mendrik
Generalization of voxelwise classifiers is hampered by differences between MRI-scanners, e. g. different acquisition protocols and field strengths.
no code implementations • 28 Aug 2018 • Lex Razoux Schultz, Marco Loog, Peyman Mohajerin Esfahani
The performance of the proposed methodology is validated through an SC case study in which our numerical experiments suggest a significant improvement in the cross domain classification error in comparison with a random selected source domain for both a naive and adaptive learning setting.
1 code implementation • 20 Jul 2018 • Julius von Kügelgen, Alexander Mey, Marco Loog
Current methods for covariate-shift adaptation use unlabelled data to compute importance weights or domain-invariant features, while the final model is trained on labelled data only.
1 code implementation • 21 Jun 2018 • Wouter M. Kouw, Marco Loog
In practice, the data distribution at test time often differs, to a smaller or larger extent, from that of the original training data.
1 code implementation • 17 May 2018 • Yazhou Yang, Marco Loog
These pseudo annotators always provide uniform and random labels whenever new unlabeled samples are queried.
1 code implementation • 19 Apr 2018 • Wouter M. Kouw, Marco Loog
For sample selection bias settings, and for small sample sizes, the importance-weighted risk estimator produces overestimates for datasets in the body of the sampling distribution, i. e. the majority of cases, and large underestimates for data sets in the tail of the sampling distribution.
no code implementations • 25 Oct 2017 • Marco Loog
The original problem of supervised classification considers the task of automatically assigning objects to their respective classes on the basis of numerical measurements derived from these objects.
1 code implementation • 17 Oct 2017 • Wouter M. Kouw, Jesse H. Krijthe, Marco Loog
Cross-validation under sample selection bias can, in principle, be done by importance-weighting the empirical risk.
1 code implementation • 22 Sep 2017 • Wouter M. Kouw, Marco Loog, Lambertus W. Bartels, Adriënne M. Mendrik
Due to this acquisition related variation, classifiers trained on data from a specific scanner fail or under-perform when applied to data that was acquired differently.
no code implementations • 19 Jul 2017 • Amogh Gudi, Nicolai van Rosmalen, Marco Loog, Jan van Gemert
To facilitate this, we propose a novel global pooling technique called Spatial Pyramid Averaged Max (SPAM) pooling for training this CAM-based network for object extent localisation with only weak image-level supervision.
no code implementations • 13 Jul 2017 • Marco Loog, Jesse H. Krijthe, Are C. Jensen
In various approaches to learning, notably in domain adaptation, active learning, learning under covariate shift, semi-supervised learning, learning with concept drift, and the like, one often wants to compare a baseline classifier to one or more advanced (or at least different) strategies.
no code implementations • 10 Jul 2017 • Marco Loog, François Lauze
We start out by demonstrating that an elementary learning task, corresponding to the training of a single linear neuron in a convolutional neural network, can be solved for feature spaces of very high dimensionality.
1 code implementation • 25 Jun 2017 • Wouter M. Kouw, Marco Loog
In domain adaptation, classifiers with information from a source domain adapt to generalize to a target domain.
1 code implementation • 23 Jun 2017 • Yazhou Yang, Marco Loog
We propose a novel approach which we refer to as maximizing variance for active learning or MVAL for short.
no code implementations • 8 Jun 2017 • Tom J. Viering, Jesse H. Krijthe, Marco Loog
In particular we show the relation between the bound of the state-of-the-art Maximum Mean Discrepancy (MMD) active learner, the bound of the Discrepancy, and a new and looser bound that we refer to as the Nuclear Discrepancy bound.
no code implementations • 15 Mar 2017 • Veronika Cheplygina, Lauge Sørensen, David M. J. Tax, Jesper Holst Pedersen, Marco Loog, Marleen de Bruijne
Chronic obstructive pulmonary disease (COPD) is a lung disease where early detection benefits the survival rate.
no code implementations • 15 Mar 2017 • Veronika Cheplygina, Lauge Sørensen, David M. J. Tax, Marleen de Bruijne, Marco Loog
We address the problem of \emph{instance label stability} in multiple instance learning (MIL) classifiers.
1 code implementation • 27 Feb 2017 • Yazhou Yang, Marco Loog
Many active learning methods belong to the retraining-based approaches, which select one unlabeled instance, add it to the training set with its possible labels, retrain the classification model, and evaluate the criteria that we base our selection on.
no code implementations • NeurIPS 2018 • Jesse H. Krijthe, Marco Loog
Consider a classification problem where we have both labeled and unlabeled data available.
no code implementations • 27 Dec 2016 • Jesse H. Krijthe, Marco Loog
In this paper, we discuss the approaches we took and trade-offs involved in making a paper on a conceptual topic in pattern recognition research fully reproducible.
no code implementations • 25 Nov 2016 • Yazhou Yang, Marco Loog
Logistic regression is by far the most widely used classifier in real-world applications.
no code implementations • 17 Oct 2016 • Jesse H. Krijthe, Marco Loog
For the supervised least squares classifier, when the number of training objects is smaller than the dimensionality of the data, adding more data to the training set may first increase the error rate before decreasing it.
no code implementations • 12 Oct 2016 • Jesse H. Krijthe, Marco Loog
The goal of semi-supervised learning is to improve supervised classifiers by using additional unlabeled training examples.
1 code implementation • 31 Jul 2016 • Wouter M. Kouw, Marco Loog
This paper identifies a problem with the usual procedure for L2-regularization parameter estimation in a domain adaptation setting.
no code implementations • 10 Mar 2016 • Erik J. Bekkers, Marco Loog, Bart M. ter Haar Romeny, Remco Duits
We propose a template matching method for the detection of 2D image objects that are characterized by orientation patterns.
no code implementations • 25 Feb 2016 • Jesse H. Krijthe, Marco Loog
For semi-supervised techniques to be applied safely in practice we at least want methods to outperform their supervised counterparts.
no code implementations • 27 Dec 2015 • Jesse H. Krijthe, Marco Loog
Experimental results show that also in the general multidimensional case performance improvements can be expected, both in terms of the squared loss that is intrinsic to the classifier, as well as in terms of the expected classification error.
no code implementations • 15 Dec 2015 • Wouter M. Kouw, Jesse H. Krijthe, Marco Loog, Laurens J. P. van der Maaten
Our empirical evaluation of FLDA focuses on problems comprising binary and count data in which the transfer can be naturally modeled via a dropout distribution, which allows the classifier to adapt to differences in the marginal probability of features in the source and the target domain.
no code implementations • 24 Jul 2015 • Jesse H. Krijthe, Marco Loog
We introduce a novel semi-supervised version of the least squares classifier.
1 code implementation • 1 Mar 2015 • Marco Loog
The latter refers to the fact that our estimates are conservative and therefore resilient to whatever form the true labeling of the unlabeled data takes on.
no code implementations • 17 Nov 2014 • Jesse H. Krijthe, Marco Loog
Using any one of these methods is not guaranteed to outperform the supervised classifier which does not take the additional unlabeled data into account.
no code implementations • 2 Jun 2014 • Veronika Cheplygina, David M. J. Tax, Marco Loog
To better deal with such problems, several extensions of supervised learning have been proposed, where either training and/or test objects are sets of feature vectors.
no code implementations • 6 Feb 2014 • David M. J. Tax, Veronika Cheplygina, Marco Loog
Considering one whole slide as a collection (a bag) of feature vectors, however, poses the problem of how to handle this bag.
no code implementations • 6 Feb 2014 • Veronika Cheplygina, David M. J. Tax, Marco Loog
In multiple instance learning, objects are sets (bags) of feature vectors (instances) rather than individual feature vectors.
no code implementations • 22 Sep 2013 • Veronika Cheplygina, David M. J. Tax, Marco Loog
Multiple instance learning (MIL) is concerned with learning from sets (bags) of objects (instances), where the individual instance labels are ambiguous.