Lidé
Ing. Vojtěch Franc, Ph.D.
Všechny publikace
Consistent and Tractable Algorithm for Markov Network Learning
 Autoři: Ing. Vojtěch Franc, Ph.D., doc. RNDr. Daniel Průša, Ph.D., Yermakov, A.
 Publikace: Machine Learning and Knowledge Discovery in Databases, Part IV. Cham: Springer, 2023. p. 435451. Lecture Notes in Computer Science. vol. 13716. ISSN 03029743. ISBN 9783031264115.
 Rok: 2023
 DOI: 10.1007/9783031264122_27
 Odkaz: https://doi.org/10.1007/9783031264122_27
 Pracoviště: Strojové učení

Anotace:
Markov network (MN) structured output classifiers provide a transparent and powerful way to model dependencies between output labels. The MN classifiers can be learned using the M3N algorithm, which, however, is not statistically consistent and requires expensive fully annotated examples. We propose an algorithm to learn MN classifiers that is based on Fisherconsistent adversarial loss minimization. Learning is transformed into a tractable convex optimization that is amenable to standard gradient methods. We also extend the algorithm to learn from examples with missing labels. We show that the extended algorithm remains convex, tractable, and statistically consistent.
Detection of Microscopic Fungi and Yeast in Clinical Samples Using Fluorescence Microscopy and Deep Learning
 Autoři: Ing. Jakub Paplhám, Ing. Vojtěch Franc, Ph.D., Lžíčařová, D.
 Publikace: Proceedings of the 18th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications. Setùbal: SciTePress, 2023. p. 777784. vol. 4. ISSN 21844321. ISBN 9789897586347.
 Rok: 2023
 DOI: 10.5220/0011616100003417
 Odkaz: https://doi.org/10.5220/0011616100003417
 Pracoviště: Skupina vizuálního rozpoznávání, Strojové učení

Anotace:
Early detection of yeast and filamentous fungi in clinical samples is critical in treating patients predisposed to severe infections caused by these organisms. The patients undergo regular screening, and the gathered samples are manually examined by trained personnel. This work uses deep neural networks to detect filamentous fungi and yeast in the clinical samples to simplify the work of the human operator by filtering out samples that are clearly negative and presenting the operator with only samples suspected of containing the contaminant. We propose data augmentation with Poisson inpainting and compare the model performance against expert and beginnerlevel humans. The method achieves humanlevel performance, theoretically reducing the amount of manual labor by 87%, given a true positive rate of 99% and incidence rate of 10%.
Optimal Strategies for Reject Option Classifiers
 Autoři: Ing. Vojtěch Franc, Ph.D., doc. RNDr. Daniel Průša, Ph.D., Voráček, V.
 Publikace: Journal of Machine Learning Research. 2023, 24(11), 149. ISSN 15324435.
 Rok: 2023
 Pracoviště: Strojové učení

Anotace:
In classification with a reject option, the classifier is allowed in uncertain cases to abstain from prediction. The classical costbased model of a reject option classifier requires the rejection cost to be defined explicitly. The alternative boundedimprovement model and the boundedabstention model avoid the notion of the reject cost. The boundedimprovement model seeks a classifier with a guaranteed selective risk and maximal cover. The boundedabstention model seeks a classifier with guaranteed cover and minimal selective risk. We prove that despite their different formulations the three rejection models lead to the same prediction strategy: the Bayes classifier endowed with a randomized Bayes selection function. We define the notion of a proper uncertainty score as a scalar summary of the prediction uncertainty sufficient to construct the randomized Bayes selection function. We propose two algorithms to learn the proper uncertainty score from examples for an arbitrary blackbox classifier. We prove that both algorithms provide Fisher consistent estimates of the proper uncertainty score and demonstrate their efficiency in different prediction problems, including classification, ordinal regression, and structured output classification.
CNN Based Predictor of Face Image Quality
 Autoři: Yermakov, A., Ing. Vojtěch Franc, Ph.D.,
 Publikace: Pattern Recognition. ICPR International Workshops and Challenges, Part VI. Cham: Springer International Publishing, 2021. p. 679693. LNCS. vol. 12666. ISSN 03029743. ISBN 9783030687793.
 Rok: 2021
 DOI: 10.1007/9783030687809_52
 Odkaz: https://doi.org/10.1007/9783030687809_52
 Pracoviště: Strojové učení

Anotace:
We propose a novel method for training Convolution Neural Network, named CNNFQ, which takes a face image and outputs a scalar summary of the image quality. The CNNFQ is trained from triplets of faces that are automatically labeled based on responses of a pretrained face matcher. The quality scores extracted by the CNNFQ are directly linked to the probability that the face matcher incorrectly ranks a randomly selected triplet of faces. We applied the proposed CNNFQ, trained on CASIA database, for selection of the best quality image from a collection of face images capturing the same identity. The quality of the single face representation was evaluated on 1:1 Verification and 1:N Identification tasks defined by the challenging IJBB protocol. We show that the recognition performance obtained when using faces selected based on the CNNFQ scores is significantly higher than what can be achieved by competing stateoftheart image quality extractors.
Dominant subject recognition by Bayesian learning
 Autoři: Ing. Vojtěch Franc, Ph.D., Yermakov, A.
 Publikace: Proc. of the 16th IEEE International Conference on Automatic Face and Gesture Recognition, 2021 (FG 2021). Los Alamitos: IEEE Computer Society Press, 2021. ISBN 9781665431767.
 Rok: 2021
 DOI: 10.1109/FG52635.2021.9666979
 Odkaz: https://doi.org/10.1109/FG52635.2021.9666979
 Pracoviště: Strojové učení

Anotace:
We tackle the problem of dominant subject recognition (DSR), which aims at identifying the faces of the subject whose faces appear most frequently in a given collection of images. We propose a simple algorithm solving the DSR problem in a principled way via Bayesian learning. The proposed algorithm has complexity quadratic in the number of detected faces, and it provides labeling of images along with an accurate estimate of the prediction confidence. The prediction confidence permits using the algorithm in semiautomatic mode when only a subset of images with uncertain labels are corrected manually. We demonstrate on a challenging IJBB database, that the algorithm significantly reduces the number of images that need to be manually annotated to get the perfect performance of face verification and face identification systems using the face database created by the method.
Hairstyle Transfer between Face Images
 Autoři: Ing. Adéla Šubrtová, Ing. Jan Čech, Ph.D., Ing. Vojtěch Franc, Ph.D.,
 Publikace: Proc. of the 16th IEEE International Conference on Automatic Face and Gesture Recognition, 2021 (FG 2021). Los Alamitos: IEEE Computer Society Press, 2021. ISBN 9781665431767.
 Rok: 2021
 DOI: 10.1109/FG52635.2021.9667038
 Odkaz: https://doi.org/10.1109/FG52635.2021.9667038
 Pracoviště: Skupina vizuálního rozpoznávání, Strojové učení

Anotace:
We propose a neural network which takes two inputs, a hair image and a face image, and produces an output image having the hair of the hair image seamlessly merged with the inner face of the face image. Our architecture consists of neural networks mapping the input images into a latent code of a pretrained StyleGAN2 which generates the output highdefinition image. We propose an algorithm for training parameters of the architecture solely from synthetic images generated by the StyleGAN2 itself without the need of any annotations or external dataset of hairstyle images. We empirically demonstrate the effectiveness of our method in applications including hairstyle transfer, hair generation for 3D morphable models, and hairstyle interpolation. Fidelity of the generated images is verified by a user study and by a novel hairstyle metric proposed in the paper.
Learning Maximum Margin Markov Networks from examples with missing labels
 Autoři: Ing. Vojtěch Franc, Ph.D., Yermakov, A.
 Publikace: Asian Machine Learning Conference. Proceedings of Machine Learning Research, 2021. ISSN 26403498.
 Rok: 2021
 Pracoviště: Strojové učení

Anotace:
Structured output classifiers based on the framework of Markov Networks provide a transparent way to model statistical dependencies between output labels. The Markov Network (MN) classifier can be efficiently learned by the maximum margin method, which however requires expensive completely annotated examples. We extend the maximum margin algorithm for learning of unrestricted MN classifiers from examples with partially missing annotation of labels. The proposed algorithm translates learning into minimization of a novel loss function which is convex, has a clear connection with the supervised marginrescaling loss, and can be efficiently optimized by firstorder methods. We demonstrate the efficacy of the proposed algorithm on a challenging structured output classification problem where it beats deep neural network models trained from a much higher number of completely annotated examples, while the proposed method used only partial annotations.
On Model Evaluation Under Nonconstant Class Imbalance
 Autoři: Brabec, J., Komárek, T., Ing. Vojtěch Franc, Ph.D., Machlica, L.
 Publikace: Computational Science  ICCS 2020. Cham: Springer, 2020. p. 7487. vol. 12140. ISSN 03029743. ISBN 9783030504229.
 Rok: 2020
 DOI: 10.1007/9783030504236_6
 Odkaz: https://doi.org/10.1007/9783030504236_6
 Pracoviště: Strojové učení

Anotace:
Many realworld classification problems are significantly classimbalanced to detriment of the class of interest. The standard set of proper evaluation metrics is wellknown but the usual assumption is that the test dataset imbalance equals the realworld imbalance. In practice, this assumption is often broken for various reasons. The reported results are then often too optimistic and may lead to wrong conclusions about industrial impact and suitability of proposed techniques. We introduce methods (Supplementary code related to techniques described in this paper is available at: https://github.com/CiscoCTA/nci_eval) focusing on evaluation under nonconstant class imbalance. We show that not only the absolute values of commonly used metrics, but even the order of classifiers in relation to the evaluation metric used is affected by the change of the imbalance rate. Finally, we demonstrate that using subsampling in order to get a test dataset with class imbalance equal to the one observed in the wild is not necessary, and eventually can lead to significant errors in classifier’s performance estimate.
On Discriminative Learning of Prediction Uncertainty
 Autoři: Ing. Vojtěch Franc, Ph.D., doc. RNDr. Daniel Průša, Ph.D.,
 Publikace: International Conference on Machine Learning. Proceedings of Machine Learning Research, 2019. p. 34653480. ISSN 26403498. ISBN 9781510886988.
 Rok: 2019
 Pracoviště: Strojové učení

Anotace:
In classification with a reject option, the classifier is allowed in uncertain cases to abstain from prediction. The classical cost based model of an optimal classifier with a reject option requires the cost of rejection to be defined explicitly. An alternative boundedimprovement model, avoiding the notion of the reject cost, seeks for a classifier with a guaranteed selective risk and maximal cover. We prove that both models share the same class of optimal strategies, and we provide an explicit relation between the reject cost and the target risk being the parameters of the two models. An optimal rejection strategy for both models is based on thresholding the conditional risk defined by posterior probabilities which are usually unavailable. We propose a discriminative algorithm learning an uncertainty function which preserves ordering of the input space induced by the conditional risk, and hence can be used to construct optimal rejection strategies.
Learning CNNs from Weakly Annotated Facial Images
 Autoři: Ing. Vojtěch Franc, Ph.D., Ing. Jan Čech, Ph.D.,
 Publikace: Image and Vision Computing. 2018, 77 1020. ISSN 02628856.
 Rok: 2018
 DOI: 10.1016/j.imavis.2018.06.011
 Odkaz: https://doi.org/10.1016/j.imavis.2018.06.011
 Pracoviště: Skupina vizuálního rozpoznávání, Strojové učení

Anotace:
Learning of convolutional neural networks (CNNs) to perform a face recognition task requires a large set of facial images each annotated with a label to be predicted. In this paper we propose a method for learning CNNs from weakly annotated images. The weak annotation in our setting means that a pair of an attribute label and a person identity label is assigned to a set of faces automatically detected in the image. The challenge is to link the annotation with the correct face. The weakly annotated images of this type can be collected by an automated process not requiring a human labor. We formulate learning from weakly annotated images as a maximum likelihood (ML) estimation of a parametric distribution describing the weakly annotated images. The ML problem is solved by an instance of the EM algorithm which in its inner loop learns a CNN to predict attribute label from facial images. Experiments on age and gender estimation problem show that the proposed algorithm significantly outperforms the existing heuristic approach for dealing with this type of data. A practical outcome of our paper is a new annotation of the IMDB database [26] containing 300 k faces each one annotated by biological age, gender and identity labels.
Learning data discretization via convex optimization
 Autoři: Ing. Vojtěch Franc, Ph.D., Fikar, O., Bartoš, K., Sofka, M.
 Publikace: Machine Learning. 2018, 107(2), 333355. ISSN 08856125.
 Rok: 2018
 DOI: 10.1007/s1099401756544
 Odkaz: https://doi.org/10.1007/s1099401756544
 Pracoviště: Strojové učení

Anotace:
Discretization of continuous input functions into piecewise constant or piecewise linear approximations is needed in many mathematical modeling problems. It has been shown that choosing the length of the piecewise segments adaptively based on data samples leads to improved accuracy of the subsequent processing such as classification. Traditional approaches are often tied to a particular classification model which results in local greedy optimization of a criterion function. This paper proposes a technique for learning the discretization parameters along with the parameters of a decision function in a convex optimization of the true objective. The general formulation is applicable to a wide range of learning problems. Empirical evaluation demonstrates that the proposed convex algorithms yield models with fewer number of parameters with comparable or better accuracy than the existing methods.
License Plate Recognition and Superresolution from LowResolution Videos by Convolutional Neural Networks
 Autoři: Vašek, V., Ing. Vojtěch Franc, Ph.D., Urban, M.
 Publikace: BMVC2018: Proceedings of the British Machine Vision Conference. London: British Machine Vision Association, 2018.
 Rok: 2018
 Pracoviště: Strojové učení

Anotace:
The paper proposes Convolutional Neural Network (CNN) for License Plate Recognition (LPR) from lowresolution videos. The CNN accepts arbitrary long sequence of geometrically registered license plate (LP) images and outputs a distribution over a set of strings with an admissible length. Evaluation on 31k lowresolution videos shows that the proposed CNN significantly outperforms both baseline methods and humans by a large margin. Our second contribution is a CNN based superresolution generator of LP images. The generator converts input lowresolution LP image into its highresolution counterpart which i) preserves the structure of the input and ii) depicts a string that was previously recognized from video.
Visual Heart Rate Estimation with Convolutional Neural Network
 Autoři: Ing. Bc. Radim Špetlík, Ing. Vojtěch Franc, Ph.D., Ing. Jan Čech, Ph.D., prof. Ing. Jiří Matas, Ph.D.,
 Publikace: BMVC2018: Proceedings of the British Machine Vision Conference. London: British Machine Vision Association, 2018.
 Rok: 2018
 Pracoviště: Skupina vizuálního rozpoznávání, Strojové učení

Anotace:
We propose a novel twostep convolutional neural network to estimate a heart rate from a sequence of facial images. The network is trained endtoend by alternating op timization and validated on three publicly available datasets yielding stateoftheart results against three baseline methods. The network performs better by a 40% margin to the stateoftheart method on a newly collected dataset. A challenging dataset of 204 fitnessthemed videos is introduced. The dataset is designed to test the robustness of heart rate estimation methods to illumination changes and subject’s motion. 17 subjects perform 4 activities (talking, rowing, exercising on a stationary bike and an elliptical trainer) in 3 lighting setups. Each activity is captured by two RGB webcameras, one is placed on a tripod, the other is attached to the fitness machine which vibrates significantly. Subject’s age ranges from 20 to 53 years, the mean heart rate is ≈ 110, the standard deviation ≈ 25.
Largescale robust transductive support vector machines
 Autoři: Cevikalp, H., Ing. Vojtěch Franc, Ph.D.,
 Publikace: Neurocomputing. 2017, 235 199209. ISSN 09252312.
 Rok: 2017
 DOI: 10.1016/j.neucom.2017.01.012
 Odkaz: https://doi.org/10.1016/j.neucom.2017.01.012
 Pracoviště: Strojové učení

Anotace:
In this paper, we propose a robust and fast transductive support vector machine (RTSVM) classifier that can be applied to largescale data. To this end, we use the robust Ramp loss instead of Hinge loss for labeled data samples. The resulting optimization problem is nonconvex but it can be decomposed to a convex and concave parts. Therefore, the optimization is accomplished iteratively by solving a sequence of convex problems known as concaveconvex procedure. Stochastic gradient (SG) is used to solve the convex problem at each iteration, thus the proposed method scales well with large training set size for the linear case (to the best of our knowledge, it is the second transductive classification method that is practical for more than a million data). To extend the proposed method to the nonlinear case, we proposed two alternatives where one uses the primal optimization problem and the other uses the dual. But in contrast to the linear case, both alternatives do not scale well with largescale data. Experimental results show that the proposed method achieves comparable results to other related transductive SVM methods, but it is faster than other transductive learning methods and it is more robust to the noisy data.
Learning CNNs for face recognition from weakly annotated images
 Autoři: Ing. Vojtěch Franc, Ph.D., Ing. Jan Čech, Ph.D.,
 Publikace: International Conference on Automatic Face and Gesture Recognition Workshops, Biometrics in the Wild. USA: IEEE Computer Society, 2017. p. 933940. ISSN 23265396. ISBN 9781509040230.
 Rok: 2017
 DOI: 10.1109/FG.2017.115
 Odkaz: https://doi.org/10.1109/FG.2017.115
 Pracoviště: Skupina vizuálního rozpoznávání, Strojové učení

Anotace:
Supervised learning of convolutional neural networks (CNNs) for face recognition requires a large set of facial images each annotated with a single attribute label to be predicted. In this paper we propose a method for learning CNNs from weakly annotated images. The weak annotation in our setting means that a pair of an attribute label and a person identity label is assigned to a set of faces automatically detected in the image. The challenge is to link the annotation with the correct face. The weakly annotated images of this type can be collected by an automated process not requiring a human labor. We formulate learning from weakly annotated images as a maximum likelihood estimation of a parametric distribution describing the data. The ML problem is solved by an instance of EM algorithm which in its inner loop learns a CNN to perform given face recognition task. Experiments on age and gender estimation problem show that the proposed EMCNN algorithm significantly outperforms the stateoftheart approach for dealing with this type of data.
Visual Language Identification from Facial Landmarks
 Autoři: Ing. Bc. Radim Špetlík, Ing. Jan Čech, Ph.D., Ing. Vojtěch Franc, Ph.D., prof. Ing. Jiří Matas, Ph.D.,
 Publikace: Image Analysis, Part II. Springer, Cham, 2017. p. 389400. Lecture Notes in Computer Science. vol. 10270. ISSN 03029743. ISBN 9783319591285.
 Rok: 2017
 DOI: 10.1007/9783319591292_33
 Odkaz: https://doi.org/10.1007/9783319591292_33
 Pracoviště: Katedra kybernetiky, Skupina vizuálního rozpoznávání, Strojové učení

Anotace:
The automatic Visual Language IDentification (VLID), i.e. a problem of using visual information to identify the language being spoken, using no audio information, is studied. The proposed method employs facial landmarks automatically detected in a video. A convex optimisation problem to find jointly both the discriminative representation (a softhistogram over a set of lip shapes) and the classifier is formulated. A 10fold crossvalidation is performed on dataset consisting of 644 videos collected from youtube.com resulting in accuracy 73% in a pairwise iscrimination between English and French (50% for a chance).Astudy, inwhich 10 videos were used, suggests that the proposed method performs better than average human in discriminating between the languages.
Learning Invariant Representation for Malicious Network Traffic Detection
 Autoři: Bartos, K., Sofka, M., Ing. Vojtěch Franc, Ph.D.,
 Publikace: European Conference on Artificial Intelligence. Amsterdam: IOS Press, 2016. pp. 11321139. Frontiers in Artificial Intelligence and Applications. vol. 285. ISSN 09226389. ISBN 9781614996712.
 Rok: 2016
 DOI: 10.3233/97816149967291132
 Odkaz: https://doi.org/10.3233/97816149967291132
 Pracoviště: Strojové učení

Anotace:
Statistical learning theory relies on an assumption that the joint distributions of observations and labels are the same in training and testing data. However, this assumption is violated in many real world problems, such as training a detector of malicious network traffic that can change over time as a result of attacker's detection evasion efforts. We propose to address this problem by creating an optimized representation, which significantly increases the robustness of detectors or classifiers trained under this distributional shift. The representation is created from bags of samples (e.g. network traffic logs) and is designed to be invariant under shifting and scaling of the feature values extracted from the logs and under permutation and size changes of the bags. The invariance is achieved by combining feature histograms with feature selfsimilarity matrices computed for each bag and significantly reduces the difference between the training and testing data. The parameters of the representation, such as histogram bin boundaries, are learned jointly with the classifier. We show that the representation is effective for training a detector of malicious traffic, achieving 90% precision and 67% recall on samples of previously unseen malware variants.
Multiview facial landmark detection by using a 3D shape model
 Autoři: Ing. Jan Čech, Ph.D., Ing. Vojtěch Franc, Ph.D., Uřičář, M., prof. Ing. Jiří Matas, Ph.D.,
 Publikace: Image and Vision Computing. 2016, 47 6070. ISSN 02628856.
 Rok: 2016
 DOI: 10.1016/j.imavis.2015.11.003
 Odkaz: https://doi.org/10.1016/j.imavis.2015.11.003
 Pracoviště: Katedra kybernetiky

Anotace:
An algorithm for accurate localization of facial landmarks coupled with a head pose estimation from a single monocular image is proposed. The algorithm is formulated as an optimization problem where the sum of individual landmark scoring functions is maximized with respect to the camera pose by fitting a parametric 3D shape model. The landmark scoring functions are trained by a structured output SVM classifier that takes a distance to the true landmark position into account when learning. The optimization criterion is nonconvex and we propose a robust initialization scheme which employs a global method to detect a raw but reliable initial landmark position. Selfocclusions causing landmarks invisibility are handled explicitly by excluding the corresponding contributions from the data term. This allows the algorithm to operate correctly for a large range of viewing angles. Experiments on standard ``inthewild'' datasets demonstrate that the proposed algorithm outperforms several stateoftheart landmark detectors especially for nonfrontal face images. The algorithm achieves the average relative landmark localization error below 10% of the interocular distance in 98.3% of the 300W dataset test images.
Multiview facial landmark detector learned by the Structured Output SVM
 Autoři: Uřičář, M., Ing. Vojtěch Franc, Ph.D., Thomas, D., Sugimoto, A., Hlaváč, V.
 Publikace: Image and Vision Computing. 2016, 47 4559. ISSN 02628856.
 Rok: 2016
 DOI: 10.1016/j.imavis.2016.02.004
 Odkaz: https://doi.org/10.1016/j.imavis.2016.02.004
 Pracoviště: Katedra kybernetiky

Anotace:
Abstract We propose a realtime multiview landmark detector based on Deformable Part Models (DPM). The detector is composed of a mixture of tree based DPMs, each component describing landmark configurations in a specific range of viewing angles. The usage of view specific DPMs allows to capture a large range of poses and to deal with the problem of selfocclusions. Parameters of the detector are learned from annotated examples by the Structured Output Support Vector Machines algorithm. The learning objective is directly related to the performance measure used for detector evaluation. The tree based DPM allows to find a globally optimal landmark configuration by the dynamic programming. We propose a coarsetofine search strategy which allows realtime processing by the dynamic programming also on high resolution images. Empirical evaluation on “in the wild” images shows that the proposed detector is competitive with the stateoftheart methods in terms of speed and accuracy yet it keeps the guarantee of finding a globally optimal estimate in contrast to other methods.
Optimized Invariant Representation of Network Traffic for Detecting Unseen Malware Variants
 Autoři: Bartoš, K., Sofka, M., Ing. Vojtěch Franc, Ph.D.,
 Publikace: Proceedings of the 25th USENIX Security Symposium. The USENIX Association, 2016. p. 807822. ISBN 9781931971324.
 Rok: 2016
 Pracoviště: Strojové učení

Anotace:
New and unseen polymorphic malware, zeroday attacks, or other types of advanced persistent threats are usually not detected by signaturebased security devices, firewalls, or antiviruses. This represents a challenge to the network security industry as the amount and variability of incidents has been increasing. Consequently, this complicates the design of learningbased detection systems relying on features extracted from network data. The problem is caused by different joint distribution of observation (features) and labels in the training and testing data sets. This paper proposes a classification system designed to detect both known as well as previouslyunseen security threats. The classifiers use statistical feature representation computed from the network traffic and learn to recognize malicious behavior. The representation is designed and optimized to be invariant to the most common changes of malware behaviors. This is achieved in part by a feature histogram constructed for each group of HTTP flows (proxy log records) of a user visiting a particular hostname and in part by a feature selfsimilarity matrix computed for each group. The parameters of the representation (histogram bins) are optimized and learned based on the training samples along with the classifiers. The proposed classification system was deployed on large corporate networks, where it detected 2,090 new and unseen variants of malware samples with 90% precision (9 of 10 alerts were malicious), which is a considerable improvement when compared to the current flowbased approaches or existing signaturebased web security devices.
Vshaped interval insensitive loss for ordinal classification
 Autoři: Antoniuk, K., Ing. Vojtěch Franc, Ph.D., Hlaváč, V.
 Publikace: Machine Learning. 2016, 103(2), 261283. ISSN 08856125.
 Rok: 2016
 DOI: 10.1007/s1099401555419
 Odkaz: https://doi.org/10.1007/s1099401555419
 Pracoviště: Katedra kybernetiky

Anotace:
We address a problem of learning ordinal classifiers from partially annotated examples. We introduce a Vshaped intervalinsensitive loss function to measure discrepancy between predictions of an ordinal classifier and a partial annotation provided in the form of intervals of candidate labels. We show that under reasonable assumptions on the annotation process the Bayes risk of the ordinal classifier can be bounded by the expectation of an associated intervalinsensitive loss. We propose several convex surrogates of the intervalinsensitive loss which are used to formulate convex learning problems. We described a variant of the cutting plane method which can solve large instances of the learning problems. Experiments on a reallife application of human age estimation show that the ordinal classifier learned from cheap partially annotated examples can achieve accuracy matching the results of the sofar used supervised methods which require expensive precisely annotated examples.
Consistency of structured output learning with missing labels
 Autoři: Antoniuk, K., Ing. Vojtěch Franc, Ph.D., Hlaváč, V.
 Publikace: Proceedings of the 7th Asian Conference on Machine Learning. Brookline: Microtome Publishing, 2015. pp. 8195. Proc. of Asian Conference on Machine Learning (ACML). ISSN 15324435.
 Rok: 2015
 Pracoviště: Katedra kybernetiky

Anotace:
In this paper we study statistical consistency of partial losses suitable for learning structured output predictors from examples containing missing labels. We provide sufficient conditions on data generating distribution which admit to prove that the expected risk of the structured predictor learned by minimizing the partial loss converges to the optimal Bayes risk defined by an associated complete loss. We define a concept of surrogate classification calibrated partial losses which are easier to optimize yet their minimization preserves the statistical consistency. We give some concrete examples of surrogate partial losses which are classification calibrated. In particular, we show that the ramploss which is in the core of many existing algorithms is classification calibrated.
Facial Landmark Tracking by TreeBased Deformable Part Model Based Detector
 Autoři: Uřičář, M., Ing. Vojtěch Franc, Ph.D., Hlaváč, V.
 Publikace: The IEEE International Conference on Computer Vision (ICCV) Workshops. New York: IEEE Computer Society Press, 2015. p. 963970. ISSN 15505499. ISBN 9781467383905.
 Rok: 2015
 DOI: 10.1109/ICCVW.2015.127
 Odkaz: https://doi.org/10.1109/ICCVW.2015.127
 Pracoviště: Katedra kybernetiky

Anotace:
In this paper we describe a tracker of facial landmarks submitted to the 300 Videos in the Wild (300VW) challenge. Our tracker is a straightforward extension of a well tuned treebased DPM landmark detector originally developed for static images. The tracker is obtained by applying the static detector independently in each frame and using the Kalman filter to smooth estimates of the face positions as well as to compensate possible failures of the face detector. The resulting tracker provides a robust estimate of 68 landmarks running at 5 fps on an ordinary PC. We provide an opensource implementation of the proposed tracker at (http://cmp.felk.cvut.cz/~uricamic/clandmark/).
Learning Detector of Malicious Network Traffic from Weak Labels
 Autoři: Ing. Vojtěch Franc, Ph.D., Sofka, M., Bartoš, K.
 Publikace: Machine Learning and Knowledge Discovery in Databases, Part III. Heidelberg: Springer, 2015. p. 8599. Lecture notes in artificial intelligence. ISSN 03029743. ISBN 9783319234601.
 Rok: 2015
 DOI: 10.1007/9783319234618_6
 Odkaz: https://doi.org/10.1007/9783319234618_6
 Pracoviště: Katedra kybernetiky

Anotace:
We address the problem of learning a detector of malicious behavior in network traffic. The malicious behavior is detected based on the analysis of network proxy logs that capture malware communication between client and server computers. The conceptual problem in using the standard supervised learning methods is the lack of sufficiently representative training set containing examples of malicious and legitimate communication. Annotation of individual proxy logs is an expensive process involving security experts and does not scale with constantly evolving malware. However, weak supervision can be achieved on the level of properly defined bags of proxy logs by leveraging internet domain black lists, security reports, and sandboxing analysis. We demonstrate that an accurate detector can be obtained from the collected security intelligence data by using a Multiple Instance Learning algorithm tailored to the NeymanPearson problem. We provide a thorough experimental evaluation on a large corpus of network communications collected from various company network environments.
Realtime Multiview Facial Landmark Detector Learned by the Structured Output SVM
 Autoři: Uřičář, M., Ing. Vojtěch Franc, Ph.D., Thomas, D., Akihiro, S., Hlaváč, V.
 Publikace: BWILD'15: 11th IEEE International Conference on Automatic Face and Gestu re Recognition Workshops, Biometrics in the Wild. New York: IEEE Computer Society Press, 2015. ISBN 9781479960255.
 Rok: 2015
 DOI: 10.1109/FG.2015.7284808
 Odkaz: https://doi.org/10.1109/FG.2015.7284808
 Pracoviště: Katedra kybernetiky

Anotace:
While the problem of facial landmark detection is getting big attention i n the computer vision community recently, most of the methods deal only with nearfrontal views and there is only a few really multiview detectors available, that are capable of detection in a wide range of yaw angle (e.g. $phi in (90, 90)$). We describe a multiview facial landmark detector based on the Deformable Part Models, which treats the problem of the simultaneous landmark detection and the viewing angle estimation within a structured output classification framework. We prese nt an easily extensible and flexible framework which provides a realtime performance on the ``in the wild'' images, evaluated on a challenging ``Annotated Facial Landmarks in the Wild'' database. We show that our det ector achieves better results than the current state of the art in terms of the localization error.
A 3D Approach to Facial Landmarks: Detection, Refinement, and Tracking
 Autoři: Ing. Jan Čech, Ph.D., Ing. Vojtěch Franc, Ph.D., prof. Ing. Jiří Matas, Ph.D.,
 Publikace: Proc. of 22nd International Conference on Pattern Recognition. Los Alamitos: IEEE Computer Society Press, 2014. p. 21732178. ISSN 10514651. ISBN 9781479952083.
 Rok: 2014
 DOI: 10.1109/ICPR.2014.378
 Odkaz: https://doi.org/10.1109/ICPR.2014.378
 Pracoviště: Katedra kybernetiky

Anotace:
A realtime algorithm for accurate localization of facial landmarks in a single monocular image is proposed. The algorithm is formulated as an optimization problem, in which the sum of responses of local classifiers is maximized with respect to the camera pose by fit ting a generic (not a person specific) 3D model. The algorithm simultaneously estimates a head position and orientation and detects the facial landmarks in the image. Despite being local, we show that the basin of attraction is large to the extent it can be initialized by a scannin g window face detector. Other experiments on standard datasets demonstrate that the proposed a lgorithm outperforms a stateoftheart landmark detector especially for nonfrontal face imag es, and that it is capable of reliable and stable tracking for large set of viewing angles.
FASOLE: Fast Algorithm for Structured Output LEarning
 Autoři: Ing. Vojtěch Franc, Ph.D.,
 Publikace: Machine Learning and Knowledge Discovery in Databases  ECML PKDD 2013, part I. Heidelberg: Springer, 2014, pp. 402417. Lecture Notes in Computer Science. ISSN 03029743. ISBN 9783662448472. Available from: ftp://cmp.felk.cvut.cz/pub/cmp/articles/franc/FrancFasoleECML2014.pdf
 Rok: 2014
 DOI: 10.1007/9783662448489_26
 Odkaz: https://doi.org/10.1007/9783662448489_26
 Pracoviště: Katedra kybernetiky

Anotace:
This paper proposes a novel Fast Algorithm for Structured Ouput LEarning (FASOLE). FASOLE implements the dual coordinate ascent (DCA) algorithm for solving the dual problem of the Structured Output Support Vector Machines (SOSVM). Unlike existing instances of DCA algorithm applied for SOSVM, the proposed FASOLE uses a different working set selection strategy which provides nearly maximal improvement of the objective function in each update. FASOLE processes examples in online fashion and it provides certificate of optimality. FASOLE is guaranteed to find the {$veps$}optimal solution in {$SO(frac{1}{veps^2})$} time in the worst case. In the empirical comparison FASOLE consistently outperforms the existing stateoftheart solvers, like the Cutting Plane Algorithm or the BlockCoordinate FrankWolfe algorithm, achieving up to an order of magnitude speedups while obtaining the same precise solution.
Interval Insensitive Loss for Ordinal Classification
 Autoři: Antoniuk, K., Ing. Vojtěch Franc, Ph.D., Hlaváč, V.
 Publikace: JMLR Workshop and Conference Proceedings. Brookline: Microtome Publishing, 2014. pp. 189204. Proceedings of the Sixth Asian Conference on Machine Learning. ISSN 15324435.
 Rok: 2014
 Pracoviště: Katedra kybernetiky

Anotace:
We address a problem of learning ordinal classifier from partially annotated examples. We introduce an intervalinsensitive loss function to measure discrepancy between predictions of an ordinal classifier and a partial annotation provided in the form of intervals of admissible labels. The proposed intervalinsensitive loss is an instance of loss functions previously used for learning of different classification models from partially annotated examples. We propose several convex surrogates of the intervalinsensitive loss which can be efficiently optimized by existing solvers. Experiments on standard benchmarks and a reallife application show that ordinal classifiers learned from partially annotated examples can achieve accuracy close to the accuracy of classifiers learned from completely annotated examples.
Bundle Methods for Structured Output Learning  Back to the Roots
 Autoři: Uřičář, M., Ing. Vojtěch Franc, Ph.D., Hlaváč, V.
 Publikace: SCIA 2013: Proceedings of the 18th Scandinavian Conference on Image Analysis. Heidelberg: Springer, 2013, pp. 162171. Lecture Notes in Computer Science. ISSN 03029743. ISBN 9783642388859. Available from: http://link.springer.com/chapter/10.1007/9783642388866_16
 Rok: 2013
 DOI: 10.1007/9783642388866_16
 Odkaz: https://doi.org/10.1007/9783642388866_16
 Pracoviště: Katedra kybernetiky

Anotace:
Discriminative methods for learning structured output classifiers have been gaining popularity in recent years due to their successful applications in fields like computer vision, natural language processing, etc. Learning of the structured output classifiers leads to solving a convex minimization problem, still hard to solve by standard algorithms in reallife settings. A significant effort has been put to development of specialized solvers among which the Bundle Method for Risk Minimization (BMRM) [1] is one of the most successful. The BMRM is a simplified variant of bundle methods well known in the filed of nonsmooth optimization. In this paper, we propose two speedup improvements of the BMRM: i) using the adaptive proxterm known from the original bundle methods, ii) starting optimization from a nontrivial initial solution. We combine both improvements with the multiple cutting plane model approximation [2]. Experiments on reallife data show consistently faster convergence achieving speedup up to factor of 9.7.
Face and Landmark Detection by Using Cascade of Classifiers
 Autoři: Cevikalp, H., Triggs, B., Ing. Vojtěch Franc, Ph.D.,
 Publikace: Automatic Face and Gesture Recognition (FG), 2013 10th IEEE International Conference and Workshops on. Piscataway: IEEE, 2013. p. 17. ISBN 9781467355452.
 Rok: 2013
 DOI: 10.1109/FG.2013.6553705
 Odkaz: https://doi.org/10.1109/FG.2013.6553705
 Pracoviště: Katedra kybernetiky

Anotace:
In this paper, we consider face detection along with facial landmark localization inspired by the recent studies showing that incorporating object parts improves the detection accuracy. To this end, we train roots and parts detectors where the roots detector returns candidate image regions that cover the entire face, and the parts detector searches for the landmark locations within the candidate region. We use a cascade of binary and oneclass type classifiers for the roots detection and SVM like learning algorithm for the parts detection. Our proposed face detector outperforms the most of the successful face detection algorithms in the literature and gives the second best result on all tested challenging face detection databases. Experimental results show that including parts improves the detection performance when face images are large and the details of eyes and mouth are clearly visible, but does not introduce any improvement when the images are small.
Facial Landmarks Detector Learned by the Structured Output SVM
 Autoři: Uřičář, M., Ing. Vojtěch Franc, Ph.D., Hlaváč, V.
 Publikace: Computer Vision, Imaging and Computer Graphics. Theory and Application. Heidelberg: Springer, 2013. p. 383398. Communications in Computer and Information Science. vol. 359. ISSN 18650929. ISBN 9783642382406.
 Rok: 2013
 DOI: 10.1007/9783642382413_26
 Odkaz: https://doi.org/10.1007/9783642382413_26
 Pracoviště: Katedra kybernetiky

Anotace:
We propose a principled approach to supervised learning of facial landmarks detector based on the Deformable Part Models (DPM). We treat the task of landmarks detection as an instance of the structured output classification. To learn the parameters of the detector we use the Structured Output Support Vector Machines algorithm. The objective function of the learning algorithm is directly related to the performance of the detector and controlled by the userdefined loss function, in contrast to the previous works. Our proposed detector is realtime on a standard computer, simple to implement and easily modifiable for detection of various set of landmarks. We evaluate the performance of our detector on a challenging ``Labeled Faces in the Wild'' (LFW) database. The empirical results show that our detector consistently outperforms two public domain implementations based on the Active Appearance Models and the DPM. We are releasing opensource code implementing our proposed detector along with the manual annotation of seven facial landmarks for nearly all images in the LFW database.
MORD: Multiclass Classifier for Ordinal Regression
 Autoři: Antoniuk, K., Ing. Vojtěch Franc, Ph.D., Hlaváč, V.
 Publikace: Machine Learning and Knowledge Discovery in Databases. Heidelberg: SpringerVerlag, GmbH, 2013, pp. 96111. ISSN 03029743. ISBN 9783642409936. Available from: ftp://cmp.felk.cvut.cz/pub/cmp/articles/antoniuk/AntoniukFrancHlavacECML2013.pdf
 Rok: 2013
 DOI: 10.1007/9783642409943_7
 Odkaz: https://doi.org/10.1007/9783642409943_7
 Pracoviště: Katedra kybernetiky

Anotace:
We show that classification rules used in ordinal regression are equivalent to a certain class of linear multiclass classifiers. This observation not only allows to design new learning algorithms for ordinal regression using existing methods for multiclass classification but it also allows to derive new models for ordinal regression. For example, one can convert learning of ordinal classifier with (almost) arbitrary loss function to a convex unconstrained risk minimization problem for which many efficient solvers exist. The established equivalence also allows to increase discriminative power of the ordinal classifier without need to use kernels by introducing a piecewise ordinal classifier. We demonstrate advantages of the proposed models on standard benchmarks as well as in solving a reallife problem. In particular, we show that the proposed piecewise ordinal classifier applied to visual age estimation outperforms other standard prediction models.
RANSACing Optical Image Sequences for GEO and nearGEO Objects
 Autoři: doc. Dr. Ing. Radim Šára, Matoušek, M., Ing. Vojtěch Franc, Ph.D.,
 Publikace: Proceedings of the Advanced Maui Optical and Space Surveillance Technologies Conference. Kihei: Maui Economic Development Board, 2013, pp. 924933. ISSN 21524629. Available from: http://www.amostech.com/TechnicalPapers/2013/POSTER/SARA.pdf
 Rok: 2013
 Pracoviště: Katedra kybernetiky

Anotace:
This paper describes statistical models and an efficient MonteCarlo algorithm for detecting tracks of slowly moving objects in optical telescope imagery sequences. The algorithm is based on accurate robust image preregistration with respect to the star background, hot/warm pixel suppression, extracting dense normalized local image features, pixelwise statistical event detection, segmentation of event maps to putative image primitives, and finding consistent track sequences composed of the image primitives. Good performance at low SNR and robustness of detection with respect to fast or slowmoving thin overhead clouds is achieved by an event detection model which requires collecting at least 10 images of a particular spatial direction. The method does not degrade due to an accumulation of acquisition artifacts if more images are available. The track sequence detection method is similar in spirit to LINE [Yanagisawa et al, T JPN SOC AERONAUT S 2012]. The detection is performed by the RANSAC robust method modified for a concurrent detection of a fixed number of tracks, followed by an acceptance test based on a maximum posterior probability classifier. The statistical model of an image primitive track is based on the consistence between the size and the inclination angle of the image primitive, its image motion velocity, and the sidereal velocity, together with a consistence in relative magnitude. The method does not presume any particular movements of the object, as long as its motion velocity is constant. It can detect tracks without any constraints on their angular direction or length. The detection does not require repeated image transformations (rotations etc.), which makes it computationally efficient. The detection time is linear in the number of input images and, unlike in the LINE proposal method, the number of RANSAC proposals is (theoretically) independent of the number of putative image primitives. The current (unoptimized) experimental implementation run
RAVEL: an annotated corpus for training robots with audiovisual abilities
 Autoři: AlamedaPineda, X., SanchezRiera, J., Wienke, J., Ing. Vojtěch Franc, Ph.D., Ing. Jan Čech, Ph.D., Kulkarni, K., Deleforge, A., Horaud, R.
 Publikace: Journal on Multimodal User Interfaces. 2013, 7(12), 7991. ISSN 17837677.
 Rok: 2013
 DOI: 10.1007/s121930120111y
 Odkaz: https://doi.org/10.1007/s121930120111y
 Pracoviště: Katedra kybernetiky

Anotace:
We introduce Ravel (Robots with Audiovisual Abilities), a publicly available data set which covers examples of Human Robot Interaction (HRI) scenarios. These scenarios are recorded using the audiovisual robot head POPEYE, equipped with two cameras and four microphones, two of which being plugged into the ears of a dummy head. All the recordings were performed in a standard room with no special equipment, thus providing a challenging indoor scenario. This data set provides a basis to test and benchmark methods and algorithms for audiovisual scene analysis with the ultimate goal of enabling robots to interact with people in the most natural way. The data acquisition setup, sensor calibration, data annotation and data content are fully detailed. Moreover, three examples of using the recorded data are provided, illustrating its appropriateness for carrying out a large variety of HRI experiments. The Ravel data are publicly available at: http://ravel.humavips.eu/.
CuttingPlane Methods in Machine Learning
 Autoři: Ing. Vojtěch Franc, Ph.D., Sonnenburg, S., doc. Ing. Tomáš Werner, Ph.D.,
 Publikace: Optimization for Machine Learning. Cambridge: The MIT Press, 2012. p. 185218. Neural Information Processing. ISBN 9780262016469.
 Rok: 2012
 Pracoviště: Katedra kybernetiky

Anotace:
Cutting plane methods are optimization techniques that incrementally construct an approximation of a feasible set or an objective function by linear inequalities, called cutting planes. Numerous variants of this basic idea are among standard tools used in convex nonsmooth optimization and integer linear programing. Recently, cutting plane methods have seen growing interest in the field of machine learning. In this chapter, we describe the basic theory behind these methods and we show three of their successful applications to solving machine learning problems: regularized risk minimization, multiple kernel learning, and MAP inference in graphical models.
Detector of Facial Landmarks Learned by the Structured Output SVM
 Autoři: Uřičář, M., Ing. Vojtěch Franc, Ph.D., Hlaváč, V.
 Publikace: VISAPP 2012: Proceedings of the 7th International Conference on Computer Vision Theory and Applications. Porto: SciTePress  Science and Technology Publications, 2012. pp. 547556. ISBN 9789898565037.
 Rok: 2012
 Pracoviště: Katedra kybernetiky

Anotace:
In this paper we describe a detector of facial landmarks based on the Deformable Part Models. We treat the task of landmark detection as an instance of the structured output classification problem. We propose to learn the parameters of the detector from data by the Structured Output Support Vector Machines algorithm. In contrast to the previous works, the objective function of the learning algorithm is directly related to the performance of the resulting detector which is controlled by a userdefined loss function. The resulting detector is realtime on a standard PC, simple to implement and it can be easily modified for detection of a different set of landmarks. We evaluate performance of the proposed landmark detector on a challenging ''''Labeled Faces in the Wild`` (LFW) database. The empirical results demonstrate that the proposed detector is consistently more accurate than two public domain implementations based on the Active Appearance Models and the Deformable Part Models. We provide an opensource implementation of the proposed detector and the manual annotation of the facial landmarks for all images in the LFW database.
Efficient Algorithm for Regularized Risk Minimization
 Autoři: Uřičář, M., Ing. Vojtěch Franc, Ph.D.,
 Publikace: CVWW 2012: Proceedings of the 17th Computer Vision Winter Workshop. Ljubljana: Slovenian Pattern Recognition Society, 2012, pp. 5764. ISBN 9789619090169.
 Rok: 2012
 Pracoviště: Katedra kybernetiky

Anotace:
Many machine learning algorithms lead to solving a convex regularized risk minimization problem. Despite its convexity the problem is often very demanding in practice due to a high number of variables or a complex objective function. The Bundle Method for Risk Minimization (BMRM) is a recently proposed method for minimizing a generic regularized risk. Unlike the approximative methods, the BMRM algorithm comes with convergence guarantees but it is often too slow in practice. We propose a modified variant of the BMRM algorithm which decomposes the objective function into several parts and approximates each part by a separate cutting plane model instead of a single cutting plane model used in the original BMRM. The finer approximation of the objective function can significantly decrease the number of iterations at the expense of higher memory requirements. A preliminary experimental comparison shows promising results.
Learning Markov Networks by Analytic Center Cutting Plane Method
 Autoři: Antoniuk, K., Ing. Vojtěch Franc, Ph.D., Hlaváč, V.
 Publikace: ICPR 2012: Proceedings of 21st International Conference on Pattern Recognition. New York: IEEE, 2012, pp. 22502253. ISSN 10514651. ISBN 9784990644109.
 Rok: 2012
 Pracoviště: Katedra kybernetiky

Anotace:
During the last decade the supermodular Pairwise Markov Networks (SMPMN) have become a routinely used model for structured prediction. Their popularity can be attributed to efficient algorithms for the MAP inference. Comparably efficient algorithms for learning their parameters from data have not been available so far. We propose an instance of the Analytic Center Cutting Plane Method (ACCPM) for discriminative learning of the SMPMN from annotated examples. We empirically evaluate the proposed ACCPM on a problem of learning the SMPMN for image segmentation. Results obtained on two public datasets show that the proposed ACCPM significantly outperforms the current stateoftheart algorithm in terms of computational time as well as the accuracy because it can learn models which were not tractable by existing methods.
Learning Maximal Margin Markov Networks via Tractable Convex Optimization
 Autoři: Ing. Vojtěch Franc, Ph.D., Laskov, P.
 Publikace: Control Systems and Computers. 2011, 232(2), 2534. ISSN 01305395.
 Rok: 2011
 Pracoviště: Katedra kybernetiky

Anotace:
We show that learning of a general Markov network can be expressed as a convex optimization problem which is solvable efficiently without calling any external maxsum solver. The key idea is to use a linear programing relaxation of the maxsum problem directly in the formulation of the learning problem. We show that the proposed learning problem can be solved efficiently by the Generalized Proximal Point Algorithm. The empirical evaluation shows that our algorithm speeds up learning of general Markov networks by a factor of up to 20 compared to the ACP algorithm.
Support Vector Machines as Probabilistic Models
 Autoři: Ing. Vojtěch Franc, Ph.D., Zien, A., Bernhard, S.
 Publikace: Proceedings of the 28th Annual International Conference on Machine Learning (ICML 2011). New York: ACM, 2011. pp. 665672. ISBN 9781450306195.
 Rok: 2011
 Pracoviště: Katedra kybernetiky

Anotace:
We show how the SVM can be viewed as a maximum likelihood estimate of a class of probabilistic models. This model class can be viewed as a reparametrization of the SVM in a similar vein to the vSVM reparametrizing the classical (C)SVM. It is not discriminative, but has a nonuniform marginal. We illustrate the benefits of this new view by rederiving and reinvestigating two established SVMrelated algorithms.
Algoritmus pro minimalizaci regularizovaného rizika
 Autoři: Ing. Vojtěch Franc, Ph.D.,
 Publikace: Analýza dat 2010/II, Statistické metody pro technologii a výzkum. Pardubice: TriloByte, 2010, pp. 6975. ISBN 9788090405332.
 Rok: 2010
 Pracoviště: Katedra kybernetiky

Anotace:
Clanek popisuje obecny algoritmus pro minimalizaci convexnich problemu, na ktere je prevedeno velke mnostvi problemu strojoveho uceni.
COFFIN: A Computational Framework for Linear SVMs
 Autoři: Sonnenburg, S., Ing. Vojtěch Franc, Ph.D.,
 Publikace: Proceedings of the 27th Annual International Conference on Machine Learning (ICML 2010). Madison: Omnipress, 2010. pp. 9991006. ISBN 9781605589077.
 Rok: 2010
 Pracoviště: Katedra kybernetiky

Anotace:
In a variety of applications, kernel machines such as Support Vector Machines (SVMs) have been used with great success often delivering stat eoftheart results. Using the kernel trick, they work on several domains and even enable heterogeneous data fusion by concatenating feature spaces or multiple kernel learning. Unfortunately, they are not suited for truly largescale applications since they suffer from the curse of supporting vectors, i.e., the speed of applying SVMs decays linearly with the number of support vectors. In this paper we develop COFFIN  a new training strategy for linear SVMs that effectively allows the use of on demand computed kernel feature spaces and virtual examples in the primal. With linear training and prediction effort this framework leverages SVM applications to truly largescale problems: As an example, we train SVMs for human splice site recognition involving 50 million examples and sophisticated string kernels. Additionally, we learn an SVM based gende
The SHOGUN Machine Learning Toolbox
 Autoři: Sonnenburg, S., Rätsch, G., Henschel, S., Widmer, C., Behr, J., Zien, A., de Bona, F., Binder, A., Gehl, C., Ing. Vojtěch Franc, Ph.D.,
 Publikace: Journal of Machine Learning Research. 2010, 11(6), 17991802. ISSN 15324435.
 Rok: 2010
 Pracoviště: Katedra kybernetiky

Anotace:
We have developed a machine learning toolbox, called SHOGUN, which is designed for unified largescale learning for a broad range of feature types and learning settings. It offers a considerable number of machine learning models such as support vector machines, hidden Markov models, multiple kernel learning, linear discriminant analysis, and more. Most of the specific algorithms are able to deal with several different data classes. We have used this toolbox in several applications from computational biology, some of them coming with no less than 50 million training examples and others with 7 billion test examples. With more than a thousand installations worldwide, SHOGUN is already widely adopted in the machine learning community and beyond. SHOGUN is implemented in C++ and interfaces to MATLABTM, R, Octave, Python, and has a standalone command line interface. The source code is freely available under the GNU General Public License, Version 3 at http://www.shoguntoolbox.org.
Optimized Cutting Plane Algorithm for LargeScale Risk Minimization
 Autoři: Ing. Vojtěch Franc, Ph.D., Sonneburg, S.
 Publikace: Journal of Machine Learning Research. 2009, 10(4), 21572192. ISSN 15324435.
 Rok: 2009
 Pracoviště: Katedra kybernetiky

Anotace:
We have developed an optimized cutting plane algorithm (OCA) for solving largescale risk minimization problems. We prove that the number of iterations OCA requires to converge to a epsilon precise solution is approximately linear in the sample size. We also derive OCAS, an OCAbased linear binary SVM solver, and OCAM, a linear multiclass SVM solver. In an extensive empirical evaluation we show that OCAS outperforms current stateoftheart SVM solvers like svmlight, svmperf and BMRM, achieving speedup factor more than 1,200 over svmlight on some data sets and speedup factor of 29 over svmperf, while obtaining the same precise support vector solution.
Discriminative Learning of MaxSum Classifiers
 Autoři: Ing. Vojtěch Franc, Ph.D., Savchynskyy, B.
 Publikace: Journal of Machine Learning Research. 2008, 9(1), 67104. ISSN 15324435.
 Rok: 2008
 Pracoviště: Strojové učení

Anotace:
In this article, we extend an existing discriminative structured output learning approach to new three classes of maxsum classifiers with an arbitrary neighbourhood structure. We derive learning algorithms for two subclasses of maxsum classifiers whose response can be computed in polynomial time: (i) the maxsum classifiers with supermodular quality functions and (ii) the maxsum classifiers whose response can be computed exactly by a linear programming relaxation. Moreover, we show that the learning problem can be approximately solved even for a general maxsum classifier.
Greedy Kernel Principal Component Analysis
 Autoři: Ing. Vojtěch Franc, Ph.D., Hlaváč, V.
 Publikace: Cognitive Vision Systems. Heidelberg: Springer, 2006. p. 87106. ISBN 354033971X.
 Rok: 2006
 Pracoviště: Katedra kybernetiky

Anotace:
This contribution discusses one aspect of statistical learning and generalization. Theory of learning is very relevant to cognitive systems including cognitive vision. A technique allowing to approximate a huge training set is proposed. The approach aims to represent data in a low dimensional space with possibly minimal representation error which is similar to the Principal Component Analysis (PCA). In contrast to the PCA, the basis vectors of the low dimensional space used for data representation are properly selected vectors from the training set and not as their linear combinations. The basis vectors can be selected by a simple algorithm which has low computational requirements and allows online processing of huge data sets. As the computations in the proposed algorithm appear in a form of dot product, kernel methods can be used to cope with nonlinear problems. The proposed method was tested to approximate training sets of the Support Vector Machines and Kernel Fisher Linear Discr
License Plate Character Segmentation Usint Hidden Markov Chains
 Autoři: Ing. Vojtěch Franc, Ph.D., Hlaváč, V.
 Publikace: DAGM 2005: Proceedings of the 27th DAGM Symposium. Berlin: SpringerVerlag, 2005. p. 385392. ISBN 3540287035.
 Rok: 2005
 Pracoviště: Katedra kybernetiky

Anotace:
We propose a method for segmentation of a line of character in a noisy lov resolution image of a ca license plate. The Hidden Markov Chains are used to model a stochastic relation between an input image and a corresponding character segmentation. The segmentation problem is sxpressed as the maximum a posterionri estimation from set of admissible segmentations.
Sequential CoordinateWise Algorithm for the Nonnegative Least Squares Problem
 Autoři: Ing. Vojtěch Franc, Ph.D., Hlaváč, V., prof. Ing. Mirko Navara, DrSc.,
 Publikace: CAIP 2005: Computer Analysis of Images and Patterns. Berlin: Springer, 2005. p. 407414. ISBN 3540289690.
 Rok: 2005
 Pracoviště: Katedra kybernetiky

Anotace:
The paper contributes to the solution of the nonnegative least squares problem. We propose a novel sequential coordinatwise algorithm which is easy to implement and it is able to cope with larte scale problems. We derive stopping conditioons which allow to control te distance of the solution found to the optimal one in terms of the optimized objective function.
Simple Solvers for Large Guadratic Programming Tasks
 Autoři: Ing. Vojtěch Franc, Ph.D., Hlaváč, V.
 Publikace: DAGM 2005: Proceedings of the 27th DAGM Symposium. Berlin: SpringerVerlag, 2005. pp. 7584. ISBN 3540287035.
 Rok: 2005
 Pracoviště: Katedra kybernetiky

Anotace:
We describe solvers for specific quadratic programming (QP) tasks. The QP task becomes challenging when large number of variables is to be optimized. We propose QP solvers which are simple to iplement and still able to cope with problems having hudred thousanda variables.
Alignment of Sewerage Inspection Videos for Their Easier Indexing
 Autoři: Hanton, K., Smutný, V., Ing. Vojtěch Franc, Ph.D., Hlaváč, V.
 Publikace: ICVS2003: Proceedings of the Third International Conference on Vision Systems. Berlin: Springer, 2003. pp. 141150. ISBN 3540009213.
 Rok: 2003
 Pracoviště: Katedra kybernetiky
An iterative algorithm learning the maximal margin classifier
 Autoři: Ing. Vojtěch Franc, Ph.D., Hlaváč, V.
 Publikace: Pattern recognition. 2003, 36(9), 19851996. ISSN 00313203.
 Rok: 2003
 Pracoviště: Katedra kybernetiky
Convergence of the Expectation Maximization Algorithm for the Conditionally Independent Model to the Global Maximum
 Autoři: Ing. Vojtěch Franc, Ph.D., Hlaváč, V., prof. Ing. Mirko Navara, DrSc.,
 Publikace: Proceedings of Workshop 2003. Praha: České vysoké učení technické v Praze, 2003, pp. 234235. ISBN 8001027082.
 Rok: 2003
 Pracoviště: Katedra kybernetiky

Anotace:
This contribution builds on M.I. Schlesinger's results. M.I. Schlesinger discovered that in a special case of a statistical model (conditional independence with two hidden states only) the EM algorithm converges to a global extreme. Under the same assumptions we shortened and simplified the proof of global convergence. We also smoothed away two minor imprecisions in the original proof and learned that they do not affect validity.
Greedy Algorithm for a Training Set Reduction in the Kernel Methods
 Autoři: Ing. Vojtěch Franc, Ph.D., Hlaváč, V.
 Publikace: CAIP 2003: Computer Analysis of Images and Patterns. Berlin: Springer, 2003. p. 426433. ISBN 3540407308.
 Rok: 2003
 Pracoviště: Katedra kybernetiky
Rubust subspace mixture models using $t$distributions
 Autoři: De Ridder, D., Ing. Vojtěch Franc, Ph.D.,
 Publikace: BMVC 2003: Proceedings of the 14th British Machine Vision Conference. London: British Machine Vision Association, 2003, pp. 319328. ISBN 1901725235.
 Rok: 2003
 Pracoviště: Katedra kybernetiky
Training Set Approximation for Kernel Methods
 Autoři: Ing. Vojtěch Franc, Ph.D., Hlaváč, V.
 Publikace: Computer Vision  CVWW'03 : Proceedings of the 8th Computer Vision Winter Workshop. Prague: Czech Pattern Recognition Society, 2003, pp. 121126. ISBN 8023899678.
 Rok: 2003
 Pracoviště: Katedra kybernetiky
Kernel representation of the Kesler construction for Multiclass SVM classification
 Autoři: Ing. Vojtěch Franc, Ph.D., Hlaváč, V.
 Publikace: Proceedings of the CVWW'02. Wien: Pattern Recognition & Image Processing Group, Vienna University of Technology, 2002, pp. 7.
 Rok: 2002
 Pracoviště: Katedra kybernetiky
Multiclass Support Vector Machine
 Autoři: Ing. Vojtěch Franc, Ph.D., Hlaváč, V.
 Publikace: ICPR 02: Proceedings 16th International Conference on Pattern Recognition. Los Alamitos: IEEE Computer Society Press, 2002. p. 236239. ISBN 076951695X.
 Rok: 2002
 Pracoviště: Katedra kybernetiky
A Contribution to the Schlesinger's Algorithm Separating Mixtures of Gaussians
 Autoři: Ing. Vojtěch Franc, Ph.D., Hlaváč, V.
 Publikace: Computer Analysis of Images and Patterns: Proceedings of the 9th International Conference. Berlin: Springer, 2001. pp. 169176. ISBN 3540425136.
 Rok: 2001
 Pracoviště: Katedra kybernetiky
A New Feature of the Statistical Pattern Recognition Toolbox
 Autoři: Ing. Vojtěch Franc, Ph.D., Hlaváč, V.
 Publikace: Telematik. 2001, 7(3), 2225. ISSN 10285067.
 Rok: 2001
 Pracoviště: Katedra kybernetiky
A New Feature of the Statistical Pattern Recognition Toolbox
 Autoři: Ing. Vojtěch Franc, Ph.D., Hlaváč, V.
 Publikace: Computer Vision, Computer Graphics and Photogrammetry  a Common Viewpoint. Wien: Österreichische Computer Gesellschaft, 2001, pp. 143150. ISBN 3854031475.
 Rok: 2001
 Pracoviště: Katedra kybernetiky
A Simple Learning Algorithm for Maximal Margin Classifier
 Autoři: Ing. Vojtěch Franc, Ph.D., Hlaváč, V.
 Publikace: Kernel and Subspace Methods for Computer Vision. Vienna: TU Vienna, 2001, pp. 111.
 Rok: 2001
 Pracoviště: Katedra kybernetiky