Change search
Link to record
Permanent link

Direct link
BETA
Publications (10 of 12) Show all publications
Alberti, M., Pondenkandath, V., Würsch, M., Bouillon, M., Seuret, M., Ingold, R. & Liwicki, M. (2019). Are You Tampering with My Data?. In: Laura Leal-Taixé & Stefan Roth (Ed.), Computer Vision – ECCV 2018 Workshops: Proceedings, Part II. Paper presented at 15th European Conference on Computer Vision (ECCV), September 8-14, Munich, Germany (pp. 296-312). Springer
Open this publication in new window or tab >>Are You Tampering with My Data?
Show others...
2019 (English)In: Computer Vision – ECCV 2018 Workshops: Proceedings, Part II / [ed] Laura Leal-Taixé & Stefan Roth, Springer, 2019, p. 296-312Conference paper, Published paper (Refereed)
Abstract [en]

We propose a novel approach towards adversarial attacks on neural networks (NN), focusing on tampering the data used for training instead of generating attacks on trained models. Our network-agnostic method creates a backdoor during training which can be exploited at test time to force a neural network to exhibit abnormal behaviour. We demonstrate on two widely used datasets (CIFAR-10 and SVHN) that a universal modification of just one pixel per image for all the images of a class in the training set is enough to corrupt the training procedure of several state-of-the-art deep neural networks, causing the networks to misclassify any images to which the modification is applied. Our aim is to bring to the attention of the machine learning community, the possibility that even learning-based methods that are personally trained on public datasets can be subject to attacks by a skillful adversary.

Place, publisher, year, edition, pages
Springer, 2019
Series
Lecture Notes in Computer Science ; 11130
Keywords
Adversarial attack, Machine learning, Deep neural networks, Data poisoning
National Category
Computer Sciences
Research subject
Machine Learning
Identifiers
urn:nbn:se:ltu:diva-73147 (URN)10.1007/978-3-030-11012-3_25 (DOI)2-s2.0-85061797135 (Scopus ID)978-3-030-11011-6 (ISBN)
Conference
15th European Conference on Computer Vision (ECCV), September 8-14, Munich, Germany
Available from: 2019-03-11 Created: 2019-03-11 Last updated: 2019-03-11Bibliographically approved
Kovács, G., Balogh, V., Mehta, P., Shridhar, K., Alonso, P. & Liwicki, M. (2019). Author Profiling Using Semantic and Syntactic Features: Notebook for PAN at CLEF 2019. In: Linda Cappellato, Nicola Ferro, David E. Losada, Henning Müller (Ed.), CLEF 2019 Working Notes: Working Notes of CLEF 2019 - Conference and Labs of the Evaluation Forum. Paper presented at CLEF 2019.
Open this publication in new window or tab >>Author Profiling Using Semantic and Syntactic Features: Notebook for PAN at CLEF 2019
Show others...
2019 (English)In: CLEF 2019 Working Notes: Working Notes of CLEF 2019 - Conference and Labs of the Evaluation Forum / [ed] Linda Cappellato, Nicola Ferro, David E. Losada, Henning Müller, 2019Conference paper, Published paper (Refereed)
Abstract [en]

In this paper we present an approach for the PAN 2019 Author Profiling challenge. The task here is to detect Twitter bots and also to classify the gender of human Twitter users as male or female, based on a hundred select tweets from their profile. Focusing on feature engineering, we explore the semantic categories present in tweets. We combine these semantic features with part of speech tags and other stylistic features – e.g. character floodings and the use of capital letters – for our eventual feature set. We have experimented with different machine learning techniques, including ensemble techniques, and found AdaBoost to be the most successful (attaining an F1-score of 0.99 on the development set). Using this technique, we achieved an accuracy score of 89.17% for English language tweets in the bot detection subtask

National Category
Language Technology (Computational Linguistics)
Research subject
Machine Learning
Identifiers
urn:nbn:se:ltu:diva-76936 (URN)
Conference
CLEF 2019
Available from: 2019-11-28 Created: 2019-11-28 Last updated: 2019-11-28
Maergner, P., Pondenkandath, V., Alberti, M., Liwicki, M., Riesen, K., Ingold, R. & Fischer, A. (2019). Combining graph edit distance and triplet networks for offline signature verification. Pattern Recognition Letters, 125, 527-533
Open this publication in new window or tab >>Combining graph edit distance and triplet networks for offline signature verification
Show others...
2019 (English)In: Pattern Recognition Letters, ISSN 0167-8655, E-ISSN 1872-7344, Vol. 125, p. 527-533Article in journal (Refereed) Published
Abstract [en]

Offline signature verification is a challenging pattern recognition task where a writer model is inferred using only a small number of genuine signatures. A combination of complementary writer models can make it more difficult for an attacker to deceive the verification system. In this work, we propose to combine a recent structural approach based on graph edit distance with a statistical approach based on deep triplet networks. The combination of the structural and statistical models achieve significant improvements in performance on four publicly available benchmark datasets, highlighting their complementary perspectives.

Place, publisher, year, edition, pages
Elsevier, 2019
Keywords
Offline signature verification, Graph edit distance, Metric learning, Deep convolutional neural network, Triplet network
National Category
Computer Sciences
Research subject
Machine Learning
Identifiers
urn:nbn:se:ltu:diva-75255 (URN)10.1016/j.patrec.2019.06.024 (DOI)000482374500072 ()2-s2.0-85067868377 (Scopus ID)
Note

Validerad;2019;Nivå 2;2019-08-20 (johcin)

Available from: 2019-07-08 Created: 2019-07-08 Last updated: 2019-09-13Bibliographically approved
Adewumi, O., Liwicki, F. & Liwicki, M. (2019). Conversational Systems in Machine Learning from the Point of View of the Philosophy of Science—Using Alime Chat and Related Studies. Philosophies, 4(3), Article ID 41.
Open this publication in new window or tab >>Conversational Systems in Machine Learning from the Point of View of the Philosophy of Science—Using Alime Chat and Related Studies
2019 (English)In: Philosophies, ISSN 2409-9287, Vol. 4, no 3, article id 41Article in journal (Refereed) Published
Abstract [en]

This essay discusses current research efforts in conversational systems from the philosophy of science point of view and evaluates some conversational systems research activities from the standpoint of naturalism philosophical theory. Conversational systems or chatbots have advanced over the decades and now have become mainstream applications. They are software that users can communicate with, using natural language. Particular attention is given to the Alime Chat conversational system, already in industrial use, and the related research. The competitive nature of systems in production is a result of different researchers and developers trying to produce new conversational systems that can outperform previous or state-of-the-art systems. Different factors affect the quality of the conversational systems produced, and how one system is assessed as being better than another is a function of objectivity and of the relevant experimental results. This essay examines the research practices from, among others, Longino’s view on objectivity and Popper’s stand on falsification. Furthermore, the need for qualitative and large datasets is emphasized. This is in addition to the importance of the peer-review process in scientific publishing, as a means of developing, validating, or rejecting theories, claims, or methodologies in the research community. In conclusion, open data and open scientific discussion fora should become more prominent over the mere publication-focused trend.

Place, publisher, year, edition, pages
Switzerland: MDPI, 2019
Keywords
conversational systems, chatbots, philosophy of science, objectivity, verification, falsification
National Category
Computer Sciences
Research subject
Machine Learning
Identifiers
urn:nbn:se:ltu:diva-75430 (URN)10.3390/philosophies4030041 (DOI)
Note

Validerad;2019;Nivå 1;2019-09-18 (marisr)

Available from: 2019-08-08 Created: 2019-08-08 Last updated: 2019-09-18Bibliographically approved
Kovács, G., Tóth, L., Van Compernolle, D. & Liwicki, M. (2019). Examining the combination of multi-band processing and channel dropout for robust speech recognition. In: Proceedings of the Annual Conference of the International Speech Communication Association, 2019: . Paper presented at Interspeech 2019.
Open this publication in new window or tab >>Examining the combination of multi-band processing and channel dropout for robust speech recognition
2019 (English)In: Proceedings of the Annual Conference of the International Speech Communication Association, 2019, 2019Conference paper, Published paper (Refereed)
Abstract [en]

A pivotal question in Automatic Speech Recognition (ASR) is the robustness of the trained models. In this study, we investigate the combination of two methods commonly applied to increase the robustness of ASR systems. On the one hand, inspired by auditory experiments and signal processing considerations, multi-band band processing has been used for decades to improve the noise robustness of speech recognition. On the other hand, dropout is a commonly used regularization technique to prevent overfitting by keeping the model from becoming over-reliant on a small set of neurons. We hypothesize that the careful combination of the two approaches would lead to increased robustness, by preventing the resulting model from over-rely on any given band. To verify our hypothesis, we investigate various approaches for the combination of the two methods using the Aurora-4 corpus. The results obtained corroborate our initial assumption, and show that the proper combination of the two techniques leads to increased robustness, and to significantly lower word error rates (WERs). Furthermore, we find that the accuracy scores attained here compare favourably to those reported recently on the clean training scenario of the Aurora-4 corpus.

Keywords
multi-band processing, band-dropout, robust speech recognition, Aurora-4
National Category
Human Computer Interaction
Identifiers
urn:nbn:se:ltu:diva-76905 (URN)10.21437/Interspeech.2019-3215 (DOI)
Conference
Interspeech 2019
Available from: 2019-11-28 Created: 2019-11-28 Last updated: 2019-11-28
Saini, R., Dobson, D., Morrey, J., Liwicki, M. & Liwicki, F. (2019). ICDAR 2019 Historical Document Reading Challenge on Large Structured Chinese Family Records. In: ICDAR 2019: ICDAR 2019 HDRC Chinese. Paper presented at ICDAR 2019.
Open this publication in new window or tab >>ICDAR 2019 Historical Document Reading Challenge on Large Structured Chinese Family Records
Show others...
2019 (English)In: ICDAR 2019: ICDAR 2019 HDRC Chinese, 2019Conference paper, Published paper (Refereed)
Abstract [en]

In this paper, we present a large historical database of Chinese family records with the aim to develop robust systems for historical document analysis. In this direction, we propose a Historical Document Reading Challenge on Large Chinese Structured Family Records (ICDAR 2019 HDRCCHINESE).The objective of the competition is to recognizeand analyze the layout, and finally detect and recognize thetextlines and characters of the large historical document image dataset containing more than 10000 pages. Cascade R-CNN, CRNN, and U-Net based architectures were trained to evaluatethe performances in these tasks. Error rate of 0.01 has been recorded for textline recognition (Task1) whereas a Jaccard Index of 99.54% has been recorded for layout analysis (Task2).The graph edit distance based total error ratio of 1.5% has been recorded for complete integrated textline detection andrecognition (Task3).

National Category
Other Electrical Engineering, Electronic Engineering, Information Engineering Computer Systems
Identifiers
urn:nbn:se:ltu:diva-77258 (URN)
Conference
ICDAR 2019
Available from: 2019-12-27 Created: 2019-12-27 Last updated: 2020-01-23
Adewumi, O. & Liwicki, M. (2019). Inner For-Loop for Speeding Up Blockchain Mining. Open Computer Science
Open this publication in new window or tab >>Inner For-Loop for Speeding Up Blockchain Mining
2019 (English)In: Open Computer Science, ISSN 2299-1093Article in journal (Refereed) Accepted
Abstract [en]

In this paper, the authors propose to increase the efficiency of blockchain mining by using a population-based approach. Blockchain relies on solving difficult mathematical problems as proof-of-work within a network before blocks are added to the chain. Brute force approach, advocated by some as the fastest algorithm for solving partial hash collisions and implemented in Bitcoin blockchain, implies exhaustive, sequential search. It involves incrementing the nonce (number) of the header by one, then taking a double SHA-256 hash at each instance and comparing it with a target value to ascertain if lower than that target. It excessively consumes both time and power. In this paper, the authors, therefore, suggest using an inner for-loop for the population-based approach. Comparison shows that it’s a slightly faster approach than brute force, with an average speed advantage of about 1.67% or 3,420 iterations per second and 73% of the time performing better. Also, we observed that the more the total particles deployed, the better the performance until a pivotal point. Furthermore, a recommendation on taming the excessive use of power by networks, like Bitcoin’s, by using penalty by consensus is suggested.

Place, publisher, year, edition, pages
Poland: De Gruyter Open, 2019
Keywords
Blockchain, Network, Inner For-Loop, SHA-256, Brute force
National Category
Computer Sciences
Research subject
Machine Learning
Identifiers
urn:nbn:se:ltu:diva-76859 (URN)
Available from: 2019-11-28 Created: 2019-11-28 Last updated: 2019-12-09
Saini, R., Kumar, P., Patidar, S., Roy, P. & Liwicki, M. (2019). Trilingual 3D Script Identification and Recognition using Leap Motion Sensor. In: : . Paper presented at 2nd Workshop on Machine Learning, ICDAR 2019.
Open this publication in new window or tab >>Trilingual 3D Script Identification and Recognition using Leap Motion Sensor
Show others...
2019 (English)Conference paper, Published paper (Refereed)
National Category
Computer Systems Other Electrical Engineering, Electronic Engineering, Information Engineering
Identifiers
urn:nbn:se:ltu:diva-77257 (URN)
Conference
2nd Workshop on Machine Learning, ICDAR 2019
Available from: 2019-12-27 Created: 2019-12-27 Last updated: 2020-01-23
Alberti, M., Pondenkandath, V., Wursch, M., Ingold, R. & Liwicki, M. (2018). DeepDIVA: A Highly-Functional Python Framework for Reproducible Experiments. In: Proceedings of International Conference on Frontiers in Handwriting Recognition, ICFHR 2018: . Paper presented at 16th International Conference on Frontiers in Handwriting Recognition, ICFHR 2018, 5- August 2018, Niagara Fall, United States (pp. 423-428). IEEE, Article ID 8583798.
Open this publication in new window or tab >>DeepDIVA: A Highly-Functional Python Framework for Reproducible Experiments
Show others...
2018 (English)In: Proceedings of International Conference on Frontiers in Handwriting Recognition, ICFHR 2018, IEEE, 2018, p. 423-428, article id 8583798Conference paper, Published paper (Refereed)
Abstract [en]

We introduce DeepDIVA: an infrastructure designed to enable quick and intuitive setup of reproducible experiments with a large range of useful analysis functionality. Reproducing scientific results can be a frustrating experience, not only in document image analysis but in machine learning in general. Using DeepDIVA a researcher can either reproduce a given experiment or share their own experiments with others. Moreover, the framework offers a large range of functions, such as boilerplate code, keeping track of experiments, hyper-parameter optimization, and visualization of data and results. To demonstrate the effectiveness of this framework, this paper presents case studies in the area of handwritten document analysis where researchers benefit from the integrated functionality. DeepDIVA is implemented in Python and uses the deep learning framework PyTorch. It is completely open source(1), and accessible as Web Service through DIVAServices(2).

Place, publisher, year, edition, pages
IEEE, 2018
Series
International Conference on Handwriting Recognition, ISSN 2167-6445
Keywords
Framework, Open-Source, Deep Learning, Neural Networks, Reproducible Research, Machine Learning, Hyper-parameters Optimization, Python
National Category
Computer Sciences
Research subject
Machine Learning
Identifiers
urn:nbn:se:ltu:diva-73160 (URN)10.1109/ICFHR-2018.2018.00080 (DOI)000454983200071 ()2-s2.0-85052223452 (Scopus ID)978-1-5386-5875-8 (ISBN)
Conference
16th International Conference on Frontiers in Handwriting Recognition, ICFHR 2018, 5- August 2018, Niagara Fall, United States
Available from: 2019-03-11 Created: 2019-03-11 Last updated: 2019-09-13Bibliographically approved
Pondenkandath, V., Alberti, M., Eichenberger, N., Ingold, R. & Liwicki, M. (2018). Identifying cross-depicted historical motifs. In: Proceedings of International Conference on Frontiers in Handwriting Recognition, ICFHR 2018: . Paper presented at 16th International Conference on Frontiers in Handwriting Recognition, ICFHR 2018, 5- August 2018, Niagara Fall, United States (pp. 333-338). IEEE, Article ID 8583783.
Open this publication in new window or tab >>Identifying cross-depicted historical motifs
Show others...
2018 (English)In: Proceedings of International Conference on Frontiers in Handwriting Recognition, ICFHR 2018, IEEE, 2018, p. 333-338, article id 8583783Conference paper, Published paper (Refereed)
Abstract [en]

Cross-depiction is the problem of identifying the same object even when it is depicted in a variety of manners.This is a common problem in handwritten historical document image analysis, for instance when the same letter or motif is depicted in several different ways. It is a simple task for humans yet conventional computer vision methods struggle to cope with it. In this paper we address this problem using state-of-the-art deep learning techniques on a dataset of historical watermarks containing images created with different methods of reproduction, such as hand tracing, rubbing, and radiography.To study the robustness of deep learning based approaches to the cross-depiction problem, we measure their performance on two different tasks: Classification and similarity rankings. For the former we achieve a classification accuracy of 96 % using deep convolutional neural networks. For the latter we have a false positive rate at 95% recall of 0.11. These results outperform state-of-the-art methods by a significant margin

Place, publisher, year, edition, pages
IEEE, 2018
Series
International Conference on Handwriting Recognition, ISSN 2167-6445
Keywords
convolutional neural network, cross-depiction, deep learning, machine learning, watermarks
National Category
Computer Sciences
Research subject
Machine Learning
Identifiers
urn:nbn:se:ltu:diva-72981 (URN)10.1109/ICFHR-2018.2018.00065 (DOI)000454983200056 ()2-s2.0-85060032898 (Scopus ID)978-1-5386-5875-8 (ISBN)
Conference
16th International Conference on Frontiers in Handwriting Recognition, ICFHR 2018, 5- August 2018, Niagara Fall, United States
Available from: 2019-02-21 Created: 2019-02-21 Last updated: 2019-03-11Bibliographically approved
Organisations
Identifiers
ORCID iD: ORCID iD iconorcid.org/0000-0003-4029-6574

Search in DiVA

Show all publications