hh.sePublications
Change search
Link to record
Permanent link

Direct link
BETA
Publications (10 of 144) Show all publications
Bouguelia, M.-R., Nowaczyk, S., Santosh, K. C. & Verikas, A. (2018). Agreeing to disagree: active learning with noisy labels without crowdsourcing. International Journal of Machine Learning and Cybernetics, 9(8), 1307-1319
Open this publication in new window or tab >>Agreeing to disagree: active learning with noisy labels without crowdsourcing
2018 (English)In: International Journal of Machine Learning and Cybernetics, ISSN 1868-8071, E-ISSN 1868-808X, Vol. 9, no 8, p. 1307-1319Article in journal (Refereed) Published
Abstract [en]

We propose a new active learning method for classification, which handles label noise without relying on multiple oracles (i.e., crowdsourcing). We propose a strategy that selects (for labeling) instances with a high influence on the learned model. An instance x is said to have a high influence on the model h, if training h on x (with label y = h(x)) would result in a model that greatly disagrees with h on labeling other instances. Then, we propose another strategy that selects (for labeling) instances that are highly influenced by changes in the learned model. An instance x is said to be highly influenced, if training h with a set of instances would result in a committee of models that agree on a common label for x but disagree with h(x). We compare the two strategies and we show, on different publicly available datasets, that selecting instances according to the first strategy while eliminating noisy labels according to the second strategy, greatly improves the accuracy compared to several benchmarking methods, even when a significant amount of instances are mislabeled. © Springer-Verlag Berlin Heidelberg 2017

Place, publisher, year, edition, pages
Heidelberg: Springer, 2018
Keywords
Active learning, Classification, Label noise, Mislabeling, Interactive learning, Machine learning, Data mining
National Category
Signal Processing Computer Systems Computer Sciences
Identifiers
urn:nbn:se:hh:diva-33365 (URN)10.1007/s13042-017-0645-0 (DOI)
Available from: 2017-02-27 Created: 2017-02-27 Last updated: 2018-07-23Bibliographically approved
Vaiciukynas, E., Gelzinis, A., Verikas, A. & Bacauskiene, M. (2018). Parkinson’s Disease Detection from Speech Using Convolutional Neural Networks. In: Guidi, B., Ricci, L., Calafate, C., Gaggi, O., Marquez-Barja, J. (Ed.), Smart objects and technologies for social good: Third International Conference, GOODTECHS 2017, Pisa, Italy, November 29-30, 2017, Proceedings. Paper presented at Third EAI International Conference on Smart Objects and Technologies for Social Good, GOODTECHS 2017, Pisa, Italy, November 29-30, 2017 (pp. 206-215). Cham: Springer, 233
Open this publication in new window or tab >>Parkinson’s Disease Detection from Speech Using Convolutional Neural Networks
2018 (English)In: Smart objects and technologies for social good: Third International Conference, GOODTECHS 2017, Pisa, Italy, November 29-30, 2017, Proceedings / [ed] Guidi, B., Ricci, L., Calafate, C., Gaggi, O., Marquez-Barja, J., Cham: Springer, 2018, Vol. 233, p. 206-215Conference paper, Published paper (Refereed)
Abstract [en]

Application of deep learning tends to outperform hand-crafted features in many domains. This study uses convolutional neural networks to explore effectiveness of various segments of a speech signal,? – text-dependent pronunciation of a short sentence, – in Parkinson’s disease detection task. Besides the common Mel-frequency spectrogram and its first and second derivatives, inclusion of various other input feature maps is also considered. Image interpolation is investigated as a solution to obtain a spectrogram of fixed length. The equal error rate (EER) for sentence segments varied from 20.3% to 29.5%. Fusion of decisions from sentence segments achieved EER of 14.1%, whereas the best result when using the full sentence exhibited EER of 16.8%. Therefore, splitting speech into segments could be recommended for Parkinson’s disease detection. © ICST Institute for Computer Sciences, Social Informatics and Telecommunications Engineering 2018.

Place, publisher, year, edition, pages
Cham: Springer, 2018
Series
Lecture Notes of the Institute for Computer Sciences, Social Informatics and Telecommunications Engineering, E-ISSN 1867-8211 ; 233
Keywords
Parkinson’s disease, Audio signal processing, Convolutional neural network, Information fusion
National Category
Signal Processing
Identifiers
urn:nbn:se:hh:diva-36617 (URN)10.1007/978-3-319-76111-4_21 (DOI)978-3-319-76111-4 (ISBN)
Conference
Third EAI International Conference on Smart Objects and Technologies for Social Good, GOODTECHS 2017, Pisa, Italy, November 29-30, 2017
Note

Funding: Research Council of Lithuania (No. MIP-075/2015)

Available from: 2018-04-16 Created: 2018-04-16 Last updated: 2018-04-17Bibliographically approved
Minelga, J., Verikas, A., Vaiciukynas, E., Gelzinis, A. & Bacauskiene, M. (2017). A Transparent Decision Support Tool in Screening for Laryngeal Disorders Using Voice and Query Data. Applied Sciences: APPS, 7(10), 1-15, Article ID 1096.
Open this publication in new window or tab >>A Transparent Decision Support Tool in Screening for Laryngeal Disorders Using Voice and Query Data
Show others...
2017 (English)In: Applied Sciences: APPS, ISSN 1454-5101, E-ISSN 1454-5101, Vol. 7, no 10, p. 1-15, article id 1096Article in journal (Refereed) Published
Abstract [en]

The aim of this study is a transparent tool for analysis of voice (sustained phonation /a/) and query data capable of providing support in screening for laryngeal disorders. In this work, screening is concerned with identification of potentially pathological cases by classifying subject’s data into ’healthy’ and ’pathological’ classes as well as visual exploration of data and automatic decisions. A set of association rules and a decision tree, techniques lending themselves for exploration, were generated for pathology detection. Data pairwise similarities, estimated in a novel way, were mapped onto a 2D metric space for visual inspection and analysis. Accurate identification of pathological cases was observed on unseen subjects using the most discriminative query parameter and six audio parameters routinely used by otolaryngologists in a clinical practice: equal error rate (EER) of 11.1% was achieved using association rules and 10.2% using the decision tree. The EER was further reduced to 9.5% by combining results from these two classifiers. The developed solution can be a useful tool for Otolaryngology departments in diagnostics, education and exploratory tasks. © 2017 by the authors.

Place, publisher, year, edition, pages
Bucharest: Universitatea Politehnica din Bucuresti, 2017
Keywords
decision tree, t-SNE visualization, association rules, pathological voice
National Category
Medical Engineering
Identifiers
urn:nbn:se:hh:diva-35313 (URN)10.3390/app7101096 (DOI)000414457800134 ()2-s2.0-85032291253 (Scopus ID)
Available from: 2017-11-02 Created: 2017-11-02 Last updated: 2017-11-29Bibliographically approved
Vaiciukynas, E., Verikas, A., Gelzinis, A. & Bacauskiene, M. (2017). Detecting Parkinson's disease from sustained phonation and speech signals. PLoS ONE, 12(10), Article ID e0185613.
Open this publication in new window or tab >>Detecting Parkinson's disease from sustained phonation and speech signals
2017 (English)In: PLoS ONE, ISSN 1932-6203, E-ISSN 1932-6203, Vol. 12, no 10, article id e0185613Article in journal (Refereed) Published
Abstract [en]

This study investigates signals from sustained phonation and text-dependent speech modalities for Parkinson’s disease screening. Phonation corresponds to the vowel /a/ voicing task and speech to the pronunciation of a short sentence in Lithuanian language. Signals were recorded through two channels simultaneously, namely, acoustic cardioid (AC) and smart phone (SP) microphones. Additional modalities were obtained by splitting speech recording into voiced and unvoiced parts. Information in each modality is summarized by 18 well-known audio feature sets. Random forest (RF) is used as a machine learning algorithm, both for individual feature sets and for decision-level fusion. Detection performance is measured by the out-of-bag equal error rate (EER) and the cost of log-likelihood-ratio. Essentia audio feature set was the best using the AC speech modality and YAAFE audio feature set was the best using the SP unvoiced modality, achieving EER of 20.30% and 25.57%, respectively. Fusion of all feature sets and modalities resulted in EER of 19.27% for the AC and 23.00% for the SP channel. Non-linear projection of a RF-based proximity matrix into the 2D space enriched medical decision support by visualization. © 2017 Vaiciukynas et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Place, publisher, year, edition, pages
San Francisco, CA: Public Library of Science, 2017
Keywords
Speech analysis, Pathology detection, Parkinson's disease
National Category
Signal Processing
Identifiers
urn:nbn:se:hh:diva-35229 (URN)10.1371/journal.pone.0185613 (DOI)000412360300047 ()28982171 (PubMedID)2-s2.0-85030766664 (Scopus ID)
Note

Funding: The Research Council of Lithuania (No. MIP-075/2015)

Available from: 2017-10-19 Created: 2017-10-19 Last updated: 2017-11-29Bibliographically approved
Verikas, A., Parker, J., Bacauskiene, M. & Olsson, M. C. (2017). Exploring relations between EMG and biomechanical data recorded during a golf swing. Expert systems with applications, 88, 109-117
Open this publication in new window or tab >>Exploring relations between EMG and biomechanical data recorded during a golf swing
2017 (English)In: Expert systems with applications, ISSN 0957-4174, E-ISSN 1873-6793, Vol. 88, p. 109-117Article in journal (Refereed) Published
Abstract [en]

Exploring relations between patterns of peak rotational speed of thorax, pelvis and arm, and patterns of EMG signals recorded from eight muscle regions of forearms and shoulders during the golf swing is the main objective of this article. The linear canonical correlation analysis, allowing studying relations between sets of variables, was the main technique applied. To get deeper insights, linear and nonlinear random forests-based prediction models relating a single output variable, e.g. a thorax peak rotational speed, with a set of input variables, e.g. an average intensity of EMG signals were used. The experimental investigations using data from 16 golfers revealed statistically significant relations between sets of input and output variables. A strong direct linear relation was observed between lin- ear combinations of EMG averages and peak rotational speeds. The coefficient of determination values R2 = 0 . 958 and R2 = 0 . 943 obtained on unseen data by the random forest models designed to predict peak rotational speed of thorax and pelvis , indicate high modelling accuracy. However, predictions of peak rotational speed of arm were less accurate. This was expected, since peak rotational speed of arm played a minor role in the linear combination of peak speeds. The most important muscles to predict peak rotational speed of the body parts were identified. The investigations have shown that the canon- ical correlation analysis is a promising tool for studying relations between sets of biomechanical and EMG data. Better understanding of these relations will lead to guidelines concerning muscle engagement and coordination of thorax, pelvis and arms during a golf swing and will help golf coaches in providing substantiated advices. ©2017 Elsevier Ltd. All rights reserved.

Place, publisher, year, edition, pages
Kidlington, Oxford: Pergamon Press, 2017
Keywords
Canonical correlation, Random forest, Prediction, EMG, Golf
National Category
Sport and Fitness Sciences
Identifiers
urn:nbn:se:hh:diva-34611 (URN)10.1016/j.eswa.2017.06.041 (DOI)000408789300008 ()2-s2.0-85021670724 (Scopus ID)
Funder
Knowledge Foundation, 2012/0319
Available from: 2017-07-12 Created: 2017-07-12 Last updated: 2018-03-23Bibliographically approved
Ražanskas, P., Verikas, A., Viberg, P.-A. & Olsson, C. M. (2017). Predicting physiological parameters in fatiguing bicycling exercises using muscle activation timing. Biomedical Signal Processing and Control, 35, 19-29
Open this publication in new window or tab >>Predicting physiological parameters in fatiguing bicycling exercises using muscle activation timing
2017 (English)In: Biomedical Signal Processing and Control, ISSN 1746-8094, E-ISSN 1746-8108, Vol. 35, p. 19-29Article in journal (Refereed) Published
Abstract [en]

This article is concerned with a novel technique for prediction of blood lactate concentration level and oxygen uptake rate from multi-channel surface electromyography (sEMG) signals. The approach is built on predictive models exploiting a set of novel time-domain variables computed from sEMG signals. Signals from three muscles of each leg, namely, vastus lateralis, rectus femoris, and semitendinosus were used in this study. The feature set includes parameters reflecting asymmetry between legs, phase shifts between activation of different muscles, active time percentages, and sEMG amplitude. Prediction ability of both linear and non-linear (random forests-based) models was explored. The random forests models showed very good prediction accuracy and attained the coefficient of determination R2 = 0.962 for lactate concentration level and R2 = 0.980 for oxygen uptake rate. The linear models showed lower prediction accuracy. Comparable results were obtained also when sEMG amplitude data were removed from the training sets. A feature elimination algorithm allowed to build accurate random forests (R2 > 0.9) using just six (lactate concentration level) or four (oxygen uptake rate) time-domain variables. Models created to predict blood lactate concentration rate relied on variables reflecting interaction between front and back leg muscles, while parameters computed from front muscles and interactions between two legs were the most important variables for models created to predict oxygen uptake rate.© 2017 Elsevier Ltd.

Place, publisher, year, edition, pages
Amsterdam: Elsevier, 2017
Keywords
Random forests, Surface electromyography, Muscle activation patterns, Fatigue detection, Bicycling
National Category
Medical Engineering
Identifiers
urn:nbn:se:hh:diva-33966 (URN)10.1016/j.bspc.2017.02.011 (DOI)000401209300003 ()2-s2.0-85014392704 (Scopus ID)
Available from: 2017-06-03 Created: 2017-06-03 Last updated: 2017-06-09Bibliographically approved
Menezes, M. L., Samara, A., Galway, L., Pinheiro Sant'Anna, A., Verikas, A., Alonso-Fernandez, F., . . . Bond, R. (2017). Towards emotion recognition for virtual environments: an evaluation of eeg features on benchmark dataset. Personal and Ubiquitous Computing, 21(6), 1003-1013
Open this publication in new window or tab >>Towards emotion recognition for virtual environments: an evaluation of eeg features on benchmark dataset
Show others...
2017 (English)In: Personal and Ubiquitous Computing, ISSN 1617-4909, E-ISSN 1617-4917, Vol. 21, no 6, p. 1003-1013Article in journal (Refereed) Published
Abstract [en]

One of the challenges in virtual environments is the difficulty users have in interacting with these increasingly complex systems. Ultimately, endowing machines with the ability to perceive users emotions will enable a more intuitive and reliable interaction. Consequently, using the electroencephalogram as a bio-signal sensor, the affective state of a user can be modelled and subsequently utilised in order to achieve a system that can recognise and react to the user’s emotions. This paper investigates features extracted from electroencephalogram signals for the purpose of affective state modelling based on Russell’s Circumplex Model. Investigations are presented that aim to provide the foundation for future work in modelling user affect to enhance interaction experience in virtual environments. The DEAP dataset was used within this work, along with a Support Vector Machine and Random Forest, which yielded reasonable classification accuracies for Valence and Arousal using feature vectors based on statistical measurements and band power from the and waves and High Order Crossing of the EEG signal. © 2017, The Author(s).

Place, publisher, year, edition, pages
London: Springer London, 2017
Keywords
Classification (of information), Decision trees, Electroencephalography, Feature extraction, Speech recognition, Virtual reality, Affective Computing, Affective state, Benchmark datasets, Circumplex models, Classification accuracy, Electroencephalogram signals, Emotion recognition, Interaction experiences, Behavioral research
National Category
Signal Processing
Identifiers
urn:nbn:se:hh:diva-36499 (URN)10.1007/s00779-017-1072-7 (DOI)000416170900005 ()2-s2.0-85027845103 (Scopus ID)
Note

cited By 1

Available from: 2018-06-14 Created: 2018-06-14 Last updated: 2018-06-14Bibliographically approved
Lundström, J., Järpe, E. & Verikas, A. (2016). Detecting and exploring deviating behaviour of smart home residents. Expert systems with applications, 55, 429-440
Open this publication in new window or tab >>Detecting and exploring deviating behaviour of smart home residents
2016 (English)In: Expert systems with applications, ISSN 0957-4174, E-ISSN 1873-6793, Vol. 55, p. 429-440Article in journal (Refereed) Published
Abstract [en]

A system for detecting deviating human behaviour in a smart home environment is the long-term goal of this work. Clearly, such systems will be very important in ambient assisted living services. A new approach to modelling human behaviour patterns is suggested in this paper. The approach reveals promising results in unsupervised modelling of human behaviour and detection of deviations by using such a model. Human behaviour/activity in a short time interval is represented in a novel fashion by responses of simple non-intrusive sensors. Deviating behaviour is revealed through data clustering and analysis of associations between clusters and data vectors representing adjacent time intervals (analysing transitions between clusters). To obtain clusters of human behaviour patterns, first, a random forest is trained without using beforehand defined teacher signals. Then information collected in the random forest data proximity matrix is mapped onto the 2D space and data clusters are revealed there by agglomerative clustering. Transitions between clusters are modelled by the third order Markov chain.

Three types of deviations are considered: deviation in time, deviation in space and deviation in the transition between clusters of similar behaviour patterns.

The proposed modelling approach does not make any assumptions about the position, type, and relationship of sensors but is nevertheless able to successfully create and use a model for deviation detection-this is claimed as a significant result in the area of expert and intelligent systems. Results show that spatial and temporal deviations can be revealed through analysis of a 2D map of high dimensional data. It is demonstrated that such a map is stable in terms of the number of clusters formed. We show that the data clusters can be understood/explored by finding the most important variables and by analysing the structure of the most representative tree. © 2016 Elsevier Ltd. All rights reserved.

Place, publisher, year, edition, pages
Amsterdam: Elsevier, 2016
Keywords
Ambient assisted living, Random forests, Stochastic neighbour embedding, Markov chain, Intelligent environments
National Category
Other Electrical Engineering, Electronic Engineering, Information Engineering Signal Processing
Identifiers
urn:nbn:se:hh:diva-30594 (URN)10.1016/j.eswa.2016.02.030 (DOI)000374811000033 ()2-s2.0-84960082873 (Scopus ID)
Projects
CAISR / SA3L
Funder
Knowledge Foundation, 2010/0271
Available from: 2016-03-30 Created: 2016-03-30 Last updated: 2018-03-22Bibliographically approved
Verikas, A., Vaiciukynas, E., Gelzinis, A., Parker, J. & Olsson, M. C. (2016). Electromyographic Patterns during Golf Swing: Activation Sequence Profiling and Prediction of Shot Effectiveness. Sensors, 16(4), Article ID 592.
Open this publication in new window or tab >>Electromyographic Patterns during Golf Swing: Activation Sequence Profiling and Prediction of Shot Effectiveness
Show others...
2016 (English)In: Sensors, ISSN 1424-8220, E-ISSN 1424-8220, Vol. 16, no 4, article id 592Article in journal (Refereed) Published
Abstract [en]

This study analyzes muscle activity, recorded in an eight-channel electromyographic (EMG) signal stream, during the golf swing using a 7-iron club and exploits information extracted from EMG dynamics to predict the success of the resulting shot. Muscles of the arm and shoulder on both the left and right sides, namely flexor carpi radialis, extensor digitorum communis, rhomboideus and trapezius, are considered for 15 golf players (∼5 shots each). The method using Gaussian filtering is outlined for EMG onset time estimation in each channel and activation sequence profiling. Shots of each player revealed a persistent pattern of muscle activation. Profiles were plotted and insights with respect to player effectiveness were provided. Inspection of EMG dynamics revealed a pair of highest peaks in each channel as the hallmark of golf swing, and a custom application of peak detection for automatic extraction of swing segment was introduced. Various EMG features, encompassing 22 feature sets, were constructed. Feature sets were used individually and also in decision-level fusion for the prediction of shot effectiveness. The prediction of the target attribute, such as club head speed or ball carry distance, was investigated using random forest as the learner in detection and regression tasks. Detection evaluates the personal effectiveness of a shot with respect to the player-specific average, whereas regression estimates the value of target attribute, using EMG features as predictors. Fusion after decision optimization provided the best results: the equal error rate in detection was 24.3% for the speed and 31.7% for the distance; the mean absolute percentage error in regression was 3.2% for the speed and 6.4% for the distance. Proposed EMG feature sets were found to be useful, especially when used in combination. Rankings of feature sets indicated statistics for muscle activity in both the left and right body sides, correlation-based analysis of EMG dynamics and features derived from the properties of two highest peaks as important predictors of personal shot effectiveness. Activation sequence profiles helped in analyzing muscle orchestration during golf shot, exposing a specific avalanche pattern, but data from more players are needed for stronger conclusions. Results demonstrate that information arising from an EMG signal stream is useful for predicting golf shot success, in terms of club head speed and ball carry distance, with acceptable accuracy. Surface EMG data, collected with a goal to automatically evaluate golf player’s performance, enables wearable computing in the field of ambient intelligence and has potential to enhance exercising of a long carry distance drive.

Place, publisher, year, edition, pages
Basel: MDPI AG, 2016
Keywords
EMG, muscle activity onset, peak detection, random forest, decision fusion
National Category
Sport and Fitness Sciences Bioinformatics (Computational Biology)
Identifiers
urn:nbn:se:hh:diva-31870 (URN)10.3390/s16040592 (DOI)000375153700171 ()27120604 (PubMedID)2-s2.0-84964308572 (Scopus ID)
External cooperation:
Funder
Knowledge Foundation, 2012/0319
Available from: 2016-08-27 Created: 2016-08-27 Last updated: 2018-03-22Bibliographically approved
Vaiciukynas, E., Verikas, A., Gelzinis, A., Bacauskiene, M., Vaskevicius, K., Uloza, V., . . . Ciceliene, J. (2016). Fusing Various Audio Feature Sets for Detection of Parkinson’s Disease from Sustained Voice and Speech Recordings. Paper presented at 18th International Conference, SPECOM 2016, Budapest, Hungary, August 23-27, 2016. Lecture Notes in Computer Science, 9811, 328-337
Open this publication in new window or tab >>Fusing Various Audio Feature Sets for Detection of Parkinson’s Disease from Sustained Voice and Speech Recordings
Show others...
2016 (English)In: Lecture Notes in Computer Science, ISSN 0302-9743, E-ISSN 1611-3349, Vol. 9811, p. 328-337Article in journal (Refereed) Published
Abstract [en]

The aim of this study is the analysis of voice and speech recordings for the task of Parkinson’s disease detection. Voice modality corresponds to sustained phonation /a/ and speech modality to a short sentence in Lithuanian language. Diverse information from recordings is extracted by 22 well-known audio feature sets. Random forest is used as a learner, both for individual feature sets and for decision-level fusion. Essentia descriptors were found as the best individual feature set, achieving equal error rate of 16.3 % for voice and 13.3 % for speech. Fusion of feature sets and modalities improved detection and achieved equal error rate of 10.8 %. Variable importance in fusion revealed speech modality as more important than voice. © Springer International Publishing Switzerland 2016

Place, publisher, year, edition, pages
Heidelberg: Springer Berlin/Heidelberg, 2016
Keywords
Parkinson’s disease, Audio signal processing, OpenSMILE, Essentia, MPEG-7, jAudio, YAAFE, Random forest, Information fusion
National Category
Language Technology (Computational Linguistics)
Identifiers
urn:nbn:se:hh:diva-31872 (URN)10.1007/978-3-319-43958-7_39 (DOI)000389335600039 ()2-s2.0-84984851988 (Scopus ID)
Conference
18th International Conference, SPECOM 2016, Budapest, Hungary, August 23-27, 2016
Note

Funding: Research Council of Lithuania (No. MIP-075/2015)

Available from: 2016-08-27 Created: 2016-08-27 Last updated: 2018-01-10Bibliographically approved
Organisations
Identifiers
ORCID iD: ORCID iD iconorcid.org/0000-0003-2185-8973

Search in DiVA

Show all publications