Match!

Expectation maximization transfer learning and its application for bionic hand prostheses

Published on Feb 1, 2018in Neurocomputing4.072
· DOI :10.1016/j.neucom.2017.11.072
Benjamin Paaßen7
Estimated H-index: 7
(Bielefeld University),
Alexander Schulz38
Estimated H-index: 38
(Bielefeld University)
+ 1 AuthorsBarbara Hammer33
Estimated H-index: 33
(Bielefeld University)
Sources
Abstract
Abstract Machine learning models in practical settings are typically confronted with changes to the distribution of the incoming data. Such changes can severely affect the model performance, leading for example to misclassifications of data. This is particularly apparent in the domain of bionic hand prostheses, where machine learning models promise faster and more intuitive user interfaces, but are hindered by their lack of robustness to everyday disturbances, such as electrode shifts. One way to address changes in the data distribution is transfer learning, that is, to transfer the disturbed data to a space where the original model is applicable again. In this contribution, we propose a novel expectation maximization algorithm to learn linear transformations that maximize the likelihood of disturbed data according to the undisturbed model. We also show that this approach generalizes to discriminative models, in particular learning vector quantization models. In our evaluation on data from the bionic prostheses domain we demonstrate that our approach can learn a transformation which improves classification accuracy significantly and outperforms all tested baselines, if few data or few classes are available in the target domain.
Figures & Tables
  • References (23)
  • Citations (3)
📖 Papers frequently viewed together
2016
6 Citations
2018CVPR: Computer Vision and Pattern Recognition
3 Authors (Artem Rozantsev, ..., Pascal Fua)
22 Citations
1 Citations
78% of Scinapse members use related papers. After signing in, all features are FREE.
References23
Newest
#1Cosima Prahm (Medical University of Vienna)H-Index: 5
#2Benjamin Paaßen (Citec)H-Index: 7
Last. Oskar C. Aszmann (Medical University of Vienna)H-Index: 25
view all 5 authors...
For decades, researchers have attempted to provide patients with an intuitive method to control upper limb prostheses, enabling them to manipulate multiple degrees of freedom continuously and simultaneously using only simple myoelectric signals. However, such controlling schemes are still highly vulnerable to disturbances in the myoelectric signal, due to electrode shifts, posture changes, sweat, fatigue etc. Recent research has demonstrated that such robustness problems can be alleviated by rap...
12 CitationsSource
Dec 1, 2016 in ICDM (International Conference on Data Mining)
#1Viktor Losing (Bielefeld University)H-Index: 5
#2Barbara Hammer (Bielefeld University)H-Index: 33
Last. Heiko Wersing (Honda)H-Index: 19
view all 3 authors...
Data Mining in non-stationary data streams is gaining more attentionrecently, especially in the context of Internet of Things and Big Data. It is a highly challenging task, since the fundamentally different typesof possibly occurring drift undermine classical assumptions such asi.i.d. data or stationary distributions. Available algorithms are either struggling with certain forms of drift or require a priori knowledge in terms of a task specific setting. We propose the Self Adjusting Memory (SAM)...
50 CitationsSource
#1Benjamin PaaßenH-Index: 7
#2Alexander SchulzH-Index: 38
Last. Barbara HammerH-Index: 33
view all 3 authors...
The utility of machine learning models in everyday applications critically depends on their robustness with respect to systematic changes in the input data. However, many machine learning models trained under lab conditions do break down if they are confronted with such systematic changes. Transfer learning addresses this issue by modelling changes in the input as transfer functions, which can be used to map the data to a space where the learned machine learning model is applicable again. In thi...
6 Citations
#1Gregory Ditzler (UA: University of Arizona)H-Index: 10
#2Manuel Roveri (Polytechnic University of Milan)H-Index: 19
Last. Robi Polikar (Rowan University)H-Index: 30
view all 4 authors...
The prevalence of mobile phones, the internet-of-things technology, and networks of sensors has led to an enormous and ever increasing amount of data that are now more commonly available in a streaming fashion [1]-[5]. Often, it is assumed - either implicitly or explicitly - that the process generating such a stream of data is stationary, that is, the data are drawn from a fixed, albeit unknown probability distribution. In many real-world scenarios, however, such an assumption is simply not true...
184 CitationsSource
#1Patrick BlöbaumH-Index: 1
#2Alexander Schulz (Bielefeld University)H-Index: 38
Last. Barbara Hammer (Bielefeld University)H-Index: 33
view all 3 authors...
We investigate the suitability of unsupervised dimensionality reduction (DR) for transfer learning in the context of different represen- tations of the source and target domain. Essentially, unsupervised DR establishes a link of source and target domain by representing the data in a common latent space. We consider two settings: a linear DR of source and target data which establishes correspondences of the data and an ac- cording transfer, and its combination with a non-linear DR which allows to...
3 Citations
#1Rami N. Khushaba (UTS: University of Technology, Sydney)H-Index: 19
#2Maen Takruri (American University of Ras Al Khaimah)H-Index: 9
Last. Sarath Kodagoda (UTS: University of Technology, Sydney)H-Index: 18
view all 4 authors...
Abstract Recent studies in Electromyogram (EMG) pattern recognition reveal a gap between research findings and a viable clinical implementation of myoelectric control strategies. One of the important factors contributing to the limited performance of such controllers in practice is the variation in the limb position associated with normal use as it results in different EMG patterns for the same movements when carried out at different positions. However, the end goal of the myoelectric control sc...
62 CitationsSource
#1Dario Farina (GAU: University of Göttingen)H-Index: 76
#2Ning Jiang (GAU: University of Göttingen)H-Index: 29
Last. Oskar C. Aszmann (Medical University of Vienna)H-Index: 25
view all 7 authors...
Despite not recording directly from neural cells, the surface electromyogram (EMG) signal contains information on the neural drive to muscles, i.e, the spike trains of motor neurons. Using this property, myoelectric control consists of the recording of EMG signals for extracting control signals to command external devices, such as hand prostheses. In commercial control systems, the intensity of muscle activity is extracted from the EMG and used for single degrees of freedom activation (direct co...
290 CitationsSource
#1Joseph T. Belter (Yale University)H-Index: 8
#2Jacob L. SegilH-Index: 7
Last. Richard F. WeirH-Index: 19
view all 4 authors...
In this article, we set forth a detailed analysis of the mechanical characteristics of anthropomorphic prosthetic hands. We report on an empirical study concerning the perfor- mance of several commercially available myoelectric pros- thetic hands, including the Vincent, iLimb, iLimb Pulse, Bebionic, Bebionic v2, and Michelangelo hands. We investi- gated the finger design and kinematics, mechanical joint cou- pling, and actuation methods of these commercial prosthetic hands. The empirical finding...
235 CitationsSource
Pattern recognition techniques have been applied to extract information from electromyographic (EMG) signals that can be used to control electrical powered hand prostheses. In this paper, optimized spatial filters that enhance separation properties of EMG signals are investigated. In particular, different multiclass extensions of the common spatial patterns algorithm are applied to high-density surface EMG signals acquired from the forearms of ten healthy subjects. Visualization of the obtained ...
58 CitationsSource
#1David Barber (UCL: University College London)H-Index: 23
Machine learning methods extract value from vast data sets quickly and with modest resources. They are established tools in a wide range of industrial applications, including search engines, DNA sequencing, stock market analysis, and robot locomotion, and their use is spreading rapidly. People who know the methods have their choice of rewarding jobs. This hands-on text opens these opportunities to computer science students with modest mathematical backgrounds. It is designed for final-year under...
728 Citations
Cited By3
Newest
#1Guangyu Jia ('KCL': King's College London)H-Index: 1
#2Hak-Keung Lam ('KCL': King's College London)H-Index: 46
Last. Rong Wang ('KCL': King's College London)
view all 4 authors...
Abstract The electromyogram (EMG) signals from an individual’s muscles can reflect the biomechanics of human movement. The accurate classification of individual and combined finger movements using surface EMG signals is able to support many applications such as dexterous prosthetic hand control. The existing research of EMG-based hand gesture classification faces the challenges of inaccurate classification, insufficient generalization ability and weak robustness. To address these problems, this ...
Source
An important barrier to commercialization of pattern recognition myoelectric control of prostheses is the lack of robustness to confounding factors such as electrode shift, skin impedance variations, and learning effects. To overcome this challenge, a novel supervised adaptation approach based on transfer learning (TL) with convolutional neural networks (CNNs) is proposed which requires only a short training session (a few seconds for each class) to recalibrate the system. TL is proposed as a so...
Source
#1Yi Zheng (SDUST: Shandong University of Science and Technology)H-Index: 1
#2Peisi Zhong (SDUST: Shandong University of Science and Technology)H-Index: 1
Last. Qingchao YueH-Index: 1
view all 5 authors...
Abstract Aiming at the stable walking problem of lower limb exoskeleton robot, the instability mechanism of the robot during walking is analyzed based on the zero-moment point theory. It is found that the zero-moment point theory has certain limitations in a complete walking cycle. Based on this, the stability criterion of the biped projection polygon method is proposed, and the stable polygon in the walking process is established. The segmentation method is used to divide the stable polygon of ...
Source
#1Carles IgualH-Index: 1
Last. Jorge IgualH-Index: 10
view all 4 authors...
State-of-the-art high-end prostheses are electro-mechanically able to provide a great variety of movements. Nevertheless, in order to functionally replace a human limb, it is essential that each movement is properly controlled. This is the goal of prosthesis control, which has become a growing research field in the last decades, with the ultimate goal of reproducing biological limb control. Therefore, exploration and development of prosthesis control are crucial to improve many aspects of an amp...
1 CitationsSource
#1Nikolaos Passalis (A.U.Th.: Aristotle University of Thessaloniki)H-Index: 9
#2Anastasios Tefas (A.U.Th.: Aristotle University of Thessaloniki)H-Index: 30
Abstract Drones, also known as Unmanned Aerial Vehicles, are capable of capturing spectacular aerial shots and can be used to aid several cinematography-oriented tasks. However, flying drones in a professional setting requires the cooperation of several people, increasing the production cost and possibly reducing the quality of the obtained shots. In this paper, a generic way for formulating cinematography-oriented control objectives, that can be used for training any RL agent to automate the dr...
1 CitationsSource
#1Cosima Prahm (Medical University of Vienna)H-Index: 5
#2Alexander SchulzH-Index: 38
Last. Oskar C. Aszmann (Medical University of Vienna)H-Index: 25
view all 8 authors...
Research on machine learning approaches for upper-limb prosthesis control has shown impressive progress. However, translating these results from the lab to patient’s everyday lives remains a challenge because advanced control schemes tend to break down under everyday disturbances, such as electrode shifts. Recently, it has been suggested to apply adaptive transfer learning to counteract electrode shifts using as little newly recorded training data as possible. In this paper, we present a novel, ...
Source
#1Alexander Schulz (Citec)H-Index: 38
#2Jeffrey Frederic Queiber (Bielefeld University)
Last. Minoru Asada (Osaka University)H-Index: 38
view all 4 authors...
Although autonomous robots can perform particularly well at highly specific tasks, learning each task in isolation is a very costly process, not only in terms of time but also in terms of hardware wearout and energy usage. Hence, robotic systems need to be able to adapt quickly to new situations in order to be useful in everyday tasks. One way to address this issue is transfer learning, which aims at reusing knowledge obtained in one situation, in a new related one. In this contribution, we deve...
Source