Willy Cheung

University of Tuebingen, Tübingen, Baden-Württemberg, Germany

Are you Willy Cheung?

Claim your profile

Publications (10)25.47 Total impact

  • [Show abstract] [Hide abstract]
    ABSTRACT: Objective. Brain-computer interfaces (BCIs) are confronted with two fundamental challenges: (a) the uncertainty associated with decoding noisy brain signals, and (b) the need for co-adaptation between the brain and the interface so as to cooperatively achieve a common goal in a task. We seek to mitigate these challenges. Approach. We introduce a new approach to brain-computer interfacing based on partially observable Markov decision processes (POMDPs). POMDPs provide a principled approach to handling uncertainty and achieving co-adaptation in the following manner: (1) Bayesian inference is used to compute posterior probability distributions ('beliefs') over brain and environment state, and (2) actions are selected based on entire belief distributions in order to maximize total expected reward; by employing methods from reinforcement learning, the POMDP's reward function can be updated over time to allow for co-adaptive behaviour. Main results. We illustrate our approach using a simple non-invasive BCI which optimizes the speed-accuracy trade-off for individual subjects based on the signal-to-noise characteristics of their brain signals. We additionally demonstrate that the POMDP BCI can automatically detect changes in the user's control strategy and can co-adaptively switch control strategies on-the-fly to maximize expected reward. Significance. Our results suggest that the framework of POMDPs offers a promising approach for designing BCIs that can handle uncertainty in neural signals and co-adapt with the user on an ongoing basis. The fact that the POMDP BCI maintains a probability distribution over the user's brain state allows a much more powerful form of decision making than traditional BCI approaches, which have typically been based on the output of classifiers or regression techniques. Furthermore, the co-adaptation of the system allows the BCI to make online improvements to its behaviour, adjusting itself automatically to the user's changing circumstances.
    Journal of Neural Engineering 10/2013; 10(6):066008. DOI:10.1088/1741-2560/10/6/066008 · 3.42 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Neurological diseases and trauma often cause demyelination, resulting in the disruption of axonal function and integrity. Endogenous remyelination promotes recovery, but the process is not well understood because no method exists to definitively distinguish regenerated from preexisting myelin. To date, remyelinated segments have been defined as anything abnormally short and thin, without empirical data to corroborate these morphological assumptions. To definitively identify regenerated myelin, we used a transgenic mouse with an inducible membrane-bound reporter and targeted Cre recombinase expression to a subset of glial progenitor cells after spinal cord injury, yielding remarkably clear visualization of spontaneously regenerated myelin in vivo. Early after injury, the mean length of sheaths regenerated by Schwann cells and oligodendrocytes (OLs) was significantly shorter than control, uninjured myelin, confirming past assumptions. However, OL-regenerated sheaths elongated progressively over 6 mo to approach control values. Moreover, OL-regenerated myelin thickness was not significantly different from control myelin at most time points after injury. Thus, many newly formed OL sheaths were neither thinner nor shorter than control myelin, vitiating accepted dogmas of what constitutes regenerated myelin. We conclude that remyelination, once thought to be static, is dynamic and elongates independently of axonal growth, in contrast to stretch-based mechanisms proposed in development. Further, without clear identification, past assessments have underestimated the extent and quality of regenerated myelin.
    Proceedings of the National Academy of Sciences 03/2013; 110(10):4075-80. DOI:10.1073/pnas.1210293110 · 9.81 Impact Factor
  • Source
    Ross K Maddox · Willy Cheung · Adrian K C Lee
    [Show abstract] [Hide abstract]
    ABSTRACT: Listeners are good at attending to one auditory stream in a crowded environment. However, is there an upper limit of streams present in an auditory scene at which this selective attention breaks down? Here, participants were asked to attend one stream of spoken letters amidst other letter streams. In half of the trials, an initial primer was played, cueing subjects to the sound configuration. Results indicate that performance increases with token repetitions. Priming provided a performance benefit, suggesting that stream selection, not formation, is the bottleneck associated with attention in an overcrowded scene. Results' implications for brain-computer interfaces are discussed.
    The Journal of the Acoustical Society of America 11/2012; 132(5):EL385-90. DOI:10.1121/1.4757696 · 1.56 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Brain-computer interfaces (BCIs) have traditionally been developed for paralyzed and locked-in individuals with no motor control. However, there is a much larger population of patients with some residual motor function as well as the general population of able-bodied individuals, both of whom could benefit significantly from BCIs. An important question that has yet to be systematically studied is: can subjects use BCIs simultaneously with overt motor activity? We present results from a preliminary study aimed at exploring this question. Three subjects used hand motor imagery in an electroencephalographic (EEG) BCI while simultaneously using a joystick to control a cursor. Particular attention was paid to preventing potential muscle artifacts from influencing imagery-based control. All three subjects were able to use the hybrid "imagery+joystick" mode of control over two days, demonstrating the ability to learn and significantly improve performance. These results suggest that subjects can potentially augment their normal human sensorimotor capability by exercising direct brain control over devices concurrently with overt motor control.
    Conference proceedings: ... Annual International Conference of the IEEE Engineering in Medicine and Biology Society. IEEE Engineering in Medicine and Biology Society. Conference 08/2012; 2012:6715-8. DOI:10.1109/EMBC.2012.6347535
  • Ross K Maddox · Willy Cheung · Adrian Kc Lee
    [Show abstract] [Hide abstract]
    ABSTRACT: As the number of elements that make up an auditory scene increases, it becomes harder to selectively attend just one of those elements. Previously, the limit of listeners' abilities to attend a target stream of repeating letters in an overcrowded scene was tested. In that experiment, each stream consisted of a repeated monotonized and localized spoken letter (an "item"), with a repetition period of 1 s. Among streams, item onset times were distributed across each repetition. Listeners were asked to detect when the attended target letter changed to an oddball "R" for a single repetition, ignoring such occurrences in the non-target streams. With a constant repetition period, adding streams to the stimulus meant that the number of items per second increased proportionally. The decrease in performance could thus be a result of having more streams in the scene, or because of the increased item rate. Here, a similar experiment was performed, holding the item rate constant, rather than the repetition period. The results allow us to disentangle the effects of the number of distractor streams and the item rate, yielding insight into the specific reasons for the diminished ability to selectively attend. Funded by USA-NIH-T32DC009975 (RKM) and R00DC010196 (AKCL).
    The Journal of the Acoustical Society of America 04/2012; 131(4):3513. DOI:10.1121/1.4709283 · 1.56 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Although neurogenesis occurs in discrete areas of the adult mammalian brain, neural progenitor cells (NPCs) produce fewer new neurons with age. To characterize the molecular changes that occur during aging, we performed a proteomic comparison between primary-cultured NPCs from the young adult and aged mouse forebrain. This analysis yielded changes in proteins necessary for cellular metabolism. Mitochondrial quantity and oxygen consumption rates decrease with aging, although mitochondrial DNA in aged NPCs does not have increased mutation rates. In addition, aged cells are resistant to the mitochondrial inhibitor rotenone and proliferate in response to lowered oxygen conditions. These results demonstrate that aging NPCs display an altered metabolic phenotype, characterized by a coordinated shift in protein expression, subcellular structure, and metabolic physiology.
    Journal of Biological Chemistry 11/2011; 286(44):38592-38601. · 4.57 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Although neurogenesis occurs in discrete areas of the adult mammalian brain, neural progenitor cells (NPCs) produce fewer new neurons with age. To characterize the molecular changes that occur during aging, we performed a proteomic comparison between primary-cultured NPCs from the young adult and aged mouse forebrain. This analysis yielded changes in proteins necessary for cellular metabolism. Mitochondrial quantity and oxygen consumption rates decrease with aging, although mitochondrial DNA in aged NPCs does not have increased mutation rates. In addition, aged cells are resistant to the mitochondrial inhibitor rotenone and proliferate in response to lowered oxygen conditions. These results demonstrate that aging NPCs display an altered metabolic phenotype, characterized by a coordinated shift in protein expression, subcellular structure, and metabolic physiology.
    Journal of Biological Chemistry 09/2011; 286(44):38592-601. DOI:10.1074/jbc.M111.252171 · 4.57 Impact Factor
  • M. Chung · W. Cheung · R. Scherer · R.P.N. Rao
    [Show abstract] [Hide abstract]
    ABSTRACT: There has been growing interest in brain-computer interfaces (BCIs) for controlling robotic devices and prosthetics directly using brain signals. Non-invasive BCIs, such as those based on electroencephalographic (EEG) signals, suffer from low signal-to-noise ratio, limiting the bandwidth of control. Invasive BCIs, on the other hand, allow fine-grained control but can leave users exhausted over long periods of time because of the amount of attention required for control on a moment-by-moment basis. In this paper, we address these problems using a new adaptive and hierarchical approach to brain-computer interfacing. The approach allows a user to teach the BCI system new skills on-the-fly; these learned skills are later invoked directly as high-level commands, relieving the user of tedious lower-level control. We demonstrate the approach using a hierarchical EEG-based BCI for controlling a humanoid robot. In a study involving four human subjects controlling the robot in a simulated home environment, each subject successfully used the BCI to teach the robot a new navigational task. They later were able to execute the same task by selecting the newly learned command from the BCI's adaptive menu, avoiding the need for low-level control. A comparison of the performance of the system under low-level and hierarchical control revealed that hierarchical control is both faster and more accurate. Our results suggest that hierarchical BCIs can provide a flexible and robust way of controlling complex robotic devices, satisfying the dual goals of decreasing the cognitive load on the user while maintaining the ability to adapt to the user's needs.
    Neural Engineering (NER), 2011 5th International IEEE/EMBS Conference on; 06/2011
  • [Show abstract] [Hide abstract]
    ABSTRACT: The performance of non-invasive electroencephalogram-based (EEG) brain-computer interfacing (BCI) has improved significantly in recent years. However, remaining challenges include the non-stationarity and the low signal-to-noise ratio (SNR) of the EEG, which limit the bandwidth and hence the available applications. In this paper, we review ongoing research in our labs and introduce novel concepts and applications. First, we present an enhancement of the 3-class self-paced Graz-BCI that allows interaction with the massive multiplayer online role playing game World of Warcraft. Second, we report on the long-term stability and robustness of detection of oscillatory components modulated by distinct mental tasks. Third, we describe a scalable, adaptive learning framework, which allows users to teach the BCI new skills on-the-fly. Using this hierarchical BCI, we successfully train and control a humanoid robot in a virtual home environment.
    Advances in Computational Intelligence - 11th International Work-Conference on Artificial Neural Networks, IWANN 2011, Torremolinos-Málaga, Spain, June 8-10, 2011, Proceedings, Part I; 01/2011
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Brain-computer interfaces (BCIs) allow a user to directly control devices such as cursors and robots using brain signals. Non-invasive BCIs, e.g., those based on electroencephalographic (EEG) signals recorded from the scalp, suffer from low signal-to-noise ratio which limits the bandwidth of control. Invasive BCIs allow fine-grained control but can leave users exhausted since control is typically exerted on a moment-by-moment basis. In this paper, we address these problems by proposing a new adaptive hierarchical architecture for brain-computer interfacing. The approach allows a user to teach the BCI new skills on-the-fly; these learned skills are later invoked directly as high-level commands, relieving the user of tedious low-level control. We report results from four subjects who used a hierarchical EEG-based BCI to successfully train and control a humanoid robot in a virtual home environment. Gaussian processes were used for learning high-level commands, allowing a BCI to switch between autonomous and user-guided modes based on the current estimate of uncertainty. We also report the first instance of multi-tasking in a BCI, involving simultaneous control of two different devices by a single user. Our results suggest that hierarchical BCIs can provide a flexible and robust way of controlling complex robotic devices in real-world environments.
    IJCAI 2011, Proceedings of the 22nd International Joint Conference on Artificial Intelligence, Barcelona, Catalonia, Spain, July 16-22, 2011; 01/2011