Summerschool Programme

All practical sessions have a duration of four hours. Sessions on different topics are run in parallel. While each participant has the opportunity to visit one practical session each afternoon, topics have only a limited space for participants. Participants will be assigned to one specific practical session for each day by the organizers based on the preferences specified by the participants. Courses that appear multiple times in the schedule are repetitions of the same session, in order to provide the possibility for participation to more people (i.e., it is not the continuation of the course).

While the schedule below only provides only the titles of the practical sessions, their abstracts can be found here.

Thursday, September 20, 2012

13:30 - 14:30   Welcome Coffee and Snacks
14:30 - 16:00   Benjamin Blankertz
Gentle Introduction to Signal Processing and Classification for Single-Trial EEG Analysis
The aim of this lecture is to provide an illustrative tutorial on the methods for single-trial EEG analysis. Basic concepts of feature extraction and classification will be explained in a way that is accessible to participants from non-technical areas for BCI research in order to facilitate the interdisciplinary exchange. The tutorial will provide the foundation for subsequent more advanced data analysis lectures.

16:00 - 16:30   Coffee Break
16:30 - 18:00   Peter Desain
New methods and application domains for stimulus driven BCI's
EEG based Brain Computer Interfaces have matured considerably since their first appearance. However, one central aspect, their online character, still seems underexploited. This aspect allows for a more optimal stimulus selection and we will explain how this can be used for well-known (speller) BCI's and can increase their performance. When the stimulus and response space is very large, as it is in less common (word-probing) BCI's this approach becomes essential. Another domain in which our knowledge on reliable detection of EEG markers still needs be applied more thorough is plasticity and training. While general neuro-feedback on features of the mental state still seems unable to demonstrate lasting cognitive or behavioral changes, feedback on the processing of stimuli may allow for more efficient training of a whole range of perceptual skills. We will explain how we are investigating this approach for second language learning.

Friday, September 21, 2012

09:00 - 10:30   Gabriel Curio
Recording spikes … noninvasively?
The detectability of spikes defines a striking contrast between invasive (microscopic) and noninvasive (macroscopic) neurophysiological measurements. While noninvasive records are dominated by summed postsynaptic potentials reflecting neuronal input, invasive electrodes can provide easy access also to the very output of neuronal computation – spikes. This micro/macro gap, however, has been narrowed gradually over the last years by combining special physiological paradigms with neurotechnological advances. To convey the gist of this research agenda on high-frequency spike-related EEG/MEG the tripartite lecture will (i) address the basic neurophysics of near-field and far-field signals distinguishing slow from fast neuronal activities, (ii) elaborate on high-frequency (~600 Hz) somatosensory evoked responses serving as ‘workhorse’ to establish the feasibility of noninvasive spike-related recordings also in non-specialised labs, and (iii) report on recent neurotechnological progress providing a unique opportunity for high-resolution scalp mappings of EEG activities even above 1 kHz which reflect noninvasive correlates of human neocortical population spike responses.

10:30 - 11:00   Coffee Break
11:00 - 12:30   Lars-Kai Hansen
Resampling based methods for design and evaluation of neurotechnology.
Brain imaging by PET, MR, EEG, and MEG has become a cornerstone in systems level neuroscience. Statistical analyses of neuroimage datasets face many interesting challenges including non-linearity and multi-scale spatial and temporal dynamics. The objectives of neuroimaging are dual, we are interested in the most accurate, i.e., predictive, statistical model, but equally important is model interpretation and visualization which often takes the form of “brain mapping”. I will introduce some current machine learning strategies invoked for explorative and hypothesis driven neuroimage modeling, and present a general framework for model evaluation, interpretation, and visualization based on computer intensive data re-sampling schemes. Within the framework we obtain both an unbiased estimate of the predictive performance and of the reliability of the brain map visualization.

12:30 - 14:00   Lunch & Transit
14:00 - 18:00
 
[QE]
[VS]
Practical Sessions (parallel)
Quality Engineering: Assessing Audio- & Video Quality in Telecommunications (40 spaces)
BCI Experiment with a Gaze-Independent Visual Speller (20 spaces)

Monday, September 24, 2012

09:00 - 10:30   Ricardo Vigário
Source Separation and Clustering
Synchrony (or phase-locking) phenomena have been shown to exist in multiple oscillating systems such as electrical circuits, lasers, chemical reactions, and human neurons. If the measurements of these systems cannot detect the individual oscillators but rather a superposition of them, spurious phase locking will be detected. Current source-extraction techniques attempt to undo this superposition by assuming properties on the data, which are not valid when underlying sources are phase-locked. Statistical independence of the sources is one such invalid assumption, as phase-locked sources are dependent. In this presentation, we will review own methods for source separation and clustering, which make adequate assumptions for data where synchrony is present, and show with simulated data that they perform well even in cases where independent component analysis and other well-known source separation methods fail. The results provide a proof of concept that synchrony-based techniques are useful for low-noise applications. Although the presentation will focus on synchrony source separation methodology, the main motivation comes from a raising awareness of the important role synchrony plays in the interaction between distinct brain regions. For example, a muscle's electromyogram oscillates coherently with several brain regions, when a person is involved in a motor task. Also,memorization, learning, autism, Alzheimer's and Parkinson's disease, as well as epilepsy are examples of neuroscience topics associated with synchrony.

Sorry, no slides available!
10:30 - 11:00   Coffee Break
11:00 - 12:30   Fabien Lotte
EEG feature representations and associated spatial filters for Brain-Computer Interfaces
When designing EEG-based Brain-Computer Interfaces (BCI), a crucial signal processing component is the feature extraction step. It consists in representing EEG signals by a number of values that describe the relevant information they contain. This lecture will first expose the main features that are used to represent EEG signals such as Motor Imagery or P300. However, due to volume conduction, EEG signals inherently have a low spatial resolution, and the information they contain is generally spread over several channels. This makes features extracted individually from each EEG channel not as efficient as it could be. To alleviate this issue and improve the signal-to-noise ratio, it is important to use spatial filtering algorithms, in order to gather the relevant information from multiple channels. Therefore, this lecture will also present the spatial filter algorithms that can be used for each feature representation. This will include inverse solutions, Common Spatial Patterns (CSP) and variants, or the xDAWN algorithm, among other.

12:30 - 14:00   Lunch & Transit
14:00 - 18:00  
[BT]
[HF]
[AS]
[PP]
Tutorial on the BBCI Matlab Toolbox for Online Classification (30 spaces)
High Frequency Somatosensory Potentials and Cortico-Muscular Coherence (15 spaces)
Auditory Event-Related Potentials of the EEG for Spelling with a BCI (10 spaces)
Psychophysical Experiments on Visual Search  (5 spaces)

Tuesday, September 25, 2012

09:00 - 10:30   Felix Wichmann
System Identification Using Machine Learning Methods
Understanding perception and the underlying cognitive processes on a behavioral level requires a solution to the feature identification problem: Which are the features on which sensory systems base their computations? What techniques can we use to identify them? Thus one of the central challenges in psychophysics is System Identification: We need to infer the critical features, or cues, human observers make use of when they see or hear. What aspect of the visual or auditory stimulus actually influences behaviour if faced with real-world, complex stimuli? In my laboratory we have developed exploratory, data-driven system identification techniques based on modern machine learning methods to infer the critical features from human behavioural judgments. I will present these methods and show what their benefits are over the traditional “reverse-correlation” approach and the “bubbles technique”.

10:30 - 11:00   Coffee Break
11:00 - 12:30   Masashi Sugiyama
Density Ratio Estimation in Machine Learning.
In statistical machine learning, avoiding density estimation is essential because it is often more difficult than solving a target machine learning problem itself.  This is often referred to as Vapnik's principle, and the support vector machine is one of the successful realizations of this principle.  Following this spirit, a new machine learning framework based on the ratio of probability density functions has been introduced recently.  This density-ratio framework includes various important machine learning tasks such as transfer learning, outlier detection, feature selection, clustering, and conditional density estimation.  All these tasks can be effectively and efficiently solved in a unified manner by direct estimating the density ratio without going through density estimation. In this lecture, I give an overview of theory, algorithms, and application of density ratio estimation.

12:30 - 14:00   Lunch & Transit
14:00 - 18:00  
[LF]
[MR]
[HF]
[AS]
[PP]
Practical Sessions (parallel)
Low Field NMR/MRI Measurements and Analysis (10 spaces)
Introduction to fMRI Acquisition and Analysis (20 spaces)
High Frequency Somatosensory Potentials and Cortico-Muscular Coherence (15 spaces)
Auditory Event-Related Potentials of the EEG for Spelling with a BCI (10 spaces)
Psychophysical Experiments on Visual Search (5 spaces)

Wednesday, September 26, 2012

09:00 - 10:30   Aapo Hyvarinen
Separating sources and analysing connectivity in EEG/MEG using probabilistic models.
Currently, there is increasing interest in analysing brain activity in resting state, or under relatively natural conditions such as while watching a movie. When using functional magnetic resonance imaging (fMRI), such analysis is typically done by independent component analysis (ICA). However, there has not been very much work on analysing data measured by EEG or MEG in similar conditions. We have been recently developing various probabilistic methods for that purpose.  First, we have created new variants of ICA to more effectively separate sources of brain activity by exploiting the special structure of EEG/MEG data. Second, we have developed tests of the statistical significance of the independent components. Third, we have a developed a framework for analysis of causality (connectivity) which uses the non-Gaussianity of the data in the context of Bayesian networks or structural equation models. In this talk, I will give a short introduction to the theory of ICA, and then I will discuss these recent developments.

10:30 - 11:00   Coffee Break
11:00 - 12:30   Stefan Haufe & Guido Nolte
Inverse Methods for EEG and MEG source reconstruction
In this lecture we review the most popular inverse methods for EEG and MEG source reconstruction. Inverse methods can be divided into three different catagories: a) overdetermined models, b) underdetermined dipole field reconstructions with additional constraints, and c) scanning methods resulting in pseudoimages of  neural activity. We will present an introductory course into the following methods: 1. Dipole models, 2. Minimum norm solutions and variants (L2, L1, mixed norms, S-FLEX), 2. Beamformers (SAM, LCMV-beamformer, DICS) , and 4. subspace methods (MUSIC and RAP-MUSIC). The aim of this lecture is to explain the main concepts behind these methods, and to illustrate respective strengths and weaknesses in mostly simulated data.

12:30 - 14:00   Lunch & Transit
14:00 - 18:00  
[LF]
[MR]
[BT]
[PP]
[VS]
Practical Sessions (parallel)
Low Field NMR/MRI Measurements and Analysis (10 spaces)
Introduction to fMRI Acquisition and Analysis (20 spaces)
Tutorial on the BBCI Matlab Toolbox for Online Classification (15 spaces)
Psychophysical Experiments on Visual Search (5 spaces)
BCI Experiment with a Gaze-Independent Visual Speller (10 spaces)

Thursday, September 27, 2012

09:00 - 10:30   Nikolaus Kriegeskorte
Pattern-information analysis: columnar sensitivity, stimulus decoding, and computational-model testing
Pattern-information fMRI has become a popular method in neuroscience. The technique is motivated by the idea that spatial patterns of fMRI activity reflect the neuronal population codes of perception, cognition, and action. I will address the question to what extent we can expect to investigate columnar-scale neuronal pattern information with fMRI at 3T using conventional resolutions of 2-3 mm voxel width. I will then compare existing approaches with a focus on the question of what we can learn from them in terms of brain theory. The most popular and widespread method is stimulus decoding by response-pattern classification. This approach addresses the question whether activity patterns in a given region carry information about the stimulus category. Pattern classification uses generic models of the stimulus-response relationship that do not mimic brain information processing and treats the stimulus space as categorical – a simplification that is often helpful, but also limiting in terms of the questions that can be addressed. Beyond pattern classification, a major new direction is the integration of computational models of brain information processing into pattern-information analysis. This approach enables us to address the question to what extent competing computational models are consistent with the stimulus representations in a brain region. Two methods that test computational models are voxel receptive-field modeling and representational similarity analysis. These methods sample the stimulus (or mental-state) space more richly, estimate a separate response pattern for each stimulus, and can generalize from the stimulus sample to a stimulus population.

10:30 - 11:00   Coffee Break
11:00 - 12:30   Jens Steinbrink, Christoph Schmitz & Jan Mehnert
Near-Infrared Spectroscopy: Impact on BCI
Near-Infrared Spectroscopy (NIRS) measures the hemodynamic changes following focal neuronal activity in the brain non-invasively. NIRS affords imaging of neocortical functions in a mobile, unobtrusive, and economical way, promising a high impact on future BCI applications. In this session we will introduce the principles of NIRS, i.e. the biological and physical background of the measurement, its instrumentation, and some recent applications in the field of BCI.
In the experimental session we will explain the principles of a NIRS scanner and demonstrate its use and the preparation of a typical experimental setup. We will perform a physiological experiment of measuring the hemodynamic changes caused by executed and imaginary hand-gripping. The experimental results will then be analyzed using methods commonly applied to NIRS signals.

12:30 - 14:00   Lunch & Transit
14:00 - 18:00  
[LF]
[NI]
[PY]
[PP]
[VS]
Practical Sessions (parallel)
Low Field NMR/MRI Measurements and Analysis (10 spaces)
Near-Infrared Spectroscopy (NIRS) Experiment and Analysis (20 spaces)
Programming Stimulus Presentation and Feedback in the Pyff Framework (10 spaces)
Psychophysical Experiments on Visual Search (5 spaces)
BCI Experiment with a Gaze-Independent Visual Speller (15 spaces)

Friday, September 28, 2012

09:00 - 10:30   Femke Nijboer
Neurotechnologies on the rise - an interactive introduction to moral, legal and social issues
Existing and emerging neurotechnologies confront us - like no other technology before - with the question what it means to be human. How make-able are we? Are we just our brains? How much technology do we want in our body? Can we become superhumans? In this lecture we will discuss several ethical, moral, legal and social issues related to the research and development of neurotechnologies. Topics we will touch upon include informed consent, risk/benefit analysis, communication to the media, mind reading, personhood and cognitive enhancement.

10:30 - 11:00   Coffee Break
11:00 - 12:30   Andrew Schwartz
Neural Population Function
A better understanding neural population function would be an important advance in systems neuroscience.  The change in emphasis from the single neuron to the neural ensemble has made it possible to extract high-fidelity information about movements that will occur in the near future.  This ability is due to the distributed nature of information processing in the brain. Neurons encode many parameters simultaneously, but the fidelity of encoding at the level of individual neurons is weak.  However, because encoding is redundant and consistent across the population, extraction methods based on multiple neurons are capable of generating a faithful representation of intended movement.  The realization that useful information is embedded in the population has spawned the current success of brain-controlled interfaces.  Since multiple movement parameters are encoded simultaneously in the same population of neurons, we have been gradually increasing the degrees of freedom (DOF) that a subject can control through the interface.  Our early work showed that 3-dimensions could be controlled in a virtual reality task.  We then demonstrated control of an anthropomorphic physical device with 4 DOF in a self-feeding task.  Currently, monkeys in our laboratory are using this interface to control a 7-DOF arm, wrist and hand to grasp objects in different locations and orientations.  Our recent data show that we can extract 10-DOF to add hand shape and dexterity to our control set.

Sorry, no slides available!
12:30 - 14:00   Lunch & Transit
14:00 - 18:00  
[LF]
[NI]
[PY]
[PP]
Practical Sessions (parallel)
Low Field NMR/MRI Measurements and Analysis (10 spaces)
Near-Infrared Spectroscopy (NIRS) Experiment and Analysis (20 spaces)
Programming Stimulus Presentation and Feedback in the Pyff Framework (25 spaces)
Psychophysical Experiments on Visual Search (5 spaces)