Learning from other Subjects Helps Reducing Brain-Computer Interface Calibration Time Fabien LOTTE, Cuntai GUAN, Institute for Infocomm Research (I2R), Singapore Introduction

Learning from other subjects

Evaluation

• Brain-Computer Interfaces (BCI) are communication systems that enable users to send commands to computer by using only their brain activity, usually measured by electroencephalography (EEG)

• CSP and LDA are both based on covariance matrix estimation

EEG data • BCI competition IV, data set 2a [4] • 9 subjects performing motor imagery (left and right hand) • 72 trials per class for training & testing

• BCI suffer from long calibration times due to the need to collect numerous EEG signals from the target user, for machine learning purposes. This is both inconvenient and uncomfortable • We propose an algorithm to use EEG data collected from previous subjects in the BCI learning process, to significantly reduce its calibration time

Method • We focus on a popular and efficient BCI design, based on Common Spatial Patterns (CSP) [1] and Linear Discriminant Analysis (LDA) [2] • CSP learns discriminative spatial filters w which extremize the following objective function:

wC1 wT w(C1 + C 2 ) wT Ci: EEG spatial covariance matrix for class I

• LDA is a linear classifier which learns a discriminative hyperplane between classes with normal vector a and intercept b as follows

1 T b = − (µ1 + µ 2 )C −1 (µ1 − µ 2 ) 2

a=C

−1

(µ1 − µ 2 )

µi: mean for class i, C: global covariance matrix

T

• Proper estimation requires numerous examples which leads to long calibration times to collect these examples • We propose to solve this problem by using data from some other subjects (previously recorded) as a regularization term in the estimation process:

~ Ct = (1 − λ )Ct + λ

1 S t ( Ω)

∑C

i∈St ( Ω )

i

(1)

Results • Evaluation for various sizes of the training set • 1 subject is the target subject, the other 8 are in Ω • Comparison with • Standard CSP and LDA • Regularized CSP and LDA with diagonal loading [5]

Ct: covariance matrix for the target subject Ci: covariance matrix for the ith previous subject St(Ω): selected subset of previous subjects Ω λ: regularization parameter (λ ϵ [0,1]) ,

Selecting a relevant subset of additional subjects In order to build St(Ω) we adopted a Forward Floating Search [3] subject selection approach: Φ0={}; θ0=Ω ; set λ=1 in Eq. 1 While i<|Ω| Add step: • Select the subject sA whose addition leads to the best CSP+LDA accuracy on the target subject training set • Φi+1= Φi+{sA}; θi+1 = θi –{sA}; i=i+1 //i.e., Add this subject Remove step: • If removing subject sR improves the accuracy Then • Φi-1 = Φi –{sR}; θi-1 = θi +{sR};i=i-1 //i.e., Remove this subject • Go to “Remove step” Else • Go to “Add step” End If End While St(Ω) is the set Φi which led to the best accuracy The regularization parameter λ is selected using a heuristic

Conclusion • A new method to incorporate EEG data from other subjects into the BCI learning process • Applied to CSP and LDA, our method significantly outperformed standard CSP and LDA designs as well as CSP and LDA designs based on diagonal loading • Best improvements for small training set sizes, hence effectively reducing the BCI calibration time References [1] Blankertz et al, IEEE Sig. Proc. Magazine, 2008 [3] Pudil et al, Pattern Recognition, 1994 [5] Ledoit and Wolf, J. of Multivariate Analysis, 2004

[2] Lotte et al, J. of Neural Eng., 2007 [4] http://www.bbci.de/competition/iv/

Learning from other Subjects Helps Reducing Brain-Computer ...

Learning from other Subjects Helps Reducing Brain-Computer Interface Calibration Time. Fabien LOTTE, Cuntai GUAN, Institute for Infocomm Research (I2R), Singapore. Introduction ... Select the subject sA whose addition leads to the best.

53KB Sizes 3 Downloads 193 Views

Recommend Documents

Carroll School helps students overcome learning disabilities with ...
Margaret Kuzmicz, director of technology at Carroll School, wanted to give ... “We chose Google Apps for Education, [touchscreen] Chromebooks, and Google.

Learning to diminish the effects of proactive interference: Reducing ...
sulting from the prior learning of related materials and has been shown to play an ... adults are more susceptible to such interference than are young adults ...

Reducing Emissions from the Road Transport Sector in Turkey: The ...
Feb 17, 2017 - One of the most important causes of air pollution and climate change related fossil fuel emissions is the road transport sector. There is a close ...

Reducing Risk, Increasing Protective Factors: Findings from the ...
predicting the outcomes given combinations of the risk and protective factors. Results: Rage was the strongest risk factor for every health-compromising behavior ...

Public Health Reducing inequalities from injuries in ...
Jun 26, 2006 - Public-health burden from injuries and violence .... The costs to health services for treating injuries in the ...... 82 European Commission.

MES-English.com - flashcards - subjects
MES-English.com Flash Cards school subjects www.mes-english.com. Page 2. MES-English.com Flash Cards school subjects www.mes-english.com. + - x - ...

Complete Subjects and Predicates
Ms. Hale took a cab to the convention center. 3. ... The subways were nearly empty during the late night hours. 8. ..... What a close call that was .... Page 24 ...

Brown Memo from other faculty.pdf
Page 1 of 3. MEMO. Date: July 14, 2017. To: President Sylvia Mathews Burwell. Cc: Caleen Jennings, Chair of the President's Council on Diversity and. Inclusion; Don Williamson, Chair of the Grievance Committee. Re: Tenure Denial of Carolyn Brown. Dea

CUTTInG-EDGE STRATEGIES FROM OTHER ...
The vast expansion of the internet and broadband access to its resources have ..... best. So its telemarketers randomly made different offers (a steak dinner, free.

Other directives from the overlord.pdf
Page 1 of 1. Other Directives from the Overlord. Cleaning. Cleaning should be done weekly. Mark on the calendar when cleaning and related tasks are ...

Learning from Perfection
The ideological objection against adding human defined base features often leads to machine .... Van Belle attempted to apply genetic algorithms to checkers endgame databases, which proved to be unsuccessful. Utgoff developed the. ELF learning algori

Learning from Streams
team member gets as input a stream whose range is a subset of the set to be ... members see the same data, while in learning from streams, each team member.

Odhar subjects 90_fqp.pdf
Page 1 of 2. PÀ£Á9lPÀ ̧ÀPÁ9gÀ. ¥ÀzÀ« ¥ÀÆaÀð 2PÀët E ̄ÁSÉ ¥ÀzÀ« ¥ÀÆaÀð 2PÀët E ̄ÁSÉ. Formative Assessment (Test)Question Papers Model Pattern.

Reducing Systemic Cybersecurity Risk - OECD
Jan 14, 2011 - views of the OECD or of the governments of its member countries. ...... seeking to punish downloaders of copyright material, against the .... to focus more on the process of analysing risk rather than simply having a long list ... abou

Reducing Systemic Cybersecurity Risk - OECD.org
Jan 14, 2011 - patches to operating systems and applications; the deployment of anti-malware, firewall and intrusion detection products and services; the use ...

Protecting Vulnerable Subjects in Clinical ... - Semantic Scholar
States Department of Health and Human Services. The. Office for Human ... The list of human-subject research abuses in the United. States is regrettably long. ... cal investigators protected vulnerable research subjects by excluding them from ...

REDUCING NUCLEAR AND RADIOLOGICAL TERRORISM ...
existing security and safety systems mean that it would not be easy for ..... introduce at least modest security improvements, including alarm systems and ...

2ND SEM ALL SUBJECTS SYLLABUS.pdf
2ND SEM ALL SUBJECTS SYLLABUS.pdf. 2ND SEM ALL SUBJECTS SYLLABUS.pdf. Open. Extract. Open with. Sign In. Main menu. Displaying 2ND SEM ALL ...

CorrActive Learning: Learning from Noisy Data through ...
we present the past work done on modeling noisy data and also work done in the related .... suggest the utility of the corrActive learner system. 4.3 Synthetic experiments ... placing the human with an autmoatic oracle that has the true labels.

Learning to Design Organizations and Learning from ...
management texts include typologies of organizational structures, departmental ... a system that requires participants to adhere strictly to ... Expanding business degree programs .... enced by, broadening networks of widely distributed ele-.

Reducing Systemic Cybersecurity Risk - OECD
Jan 14, 2011 - such as satellites, cellular base stations and switches. ..... may also be difficult: investigations can be technically challenging and cross national.