# MI-017: Stationary Common Spatial Patterns for Brain-Computer Interfacing

> Generated from a local paper card. Do not edit this page directly; edit the source card and rerun `vp run docs:generate`.

> Internal wiki boundary: paper-card evidence and user-provided PDF downloads are available for private reading. Extracted full-paper text is not published.

> Source: `library/paper_cards/MI-017.md`

## Paper Access

* Internal PDF: <a href={"/papers/MI-017.pdf"} download style={{ display: "inline-flex", alignItems: "center", justifyContent: "center", minHeight: "2.25rem", padding: "0.45rem 0.8rem", borderRadius: "6px", backgroundColor: "#047857", color: "#ffffff", fontWeight: 700, lineHeight: 1, textDecoration: "none", boxShadow: "0 1px 2px rgba(15, 23, 42, 0.22)" }}>Download Paper</a>
* Deployment boundary: these PDF links are intended only for a private/protected internal wiki.

## MI-017: Stationary Common Spatial Patterns for Brain-Computer Interfacing

## Metadata

* ID: MI-017
* Title: Stationary Common Spatial Patterns for Brain-Computer Interfacing
* Year: 2012
* DOI / URL: needs confirmation
* Local PDF: see Paper Access section above
* Text artifact: local-only path withheld from docs site
* Review status: `extracted`

## Study Type

* Track: MI
* Task: address nonstationarity in CSP-based BCI classification
* Participants or dataset: three different datasets; exact names and participant counts need deeper extraction
* Device/electrode setup: EEG datasets with CSP-based processing
* Protocol/task: motion-intention / motor-imagery BCI classification

## Methods

* Signal processing or analysis: stationary CSP regularizes CSP toward stationary subspaces
* Comparator: state-of-the-art CSP variants
* Online/offline: offline evaluation on multiple datasets

## Key Results

* The abstract states that sCSP increases classification accuracy, especially for subjects with nonstationary signal components.
* The paper directly targets CSP sensitivity to changing EEG signal properties.

## Limitations

* DOI/landing page needs confirmation.
* Offline datasets do not prove online stability in SAH-BRI-Grasp.
* Exact dataset and statistical details need extraction before quantitative claims.

## Relevance To Current Review

* Supports the MI narrative that nonstationarity is a core reason offline MI performance does not directly transfer to robust online control.
* Useful bridge from CSP/FBCSP to adaptation and transfer-learning methods.

## Evidence Status

| Claim | Status | Evidence Note |
| --- | --- | --- |
| CSP is sensitive to nonstationarity. | verified | Abstract states CSP is not invariant to variations in signal properties. |
| sCSP regularizes CSP toward stationary subspaces. | verified | Abstract and methods describe this goal. |
| sCSP solves project false activations. | needs confirmation | No SAH-BRI-Grasp no-control experiment is reported. |

## Open Questions

* Confirm DOI and final publication metadata.
* Which nonstationarity tests should be included in Exp2?