Improved Activity Recognition Through Fusion of Earable Pairs
Bibliographic record
Abstract
Activity recognition using head-mounted sensors within devices like earbuds and hearing aids is an appealing option because of their wide adoption and social acceptance. Since these devices are typically used in pairs, combining the data for activity recognition could possibly increase accuracy. Inertial measurement units (IMUs) combine an accelerometer and gyroscope allowing for the recording of motion in devices. In this paper, we positioned an IMU near each ear by mounting them on eye-glass frames to record data and validate the question: Does using data from a pair of IMUs improve human activity recognition (HAR) compared to a single IMU? After recording data on 3 subjects, we applied 5 approaches for Resnet classification to the data using a train on 2, leave 1 subject out. Two methods used data from a single ear (Model 1-left or Model 2-right) while three used fusion methods combining the two ears. Fusion of the results of left and right ear prediction based on model confidence measures (Model 3) was compared to using a single method model that was trained on the combined the left and right data (Model 4) and a method that averaged the left and right data before training and testing (Model 5). We found a consistent benefit from using the fusion methods (Methods 3,4,5). The model accuracy increased by 2-4%, with the biggest improvement associated with Method 3. We then used the model’s activation layers to show that the fusion methods can improve accuracy by distinguishing between samples which the individual models are uncertain about.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.002 | 0.004 |
| Meta-epidemiology (narrow) | 0.002 | 0.000 |
| Meta-epidemiology (broad) | 0.002 | 0.002 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.002 |
| Open science | 0.001 | 0.002 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.003 | 0.002 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".