Asynchronous Averaging of Gait Cycles for Classification of Gait and Device Modes

Fredrik Gustafsson; Gustaf Hendeby; Parinaz Kasebzadeh

arxiv: 1907.02329 · v2 · pith:ELB2J7XGnew · submitted 2019-07-04 · 📡 eess.SP · cs.LG

Asynchronous Averaging of Gait Cycles for Classification of Gait and Device Modes

Parinaz Kasebzadeh , Gustaf Hendeby , Fredrik Gustafsson This is my paper

Pith reviewed 2026-05-25 09:16 UTC · model grok-4.3

classification 📡 eess.SP cs.LG

keywords gait signatureIMUdevice modeFourier seriesclassificationwearable sensorssignal segmentationactivity recognition

0 comments

The pith

Gait signatures from IMU data enable high-accuracy classification of both gait mode and device mode for each step cycle.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper proposes computing a unique gait signature that represents one full cycle of human gait from body-worn IMU measurements. The signature is created by precise segmentation and resampling of the signal, tuned by minimizing variability within each cycle, followed by a Fourier series expansion that yields a low-dimensional feature vector. This construction supports both offline and online classification of gait mode as well as joint classification of gait mode and device mode, where the latter identifies how the IMU-equipped device is carried. A sympathetic reader would care because the approach uses only standard wearable sensors to distinguish these modes without extra hardware, and the evaluation on a large multi-subject dataset with two gait modes and four device modes shows high classification rates per step cycle.

Core claim

What carries the argument

The gait signature obtained through asynchronous averaging of segmented IMU cycles, tuned by minimizing intra-cycle variability, and represented via Fourier series coefficients.

Load-bearing premise

Precise segmentation and resampling of the IMU signal, tuned by minimizing variability of the obtained signature within each gait cycle, produces a unique signature suitable for joint classification of gait and device modes.

What would settle it

A test showing that the per-cycle classification accuracy for gait and device modes using these signatures does not exceed the accuracy obtained from standard non-averaged IMU features on the same dataset.

read the original abstract

An approach for computing unique gait signature using measurements collected from body-worn inertial measurement units (IMUs) is proposed. The gait signature represents one full cycle of the human gait, and is suitable for off-line or on-line classification of the gait mode. The signature can also be used to jointly classify the gait mode and the device mode. The device mode identifies how the IMU-equipped device is being carried by the user. The method is based on precise segmentation and resampling of the measured IMU signal, as an initial step, further tuned by minimizing the variability of the obtained signature within each gait cycle. Finally, a Fourier series expansion of the gait signature is introduced which provides a low-dimensional feature vector well suited for classification purposes. The proposed method is evaluated on a large dataset involving several subjects, each one containing two different gait modes and four different device modes. The gait signatures enable a high classification rate for each step cycle.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

The paper gives a workable IMU pipeline that segments cycles, minimizes intra-cycle variability, and uses Fourier features for joint gait-plus-device classification on a multi-subject set.

read the letter

The core contribution is a concrete pipeline: precise segmentation and resampling of IMU signals, followed by variability minimization across cycles to produce a stable gait signature, then a low-order Fourier expansion for classification features. This lets them classify both gait mode and device mode (how the sensor is carried) in one go. On their dataset with several subjects, two gait modes, and four device modes, the signatures support per-cycle classification at what the abstract calls high rates. That joint-classification angle is useful for real wearable setups where placement isn't fixed. The approach is straightforward and the logic holds without circularity or hidden assumptions that break the claim. The main soft spot is that the abstract gives no numbers, baselines, or cross-validation details, so it's unclear how much the variability step actually improves over plain resampling plus FFT, or how sensitive results are to segmentation errors. If the full paper shows solid held-out performance and reasonable comparisons, the work is fine. This is for people working on activity recognition or health wearables who need to handle variable device placement. It is not foundational but it is a solid applied methods paper that deserves a serious referee to check the numbers and any implementation choices.

Referee Report

0 major / 3 minor

Summary. The manuscript proposes a method to compute unique gait signatures from body-worn IMU data via precise segmentation and resampling of the signal, followed by variability minimization within each gait cycle and Fourier series expansion to produce low-dimensional features. These signatures are used for joint classification of gait mode and device mode (how the device is carried). The approach is evaluated on a large multi-subject dataset containing two gait modes and four device modes, with the central claim that the signatures enable high per-cycle classification rates.

Significance. If the reported classification performance holds under proper validation, the work would be significant for wearable sensing applications in activity recognition and context-aware systems, as it provides a pipeline that jointly handles gait and device-mode classification without requiring subject-specific calibration. The variability-minimization step and Fourier feature extraction constitute a coherent, data-driven approach to obtaining cycle-level signatures.

minor comments (3)

[Abstract and Results] The abstract asserts 'high classification rate' without any numerical values, baselines, or error metrics; the results section should explicitly report per-class accuracies, confusion matrices, and at least one standard baseline (e.g., raw IMU features or DTW-based classification) to substantiate the claim.
[Method] The description of the variability-minimization procedure (used to tune the signature) should specify the exact objective function, any regularization, and whether the procedure introduces free parameters that must be chosen per subject or per dataset.
[Feature Extraction] Notation for the Fourier coefficients and the truncation order should be defined consistently; it is currently unclear how the low-dimensional feature vector dimension is selected and whether it is fixed across all experiments.

Simulated Author's Rebuttal

0 responses · 0 unresolved

We thank the referee for the positive assessment of our manuscript, the recognition of its potential significance for wearable sensing, and the recommendation for minor revision. No specific major comments were enumerated in the report, so we have no individual points requiring direct rebuttal or clarification at this stage. We will incorporate any minor polishing or validation clarifications in the revised version to ensure the classification results are presented with full rigor.

Circularity Check

0 steps flagged

No significant circularity

full rationale

The paper's pipeline (precise segmentation/resampling of IMU signals, variability minimization per cycle, Fourier expansion to low-dimensional features, and classification of gait+device modes) is evaluated on an external multi-subject dataset. No step reduces by construction to its own inputs, fitted parameters renamed as predictions, or load-bearing self-citations. The classification performance claim is presented as an empirical outcome rather than a tautology.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

No free parameters, axioms, or invented entities are specified in the abstract.

pith-pipeline@v0.9.0 · 5693 in / 982 out tokens · 28124 ms · 2026-05-25T09:16:38.746936+00:00 · methodology

discussion (0)

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

IndisputableMonolith/Cost/FunctionalEquation.lean washburn_uniqueness_aczel unclear

?

unclear
Relation between the paper passage and the cited Recognition theorem.

precise segmentation and resampling ... minimizing the variability ... Fourier series expansion ... low-dimensional feature vector
IndisputableMonolith/Foundation/DimensionForcing.lean reality_from_one_distinction unclear

?

unclear
Relation between the paper passage and the cited Recognition theorem.

Algorithm 1 ... optimal gait cycle segments ... 8 iterations ... K=8 Fourier

What do these tags mean?

matches: The paper's claim is directly supported by a theorem in the formal canon.
supports: The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends: The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses: The paper appears to rely on the theorem as machinery.
contradicts: The paper's claim conflicts with a theorem or certificate in the canon.
unclear: Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.