Pith. sign in

REVIEW 4 major objections 5 minor 2 cited by

DGSense: A Domain Generalization Framework for Wireless Sensing

T0 review · 4 major / 5 minor · reviewed 2026-08-08 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read DGSense reports that a wireless sensing model trained only on source domains keeps working for unseen users, rooms, and locations, with no target-domain data, across WiFi, mmWave, and acoustic signals.

desk verdict A solid engineering paper on domain generalization for wireless sensing whose empirical claim is probably right, but the unstated ResNet pretraining and an unfair DA baseline comparison need fixing before I would trust the numbers. read the letter →

arxiv 2502.08155 v1 pith:AN5KCVYI submitted 2025-02-12 cs.LG cs.AI

Rui Zhou , Yu Cheng , Songlin Li , Hongwang Zhang , Chenxu Liu This is my paper · ORCID
classification cs.LGcs.AI
keywords domaingeneralizationwirelesssensingepisodictrainingvirtualdatagenerationWiFiCSIgesturerecognitionmmWaveactivityacousticfalldetectioncross-modalvariationalautoencoder
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

This paper proposes DGSense, a domain-generalization framework for wireless sensing, and reports that a sensing model trained on a few source domains keeps working in unseen domains—new users, new rooms, new locations—without any data from those domains and without retraining. The framework is designed to be general across sensing technologies, and it is demonstrated on WiFi gesture recognition, mmWave activity recognition, and acoustic fall detection. Reported average accuracies on unseen domains are 83.3% for new WiFi users, 81.7% for new WiFi rooms, 95.4% for new mmWave users and locations, and 95.0% for new acoustic users. A reader should care because existing wireless-sensing solutions for the domain-shift problem rely on domain adaptation and still need target-domain data; DGSense targets the stronger setting where the target domain is entirely absent at training time.

What carries the argument

The load-bearing object is the episodic-training loop paired with the virtual data generator. Each source domain gets its own feature extractor and classifier; a shared main network is trained by cycling through domains with three losses: the main extractor through the main classifier, the main extractor through each domain classifier, and each domain extractor through the main classifier. Forcing the main extractor to be readable by every domain classifier—and the main classifier to read every domain extractor—is what the paper identifies as the source of domain-independent features. The virtual data generator is a VAE: for images a single encoder/decoder pair, and for multi-modal WiFi data a cross-modal design in which one base modality (amplitude) is encoded and all modalities are decoded from the shared latent code, preserving inter-modal consistency. Spatial features come from ResNet18 with CBAM within each residual block, temporal features from a 1DCNN. The t-SNE plots and the rising per-episode accuracy of the held-out user (24.2% to 83.3% over five episodic steps) are presented as direct evidence that the main extractor's features become more domain-independent as episodic training proceeds.

What would settle it

Retrain the three DGSense pipelines with randomly initialized ResNet18 weights instead of pre-trained ones, holding every other component fixed; if the new-domain accuracies (83.3%, 81.7%, 95.4%, 95.0%) fall to the reported w/o-DG levels (roughly 20–25% on WiFi), then the pretrained initialization, not DGSense's domain-independence mechanism, carries the generalization result.

Watch

Extended reading notes

Core claim

On its own terms, the paper's discovery is that the domain-dependence problem in wireless sensing can be attacked with a general, task-agnostic recipe rather than task-specific signal processing: generate diverse virtual training data with a VAE-based generator (single-modal for images, cross-modal for multi-modality WiFi samples so that amplitude, phase, and spectrogram remain consistent), then train a main feature extractor and classifier episodically against per-domain feature extractors and classifiers so that the main network must classify through every domain's eyes. At test time only the main network runs. The paper reports that this recipe yields high accuracy in unseen domains—83.3% for new WiFi users, 81.7% for new rooms, 80.6% for new users in new rooms, 95.4% for new mmWave users and locations, and 95.0% for new acoustic users—compared with roughly 20–25% on WiFi without the framework. The feature extractors combine a pre-trained ResNet18 with CBAM attention for spatial inputs and a 1DCNN for temporal inputs.

Load-bearing premise

The load-bearing premise is that a ResNet18 the abstract calls pre-trained gives useful starting features for inputs that are not natural images—WiFi phase data, Doppler spectrograms, and compressed range-Doppler maps—and the paper never reports the pretraining source or an ablation without it.

Editorial extensions

If this is right

  • If DGSense generalizes as reported, a wireless sensing deployment can be built from a few users and rooms and shipped to new environments without a separate data-collection trip.
  • The same framework works across WiFi, mmWave, and acoustic signals, so a single training recipe can serve gesture recognition, activity recognition, and fall detection.
  • The cross-modal generator matters: replacing it with a multi-modal generator that noisifies each modality separately drops new-user WiFi accuracy (for example, from about 84.6% to 73.8% in one test room), so preserving inter-modal consistency is part of the generalization gain.
  • Performance scales with source diversity: 4–5 source domains and 16–20 real samples per class suffice, so the framework is practical in data-scarce settings.
  • Inference is real-time (42.7 ms for WiFi, 64.1 ms for mmWave, and 849.6 ms for acoustic including preprocessing), so the generalization benefit does not come at the cost of deployability.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • One implication the paper leaves implicit is that the claimed domain independence is only demonstrated within the range of variation spanned by the source domains (users, rooms, locations); a harder hold-out that differs in channel geometry, hardware, or frequency band would test whether the learned features are truly domain-independent rather than interpolation across familiar variations.
  • Because the VAE is trained on source-domain data, the virtual samples can only diversify the training set within the source distribution. A testable extension would be to monitor virtual-data quality (the reported roughly 97% agreement with real data) as the number of source domains shrinks, since the generator's diversity is bounded exactly where the framework needs it most.
  • The abstract's "pre-trained" ResNet is never ablated; if the pretrained initialization is the dominant contributor, then a much simpler sensing model might reproduce the new-domain accuracies, and the episodic-training contribution would need to be re-measured relative to that baseline.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 5 minor

Summary. The paper proposes DGSense, a domain generalization framework for wireless sensing. The framework has three main components: a VAE-based virtual data generator (single-modal or cross-modal) that augments the source-domain training set, a spatial-temporal feature extractor (ResNet18 with CBAM for images and 1DCNN for time series), and a training procedure in which a main network is trained alongside per-domain networks via losses that encourage the main feature extractor and main classifier to be compatible with all domain-specific components. The authors evaluate DGSense on WiFi gesture recognition, mmWave activity recognition, and acoustic fall detection, reporting average accuracies of 83.3% for new WiFi users, 81.7% for new WiFi rooms, 95.4% for new mmWave users and locations, and 95.0% for new acoustic users, with comparisons against no-DG baselines, domain adaptation methods, and prior wireless sensing systems. The central claim is that once the model is built, it generalizes to unseen domains without any target-domain data.

Significance. If the reported results hold, the paper would be a useful step toward practical wireless sensing because it addresses the domain-shift problem without target-domain data and demonstrates the approach across three different wireless modalities on commodity hardware. The evaluation is genuinely broad: three sensing tasks, different signal types, multiple leave-one-domain-out protocols, and ablations of the generator type, generation method, generalization method, and number of source domains/samples. The paper also explicitly acknowledges limitations such as single-person scenarios and predefined gesture vocabularies in Section VIII. However, the empirical claims currently rest on several unspecified design choices and on comparisons that are not always aligned with the baselines' intended settings. The main strength is the breadth of real-world evaluation; the main weakness is that the attribution of the gains to the proposed mechanism is not yet fully supported by the reported experiments.

major comments (4)
  1. [Abstract, Secs. III-E, V-B, VI-B] The abstract and the framework description call the spatial feature extractor a 'pre-trained Residual Network (ResNet)', and Sections V-B and VI-B say the feature extractor is 'ResNet18' (with CBAM in Section V-B), but the manuscript never states the pretraining source, the initialization procedure, whether the backbone is frozen or fine-tuned, or any ablation of this choice. Since this backbone is used in every reported system and in the w/o DG baseline, the cross-domain improvements attributed to virtual data generation and episodic training cannot be separated from the effect of the pretrained initialization. Please specify the pretraining setup and add an ablation (e.g., random initialization versus the chosen pretrained weights) with both in-domain and cross-domain accuracy; this is needed for reproducibility and for the attribution claim.
  2. [Sec. IV-C6] The comparison protocol is inconsistent with the design of the baselines. The text states that OneFi requires one or few labeled target samples (Sec. II-A1) and CsiGAN uses semi-supervised GAN training with some target-domain data (Sec. II-A1), yet Fig. 8 is obtained by letting 'each method utilized only the source domain data to train the model' (Sec. IV-C6). This removes the very information those methods are designed to exploit, so the conclusion that 'our method consistently achieved the highest accuracy' in this figure does not support superiority over OneFi or CsiGAN in their intended settings. Please either compare in the settings those methods define, or clearly report the comparison as a zero-shot constraint that lies outside the baselines' design, and include the numerical w/o DG values underlying Fig. 8.
  3. [Sec. III-E3, Eqs. (13)-(16)] The procedure described is not the episodic training of Li et al. [34], contrary to the statement that the paper adopts their strategy. The cited method forms meta-train/meta-test episodes from the source domains and updates the model on a pseudo-test domain to simulate domain shift. Here, for each source domain i, losses (13)-(15) are all computed on samples from that same domain with no held-out domain; the domain networks and the main network are trained on the same data, and the main network is never evaluated against a domain that was excluded from its parameter update during training. At minimum, the name should be changed to something like a multi-domain feature-critic training procedure, or the algorithm should be extended with actual meta-test episodes; as written, the paper does not provide evidence that 'episodic training' in the literature's sense is the mechanism behind the reported gains.
  4. [Secs. III-D, III-E, VII] The framework has at least five free parameters — λ in Eqs. (5) and (7), ω1 and ω2 in Eqs. (6) and (9), θ1 and θ2 in Eq. (16), and the virtual-to-real sample ratio — but no values or sensitivity analysis are reported anywhere, and no code or trained models are released. Given that Tables IV-VI and Figs. 7-14 are the entire empirical support for the central generalization claim, the absence of these settings makes the results unverifiable and prevents a reader from assessing whether the reported margins are robust. Please add a hyperparameter table and per-fold or leave-one-domain-out results with variance, and consider releasing the code.
minor comments (5)
  1. [Sec. IV-C2, Sec. V-C2] The virtual-data quality check is presented as verifying that virtual data 'followed the same distribution' as real data, but the protocol trains on real data and tests on virtual data generated from the same real samples via a VAE trained on those samples; high accuracy in this setup mostly reflects reconstruction fidelity. Please rephrase as a reconstruction-quality check and rely on the held-out-domain results in Section VII for the claim that virtual data improve generalization.
  2. [Sec. III-A] The symbol Ds is overloaded: Eq. (1) defines Ds as a set of source domains, while Eq. (3) redefines Ds as the training set containing all training samples. Please use distinct notation for the domain set and the training set.
  3. [Sec. VIII] The outlier detector mentioned in Section VIII is not described or evaluated anywhere in the paper; either remove the reference or provide details of how it is trained and how it affects the reported results.
  4. [Tables I-VI, Figs. 7-14] All reported accuracies are point estimates without standard deviations, confidence intervals, or per-fold values. With only four to six domains per experiment, the differences in tables such as Table V (95.3 vs. 96.5 vs. 97.5) may be within noise; please report the variability across folds or leave-one-domain-out splits.
  5. [Sec. III-C] The preprocessing description is underspecified for reproducibility: window sizes for the moving average or median filter, the threshold values for filtering, and the parameters of the Power Burst Curve segmentation are not given.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: cross-domain accuracies are measured on held-out real domains, not derived from fitted parameters or self-citations.

full rationale

No significant circularity found. The paper's central claim—that a model trained only on source domains can recognize new users, rooms, and locations—is an empirical generalization result evaluated on genuinely held-out target domains (Secs. IV-C3 to IV-C5, V-C3 to V-C4, VI-C2 to VI-C3) with no target-domain data in training. The main machinery, VAE-based virtual data generation (Eqs. 5-9) and episodic training imported from Li et al. [34] (Eqs. 13-16), operates on source-domain data only, and the reported accuracies are measured on unseen real-world domains rather than on quantities fitted from those domains. The virtual data quality check in Sec. IV-C2 is self-referential in that virtual samples are autoencoder reconstructions of real samples, so a classifier trained on real data can label them if reconstruction is faithful; however, this check is not load-bearing for the cross-domain claim, which rests on the held-out real-domain evaluations. The only self-citation (AdapLoc [28]) appears in the related-work discussion and is not used to justify the framework. The unspecified 'pre-trained' ResNet initialization is a legitimate reproducibility and confound concern, but it is not a circularity: the pretrained weights are not defined in terms of the reported accuracies, and the claimed generalization results would remain an empirical comparison even if the pretraining choice were a confound.

Assumptions & free parameters 4 free parameters · 4 assumptions · 0 invented entities

The central claim depends on assumptions about domain structure, virtual data validity, and transferability of pretrained features. The paper does not derive these from first principles; it evaluates them empirically on three datasets. Hyperparameters such as lambda, omega1/2, and theta1/2 are not reported, so the exact configuration is underspecified.

free parameters (4)
  • lambda (KL divergence weight in VAE loss) = not reported
    Eqs. (5) and (7) add lambda times the KL divergence to the reconstruction loss; the value is never specified, yet it controls how closely the latent space matches a normal distribution and thus the character of the generated virtual data.
  • omega1 and omega2 (noise mixing proportions in virtual data generation) = not reported
    Eqs. (6) and (9) define z_tilde = omega1 * z + omega2 * noise; the proportions are not given, and they determine the diversity versus fidelity of the virtual samples.
  • theta1 and theta2 (weights of episodic loss terms) = not reported
    Eq. (16) sums Loss1 + theta1 * Loss2 + theta2 * Loss3; these weights affect how strongly the main network is pushed toward domain-invariant features, and no values are reported.
  • virtual-to-real sample ratio = 1x in most experiments; 40/20 per fall in acoustic
    The number of generated virtual samples is an experimental choice: WiFi and mmWave use a 1:1 ratio, while the acoustic fall case generates 40 virtual falls per person or 20 per location to balance classes; Section VII-F shows accuracy depends on this ratio.
assumptions (4)
  • domain assumption Source and target domains share the same label space, and domain shifts are fully captured by user, location, and environment changes.
    The problem formulation in Section III-A defines domains exactly as individuals, locations, and environments, and the framework does not handle unseen classes or multi-person interactions, as acknowledged in Section VIII.
  • domain assumption Virtual samples generated by the VAE are label-preserving and follow the same distribution as real data, so augmenting the training set with them improves generalization.
    Section III-D assumes the generated samples can inherit the source label and increase diversity; Section IV-C2 checks distribution overlap on real data, but the assumption is carried into unseen-domain generalization.
  • domain assumption Episodic training from Li et al. [34] transfers from image recognition to wireless signals, so a main feature extractor that is classifiable by every domain classifier will produce domain-invariant features.
    Section III-E adopts the strategy wholesale, with no argument or proof specific to WiFi, mmWave, or acoustic data.
  • ad hoc to paper A ResNet18 pretrained on ImageNet provides a useful feature extractor for non-natural-image inputs such as phase spectrograms, Doppler spectrograms, and compressed range-Doppler maps.
    The abstract says 'pre-trained' but the body does not state the pretraining source or ablate the choice; the transfer from ImageNet to these signal representations is an unexamined assumption.

how reviews work

0 comments
Cite this review

Pith. "Pith review of DGSense: A Domain Generalization Framework for Wireless Sensing." pith.science (2026). https://pith.science/paper/AN5KCVYI

@misc{pith2026250208155,
  author       = {Pith},
  title        = {Pith review of: DGSense: A Domain Generalization Framework for Wireless Sensing},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/AN5KCVYI}},
  note         = {Machine review of arXiv:2502.08155}
}
read the original abstract

Wireless sensing is of great benefits to our daily lives. However, wireless signals are sensitive to the surroundings. Various factors, e.g. environments, locations, and individuals, may induce extra impact on wireless propagation. Such a change can be regarded as a domain, in which the data distribution shifts. A vast majority of the sensing schemes are learning-based. They are dependent on the training domains, resulting in performance degradation in unseen domains. Researchers have proposed various solutions to address this issue. But these solutions leverage either semi-supervised or unsupervised domain adaptation techniques. They still require some data in the target domains and do not perform well in unseen domains. In this paper, we propose a domain generalization framework DGSense, to eliminate the domain dependence problem in wireless sensing. The framework is a general solution working across diverse sensing tasks and wireless technologies. Once the sensing model is built, it can generalize to unseen domains without any data from the target domain. To achieve the goal, we first increase the diversity of the training set by a virtual data generator, and then extract the domain independent features via episodic training between the main feature extractor and the domain feature extractors. The feature extractors employ a pre-trained Residual Network (ResNet) with an attention mechanism for spatial features, and a 1D Convolutional Neural Network (1DCNN) for temporal features. To demonstrate the effectiveness and generality of DGSense, we evaluated on WiFi gesture recognition, Millimeter Wave (mmWave) activity recognition, and acoustic fall detection. All the systems exhibited high generalization capability to unseen domains, including new users, locations, and environments, free of new data and retraining.

Figures

Figures reproduced from arXiv: 2502.08155 by the authors.

Figure 1
Figure 1. The domain generalization framework (DGSense). [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. The virtual data generator. encoder is a CNN and the decoders are DCNNs. We first select one modality as the base modality, which has relationship with every other modality and acts as the base of generating other modalities. For virtual data generation, we input solely the base modality of the real sample, and the cross-modal generator will generate all the virtual modalities from the base modality. Hence in the cr… view at source ↗
Figure 3
Figure 3. Episodic training for domain independent feature ex [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figures from the paper (10 more)
Figure 4
Figure 4. Figure 4: Process and effect of episodic training on WiFi gestu [PITH_FULL_IMAGE:figures/full_fig_p007_4.png]
Figure 5
Figure 5. Figure 5: The feature extractor and classifier of WiFi gesture r [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]
Figure 6
Figure 6. Figure 6: WiFi experimental setup. (a) New user (5 source 1 new) (b) New user (4 source 2 new) (c) New room (d) New user and new room [PITH_FULL_IMAGE:figures/full_fig_p008_6.png]
Figure 7
Figure 7. Figure 7: WiFi gesture recognition in new domains. [PITH_FULL_IMAGE:figures/full_fig_p008_7.png]
Figure 8
Figure 8. Figure 8: Comparison of WiFi gesture recognition with existin [PITH_FULL_IMAGE:figures/full_fig_p009_8.png]
Figure 10
Figure 10. Figure 10: The feature extractor and classifier of mmWave activ [PITH_FULL_IMAGE:figures/full_fig_p009_10.png]
Figure 11
Figure 11. Figure 11: Experimental setup and mmWave activity recognitio [PITH_FULL_IMAGE:figures/full_fig_p010_11.png]
Figure 12
Figure 12. Figure 12: Method comparison in new domains. C. Evaluations The experiments were carried out in a laboratory, as shown in [PITH_FULL_IMAGE:figures/full_fig_p010_12.png]
Figure 13
Figure 13. Figure 13: Doppler effect and acoustic spectrograms. [PITH_FULL_IMAGE:figures/full_fig_p011_13.png]
Figure 14
Figure 14. Figure 14: Impact of number of source domains and samples. [PITH_FULL_IMAGE:figures/full_fig_p014_14.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 2 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. OpenCSI: Self-Calibration Layer for Heterogeneous Mesh Wireless Sensor Networks

    cs.ET 2026-07 conditional novelty 6.0 of 10

    OpenCSI normalizes each WiFi link by its own quiet-period fluctuation, letting one occupancy model work across rooms, ESP32 chips, and WiFi generations without retraining.

  2. Learning-Driven Channel Representation for Wireless Localization: From Channel Observations to Location Inference

    eess.SP 2026-07 conditional novelty 4.0 of 10

    A survey organizes learning-driven wireless localization into observation, channel representation, and location inference, arguing representation quality is the decisive performance factor.

Reference graph

Works this paper leans on

42 extracted references · 42 canonical work pages · cited by 2 Pith papers

  1. [34]

    Episodic training for domain generalization,

    D. Li, J. Zhang, Y . Y ang, C. Liu, Y .-Z. Song, and T. Hosped ales, “Episodic training for domain generalization,” in IEEE/CVF Interna- tional Conference on Computer Vision (ICCV) , 2019, pp. 1446–1455

    work page 2019

  2. [1]

    RF-based device-free recogni tion of simul- taneously conducted activities,

    S. Sigg, S. Shi, and Y . Ji, “RF-based device-free recogni tion of simul- taneously conducted activities,” in ACM Conference on Pervasive and Ubiquitous Computing Adjunct Publication (UbiComp Adjunc t). ACM, 2013, pp. 531–540

    work page 2013

  3. [2]

    Exploiting the wireless RF fading for human activity recognition,

    S. Orphomma and N. Swangmuang, “Exploiting the wireless RF fading for human activity recognition,” in 10th International Conference on Electrical Engineering/Electronics, Computer , Telecommunications and Information Technology, 2013, pp. 1–5

    work page 2013

  4. [3]

    WiFi sensing with channel sta te information: A survey,

    Y . Ma, G. Zhou, and S. Wang, “WiFi sensing with channel sta te information: A survey,” ACM Computing Surveys , vol. 52, no. 3, 2019

    work page 2019

  5. [4]

    Wireless sen sing for human activity: A survey,

    J. Liu, H. Liu, Y . Chen, Y . Wang, and C. Wang, “Wireless sen sing for human activity: A survey,” IEEE Communications Surveys and Tutorials, vol. 22, no. 3, pp. 1629–1645, 2020

    work page 2020

  6. [5]

    Real-time arm gesture recognition in smar t home scenarios via millimeter wave sensing,

    H. Liu, Y . Wang, A. Zhou, H. He, W. Wang, K. Wang, P . Pan, Y . L u, L. Liu, and H. Ma, “Real-time arm gesture recognition in smar t home scenarios via millimeter wave sensing,” Proc. ACM Interact. Mob. W earable Ubiquitous Technol., vol. 4, no. 4, 2020

    work page 2020

  7. [6]

    Human tracking and identification through a mil limeter wave radar,

    P . Zhao, C. X. Lu, J. Wang, C. Chen, W. Wang, N. Trigoni, and A. Markham, “Human tracking and identification through a mil limeter wave radar,” Ad Hoc Networks , vol. 116, p. 102475, 2021

    work page 2021

  8. [7]

    LoRaIn: Mak ing a case for LoRa in indoor localization,

    B. Islam, M. T. Islam, J. Kaur, and S. Nirjon, “LoRaIn: Mak ing a case for LoRa in indoor localization,” in IEEE International Conference on Pervasive Computing and Communications W orkshops (PerC om W orkshops), 2019, pp. 423–426

    work page 2019

Show all 42 references
  1. [8]

    Exploring LoRa for long-range through-wall sensing,

    F. Zhang, Z. Chang, K. Niu, J. Xiong, B. Jin, Q. Lv, and D. Zh ang, “Exploring LoRa for long-range through-wall sensing,” Proc. ACM Interact. Mob. W earable Ubiquitous Technol. , vol. 4, no. 2, 2020

    2020

  2. [9]

    Ubiquitous acoustic sensin g on commod- ity iot devices: A survey,

    C. Cai, R. Zheng, and J. Luo, “Ubiquitous acoustic sensin g on commod- ity iot devices: A survey,” IEEE Communications Surveys & Tutorials , vol. 24, no. 1, pp. 432–454, 2022

    2022

  3. [10]

    Push the limit of acousti c gesture recognition,

    Y . Wang, J. Shen, and Y . Zheng, “Push the limit of acousti c gesture recognition,” IEEE Transactions on Mobile Computing , vol. 21, no. 5, pp. 1798–1811, 2022

    2022

  4. [11]

    Predict able 802.11 packet delivery from wireless channel measurements,

    D. Halperin, W. Hu, A. Sheth, and D. Wetherall, “Predict able 802.11 packet delivery from wireless channel measurements,” in ACM SIG- COMM. ACM, 2010, pp. 159–170

    2010

  5. [12]

    Precise power delay profiling wi th commodity wifi,

    Y . Xie, Z. Li, and M. Li, “Precise power delay profiling wi th commodity wifi,” in 21st Annual International Conference on Mobile Computing and Networking , ser. MobiCom’15. ACM, 2015, pp. 53–64

    2015

  6. [13]

    Towards environme nt independent device free human activity recognition,

    W. Jiang, C. Miao, F. Ma, S. Y ao, Y . Wang, Y . Y uan, H. Xue, C . Song, X. Ma, D. Koutsonikolas, W. Xu, and L. Su, “Towards environme nt independent device free human activity recognition,” in 24th Annual International Conference on Mobile Computing and Networki ng (Mobi- Co...

    2018

  7. [14]

    CsiGAN: Robust channe l state information-based activity recognition with GANs,

    C. Xiao, D. Han, Y . Ma, and Z. Qin, “CsiGAN: Robust channe l state information-based activity recognition with GANs,” IEEE Internet of Things Journal, vol. 6, no. 6, pp. 10 191–10 204, 2019

    2019

  8. [15]

    Context-aware wireles s based cross domain gesture recognition,

    H. Kang, Q. Zhang, and Q. Huang, “Context-aware wireles s based cross domain gesture recognition,” IEEE Internet of Things Journal , pp. 1–1, 2021

    2021

  9. [16]

    OneFi: One-shot reco gnition for unseen gesture via COTS WiFi,

    R. Xiao, J. Liu, J. Han, and K. Ren, “OneFi: One-shot reco gnition for unseen gesture via COTS WiFi,” in 19th ACM Conference on Embedded Networked Sensor Systems , ser. SenSys’21, 2021, pp. 206–219

    2021

  10. [17]

    Widar3.0: Zero-effort cross-domain gesture recognition with wi-fi,

    Y . Zhang, Y . Zheng, K. Qian, G. Zhang, Y . Liu, C. Wu, and Z. Y ang, “Widar3.0: Zero-effort cross-domain gesture recognition with wi-fi,” IEEE Transactions on Pattern Analysis and Machine Intellig ence, vol. 44, no. 11, pp. 8671–8688, 2022

    2022

  11. [18]

    Interacting with soli: Exploring fine-grained dynamic gesture recognition i n the radio- frequency spectrum,

    S. Wang, J. Song, J. Lien, I. Poupyrev, and O. Hilliges, “ Interacting with soli: Exploring fine-grained dynamic gesture recognition i n the radio- frequency spectrum,” in Proceedings of the 29th Annual Symposium on User Interface Software and Technology , ser. UIST’16. ACM, 20...

    2016

  12. [19]

    Temporal-range-d oppler features interpretation and recognition of hand gestures u sing mmW FMCW radar sensors,

    G. Zhang, S. Lan, K. Zhang, and L. Y e, “Temporal-range-d oppler features interpretation and recognition of hand gestures u sing mmW FMCW radar sensors,” in 14th European Conference on Antennas and Propagation (EuCAP) , 2020, pp. 1–4

    2020

  13. [20]

    Radar human motion recogni tion using motion states and two-way classifications,

    M. G. Amin and R. G. Guendel, “Radar human motion recogni tion using motion states and two-way classifications,” in IEEE International Radar Conference (RADAR) , 2020, pp. 1046–1051

    2020

  14. [21]

    Human ac- tivity recognition using temporal 3DCNN based on FMCW radar ,

    H. Chen, C. Ding, L. Zhang, H. Hong, and X. Zhu, “Human ac- tivity recognition using temporal 3DCNN based on FMCW radar ,” in IEEE MTT-S International Microwave Biomedical Conference (IM- BioC), 2022, pp. 245–247

    2022

  15. [22]

    M-gesture: Person-independent real-time in- air gesture recognition using commodity millimeter wave radar,

    H. Liu, A. Zhou, Z. Dong, Y . Sun, J. Zhang, L. Liu, H. Ma, J. Liu, and N. Y ang, “M-gesture: Person-independent real-time in- air gesture recognition using commodity millimeter wave radar,” IEEE Internet of Things Journal, vol. 9, no. 5, pp. 3397–3415, 2022

    2022

  16. [23]

    DeepRange: Acoustic ranging via deep learning,

    W. Mao, W. Sun, M. Wang, and L. Qiu, “DeepRange: Acoustic ranging via deep learning,” Proc. ACM Interact. Mob. W earable Ubiquitous Technol., vol. 4, no. 4, 2020

    2020

  17. [24]

    EchoSpot: Spottin g your locations via acoustic sensing,

    J. Lian, J. Lou, L. Chen, and X. Y uan, “EchoSpot: Spottin g your locations via acoustic sensing,” Proc. ACM Interact. Mob. W earable Ubiquitous Technol., vol. 5, no. 3, 2021

    2021

  18. [25]

    EarIO: A low-power acoustic sensing earable for continuously track ing detailed facial movements,

    K. Li, R. Zhang, B. Liang, F. Guimbreti` ere, and C. Zhang , “EarIO: A low-power acoustic sensing earable for continuously track ing detailed facial movements,” Proc. ACM Interact. Mob. W earable Ubiquitous Technol., vol. 6, no. 2, 2022

    2022

  19. [26]

    Fall detection via inaudible acoustic sensing,

    J. Lian, X. Y uan, M. Li, and N.-F. Tzeng, “Fall detection via inaudible acoustic sensing,” Proc. ACM Interact. Mob. W earable Ubiquitous Technol., vol. 5, no. 3, 2021

    2021

  20. [27]

    Watching your phone’s back: Gesture recognition by sensin g acousti- cal structure-borne propagation,

    L. Wang, X. Zhang, Y . Jiang, Y . Zhang, C. Xu, R. Gao, and D. Zhang, “Watching your phone’s back: Gesture recognition by sensin g acousti- cal structure-borne propagation,” Proc. ACM Interact. Mob. W earable Ubiquitous Technol., vol. 5, no. 2, 2021

    2021

  21. [28]

    Ad ap- tive device-free localization in dynamic environments thr ough adaptive neural networks,

    R. Zhou, H. Hou, Z. Gong, Z. Chen, K. Tang, and B. Zhou, “Ad ap- tive device-free localization in dynamic environments thr ough adaptive neural networks,” IEEE Sensors Journal , vol. 21, no. 1, pp. 548–559, 2021

    2021

  22. [29]

    Deep adaptati on networks based gesture recognition using commodity WiFi,

    Z. Han, L. Guo, Z. Lu, X. Wen, and W. Zheng, “Deep adaptati on networks based gesture recognition using commodity WiFi,” in IEEE Wireless Communications and Networking Conference (WCNC) , 2020, pp. 1–7

    2020

  23. [30]

    FiDo: Ubiquitous fine-grained WiFi-based localization for unlab elled users via domain adaptation,

    X. Chen, H. Li, C. Zhou, X. Liu, D. Wu, and G. Dudek, “FiDo: Ubiquitous fine-grained WiFi-based localization for unlab elled users via domain adaptation,” in The W eb Conference (WWW) , 2020, pp. 23–33

    2020

  24. [31]

    Data augmentation and dense-LSTM for human activity recog nition using WiFi signal,

    J. Zhang, F. Wu, B. Wei, Q. Zhang, H. Huang, S. W. Shah, and J. Cheng, “Data augmentation and dense-LSTM for human activity recog nition using WiFi signal,” IEEE Internet of Things Journal , vol. 8, no. 6, pp. 4628–4641, 2021

    2021

  25. [32]

    Generalizing to unseen domains: A survey on domain generalization,

    J. Wang, C. Lan, C. Liu, Y . Ouyang, T. Qin, W. Lu, Y . Chen, W. Zeng, and P . Y u, “Generalizing to unseen domains: A survey on domain generalization,” IEEE Transactions on Knowledge and Data Engineering, pp. 1–1, 2022

    2022

  26. [33]

    Domain generaliz ation with adversarial feature learning,

    H. Li, S. J. Pan, S. Wang, and A. C. Kot, “Domain generaliz ation with adversarial feature learning,” in IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , 2018, pp. 5400–5409

    2018

  27. [35]

    Learning to learn single d omain generalization,

    F. Qiao, L. Zhao, and X. Peng, “Learning to learn single d omain generalization,” in IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , 2020, pp. 12 553–12 562

    2020

  28. [36]

    Deep residual learni ng for image recognition,

    K. He, X. Zhang, S. Ren, and J. Sun, “Deep residual learni ng for image recognition,” in 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) , 2016, pp. 770–778

    2016

  29. [37]

    Cbam: Convol utional block attention module,

    S. Woo, J. Park, J.-Y . Lee, and I. S. Kweon, “Cbam: Convol utional block attention module,” in Computer Vision – ECCV 2018 . Springer International Publishing, 2018, pp. 3–19

    2018

  30. [38]

    Visualizing data using t -sne,

    V . D. M. Laurens and G. Hinton, “Visualizing data using t -sne,” Journal of Machine Learning Research , vol. 9, no. 2605, pp. 2579–2605, 2008

    2008

  31. [39]

    PADS: Passiv e detection of moving targets with dynamic speed using phy layer informa tion,

    K. Qian, C. Wu, Z. Y ang, Y . Liu, and Z. Zhou, “PADS: Passiv e detection of moving targets with dynamic speed using phy layer informa tion,” in ICPADS, 2014, pp. 1–8

    2014

  32. [40]

    Inf erring motion direction using commodity Wi-Fi for interactive exe rgames,

    K. Qian, C. Wu, Z. Zhou, Y . Zheng, Z. Y ang, and Y . Liu, “Inf erring motion direction using commodity Wi-Fi for interactive exe rgames,” in CHI Conference on Human Factors in Computing Systems (CHI) . ACM, 2017, pp. 1961–1972

    2017

  33. [41]

    An unsupervised acoustic fall detection system using source s eparation for sound interference suppression,

    M. Salman Khan, M. Y u, P . Feng, L. Wang, and J. Chambers, “ An unsupervised acoustic fall detection system using source s eparation for sound interference suppression,” Signal Processing, vol. 110, no. C, pp. 199–210, 2015

    2015

  34. [42]

    Acoustic cues from the floor: A new approach for fall classifi cation,

    E. Principi, D. Droghini, S. Squartini, P . Olivetti, an d F. Piazza, “Acoustic cues from the floor: A new approach for fall classifi cation,” Expert Systems with Applications , vol. 60, 2016

    2016

Pith tools

Reviewed August 8, 2026 · model on record in the stance chip above.