Pith. sign in

REVIEW 2 major objections 5 minor 28 references

Toward Personalized Social Robots for Child Well-being: Data Requirement Principles from a Recommender-System Perspective

T0 review · 2 major / 5 minor · reviewed 2026-07-11 · grok-4.5

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

Pith's one-line read Personalization for social robots that support children in clinics is blocked by how data is collected, not by the models; four data principles can supply what is missing.

desk verdict Clean LBR that correctly diagnoses why clinical robot data cannot train personalization and gives four usable collection principles. read the letter →

arxiv 2607.05110 v1 pith:S77U3R2B submitted 2026-07-06 cs.HC cs.RO

Jin Huang , Eric Nichols , Fethiye Irmak Dogan , Hatice Gunes This is my paper · ORCID
classification cs.HCcs.RO
keywords socialrobotspersonalizationrecommendersystemschildwell-beingclinicalHRIdatacollectionprinciplesuserprofilingexposurebias
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

Social robots are already used in hospitals to comfort and screen children, but effective support must adapt to each child and to how that child changes from moment to moment and visit to visit. The paper treats choosing the best robot action as a recommendation problem and shows that a recent recommender-system framework—user profiling, ranking, and responsible computing—could in principle deliver that personalization. What blocks it is not the model but the data: existing studies reduce a whole session to one anonymized score on a single construct, so they cannot track a shifting state, recover which actions helped, link sparse visits, or correct for the fact that care is never randomized. The authors answer with four collection principles: an integrated multi-layer profile, per-action effectiveness signals (both explicit and implicit), linkable coverage across children and visits, and an exposure record of why each action was chosen, logged at the moment of selection. If these guidelines are followed, the same framework can begin to personalize within a session, across return visits, and even for children seen only once.

What carries the argument

The four data principles (integrated profile, effectiveness signals, linkable coverage, exposure record). They convert the four familiar recommender problems—non-stationary user state, weak implicit feedback, cold-start and unlinkable histories, and off-policy bias—into concrete collection guidelines that supply what user profiling, ranking, and responsible computing require.

What would settle it

Instrument a multi-site pediatric robot deployment with the four principles, train collaborative-filtering and inverse-propensity models on the resulting action-level logs, and test whether the ranked actions improve measured child anxiety or engagement relative to existing scripted or single-score baselines; clear failure of transfer would falsify the claim that the data principles alone unblock personalization.

Watch

Extended reading notes

Core claim

Instantiating a recommender-system framework for personalizing social-robot actions in child well-being is blocked not by the model but by the data. Existing hospital studies yield only fixed, single-construct, end-of-session scores that cannot track a shifting child state, recover per-action feedback, link sparse visits, or correct for non-random assignment. Four data principles—an integrated profile, effectiveness signals, linkable representative coverage, and an exposure record logged at collection time—directly answer these four challenges and map onto the framework’s profiling, ranking, and responsible-computing components.

Load-bearing premise

The claim rests on the premise that once multi-dimensional, linkable profiles and per-action signals exist, ordinary recommender techniques will usefully transfer preferences across sparse, high-stakes pediatric clinical interactions the way they do in everyday recommendation settings.

Editorial extensions

If this is right

  • Data collection for clinical social robots must log one record per robot action rather than one score per session.
  • Within-session personalization needs per-action signals plus population coverage; cross-session personalization additionally needs stable linkable identities.
  • Exposure (why an action was chosen) must be recorded at the moment of selection or it cannot be recovered later for debiasing.
  • Capabilities such as cold-start and cross-domain transfer can run on subsets of the four principles rather than requiring all of them at once.
  • Whether social robots can personalize therefore turns less on new models than on collecting the right data from the start.

Reading between the lines

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

  • Hospitals that adopt action-level logging with exposure records would create datasets usable for offline policy evaluation, reducing the need for new randomized trials each time a robot behavior is proposed.
  • The same four principles could transfer to other high-stakes, sparse-interaction domains such as elderly care or special education where randomization is ethically limited.
  • Privacy-preserving or federated collection will almost certainly be required to reconcile the demand for linkable profiles with the anonymization constraints the paper itself flags as open.
  • Empirical thresholds for scale, diversity, and temporal depth remain untested; small pilots that instrument only integrated profiles and effectiveness signals could already enable basic within-session ranking.
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

2 major / 5 minor

Summary. This Late Breaking Report argues that personalizing social robots for child well-being in clinical settings—framed as a recommendation problem via the authors’ prior RS-for-robots framework—is blocked by data design rather than by models. From a literature survey of robot-administered assessment studies (Table I), it diagnoses four challenges: non-stationary multi-dimensional child state, weak/indirect per-action feedback, sparse and unlinkable cross-session histories under anonymization, and observational (non-randomized) exposure. It proposes four corresponding data principles—integrated multi-layer profile, effectiveness signals aligned to actions, linkable representative coverage, and exposure logging at collection time—maps them to user profiling, ranking, and responsible computing (Table II), and states which subset each capability (within-session, cross-session, cold-start, cross-domain) requires. The discussion flags open practical questions of consent-constrained collection and empirical scale/diversity/depth.

Significance. If the diagnosis and principles hold, the paper supplies a clear, reusable checklist for data collection in personalized clinical HRI, connecting standard recommender-system problems (implicit feedback, cold-start, off-policy bias) to a high-stakes pediatric setting where existing practice yields only session-level anonymized scores. The literature mapping in Table I and the principle-to-component mapping in Table II are concrete contributions that can orient future datasets and deployments (e.g., Haru-style oncology screening). As a conceptual LBR it does not claim empirical transfer of CF/IPW to sparse pediatric interactions; that open question is appropriately deferred to §VI. The work is therefore significant as an agenda-setting data-requirements paper rather than as a validated system result.

major comments (2)
  1. [§IV Principle 3 / §VI] §IV Principle 3 and §VI: Linkable coverage is presented as essential for collaborative filtering and returning-child personalization, yet the same sections (and Challenge iii) note that consent and anonymization routinely break per-child identity. The paper gestures at “privacy-preserving or federated collection” but does not specify even a minimal mechanism (e.g., site-local stable pseudonyms, cross-site federated profiles, or what may be recorded without re-identification). Because P3 is load-bearing for cross-session and cold-start claims, a short operational sketch of how linkability can be achieved under the constraints the paper itself identifies would strengthen implementability.
  2. [§V Cold start] §V (Cold start): The suggested route of “a few probing actions, such as briefly asking the user” to obtain Principle-2 signals is in tension with the motivating setting—anxious children in oncology or procedural care—where extra questioning can itself raise burden or distress. The section should either qualify when direct elicitation is appropriate or emphasize passive implicit signals (gaze, silence, facial expression) already listed under Principle 2 as the primary cold-start path in clinical use.
minor comments (5)
  1. [Abstract / §I] Abstract and §I claim the principles are framed as “concrete guidelines for data collection.” The body gives clear principles and examples but not protocol-level guidelines (what fields to log, sampling rates, consent language). Softening “concrete guidelines” to “design principles / requirements” would better match the content of an LBR.
  2. [Table I] Table I is valuable; adding a brief note on whether any listed study logs per-action robot content (even if not used for personalization) would make the gap to Principle 2 even sharper.
  3. [Fig. 1] Fig. 1 caption and body: “P1–P4 refer to the four data principles of Section IV” is clear; ensure the figure itself labels P1–P4 consistently with the principle names (integrated profile, effectiveness signals, linkable coverage, exposure record) for readers who land on the figure first.
  4. [Author block / citations] Author-name encoding artifacts appear in the text (e.g., “Do ˘gan”, “Do˘gan”); clean for the camera-ready version.
  5. [§II.B] §II.B: A one-sentence reminder of what “ranking” and “responsible computing” consume as inputs would help readers who have not read [13], given how heavily Table II depends on those components.

Circularity Check

1 steps flagged · score 1.0 of 10

Minor self-citation of the authors' own RS framework; the four data principles are independently motivated by external clinical studies and standard RS problems, with no reduction-by-construction.

  1. self citation load bearing [Introduction / §II.B]
    "Building on a recent recommender-system framework for social robots [13], we argue that what stands between that framework and these applications is not the model but the data it can learn from. We therefore propose four data principles..."

    The target architecture (user profiling, ranking, responsible computing) is taken from the authors' own prior paper [13] (overlapping authors Huang, Doğan, Gunes). This is scaffolding rather than a circular reduction: the four data principles themselves are motivated by the independent challenges of §I and the external studies of Table I, not by any equation or uniqueness claim internal to [13]. The self-citation is therefore minor and non-load-bearing for the paper's central contribution.

full rationale

This is a conceptual Late Breaking Report with no equations, fitted parameters, uniqueness theorems, or empirical predictions that could reduce to their inputs by construction. The four challenges (non-stationary state, weak per-action signals, unlinkable histories, observational bias) are diagnosed from independent external studies catalogued in Table I and from well-known RS problems (implicit feedback, cold-start, off-policy bias). The four principles are proposed as direct responses to those challenges and are mapped onto the framework components in Table II; they do not redefine or restate the framework itself. The only self-citation of note is the load-bearing reference to the authors' prior HRI 2026 framework [13] as the target architecture to be instantiated. That citation is normal prior-work scaffolding rather than a circular reduction: the paper's novel content (data principles and per-capability minimal requirements) stands independently of any unverified claim internal to [13]. No ansatz is smuggled, no known empirical pattern is merely renamed, and no fitted quantity is re-labeled a prediction. Score 1 reflects the single non-load-bearing self-citation; the derivation chain is otherwise self-contained against external literature.

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

The paper is a conceptual requirements analysis. It introduces no fitted numerical parameters. Its load-bearing premises are domain assumptions about clinical pediatric settings and standard recommender-system facts; the four principles themselves are the invented organizing entities.

assumptions (4)
  • domain assumption A child’s anxiety, attention and understanding are multi-dimensional and non-stationary within and across visits, so no fixed single-construct description suffices.
    Stated as Challenge (i) in §I and used to motivate Principle 1; drawn from cited HRI longitudinal work but treated as given for the clinical setting.
  • domain assumption Clinical care cannot be randomized over a large action space, so any fine-grained preference data will be observational and selection-biased.
    Challenge (iv) and Principle 4; standard confounding-by-indication argument applied to robot action selection.
  • domain assumption Consent and anonymization routinely break the stable per-child identity needed to link visits, and most children are seen only once.
    Challenge (iii); used to justify the collaborative-filtering route in Principle 3.
  • domain assumption Standard recommender-system techniques (collaborative filtering, sequential models, inverse-propensity weighting) remain applicable once the four data principles are satisfied.
    Implicit throughout §IV–V; the transfer from e-commerce/media RS to sparse clinical HRI is assumed rather than demonstrated.
invented entities (4)
  • Integrated multi-layer child profile (Principle 1)
    purpose: Jointly records identity, dynamic emotional state, interaction history, situational context and multimodal signals so personalization can condition on combinations rather than isolated attributes.
    New organizing construct introduced by the paper; no independent empirical validation outside this work.
  • Effectiveness signals aligned to robot actions (Principle 2)
    purpose: Pairs each content-identified robot action with subsequent explicit and implicit child reactions so per-action preference can be inferred.
    Proposed data primitive; independent evidence would require a deployed logging pipeline that is not supplied.
  • Linkable, representative coverage (Principle 3)
    purpose: Stable extensible identity plus multi-site demographic span enabling collaborative transfer for sparse users.
    Construct that reconciles clinical sparsity with RS collaborative filtering; not independently measured here.
  • Exposure record logged at collection time (Principle 4)
    purpose: Captures the policy and features that caused each action to be shown, enabling later debiasing.
    Direct import of RS off-policy logging into clinical HRI; no dataset demonstrating it is provided.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Toward Personalized Social Robots for Child Well-being: Data Requirement Principles from a Recommender-System Perspective." pith.science (2026). https://pith.science/paper/S77U3R2B

@misc{pith2026260705110,
  author       = {Pith},
  title        = {Pith review of: Toward Personalized Social Robots for Child Well-being: Data Requirement Principles from a Recommender-System Perspective},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/S77U3R2B}},
  note         = {Machine review of arXiv:2607.05110}
}
read the original abstract

Social robots are increasingly deployed in clinical settings to support the well-being of children, where effective support must be personalized to each child. Personalization, choosing the robot action best suited to each child, can be framed as a recommendation problem, and a recently proposed recommender-system framework for social robots offers a principled approach through user profiling, ranking, and responsible computing. Instantiating it, however, is blocked not by the model but by the data, which is hard to gather. A child's state shifts within and across visits, so no fixed description of the user holds. Within a session, the few signals of whether the robot's actions helped are weak and indirect. Across sessions, children are rarely seen more than once, and anonymization breaks the identity needed to link visits. Because care cannot be randomized, existing data is observational, biased toward whatever was already done. Each is a familiar recommender-system problem, and we propose four data principles in response: an integrated profile, effectiveness signals, linkable coverage, and an exposure record logged at collection time. We identify which of these principles each capability requires, and frame them as concrete guidelines for data collection.

Figures

Figures reproduced from arXiv: 2607.05110 by the authors.

Figure 1
Figure 1. Existing vs. required data collection for personalized social robots in child well-being settings. Left: existing robot-administered assessments [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

28 extracted references

  1. [1]

    Haru: Hard- ware design of an experimental tabletop robot assistant,

    R. Gomez, D. Szapiro, K. Galindo, and K. Nakamura, “Haru: Hard- ware design of an experimental tabletop robot assistant,” inProc. ACM/IEEE Int. Conf. Hum.-Robot Interact. (HRI), 2018, pp. 233–240

    2018

  2. [2]

    Using the social robot NAO for emotional support to children at a pediatric emergency department: Randomized clinical trial,

    S. Rossi, S. J. Santini, D. Di Genova, G. Maggi, A. Verrotti, G. Farello, R. Romualdi, A. Alisi, A. E. Tozzi, and C. Balsano, “Using the social robot NAO for emotional support to children at a pediatric emergency department: Randomized clinical trial,”J. Med. Internet Res., vol. 24, no. 1, p. e29656, 2022

    2022

  3. [3]

    Exploring the use of social robots to prepare children for radiological procedures: A focus group study,

    M. Nigro, A. Righini, and M. Spitale, “Exploring the use of social robots to prepare children for radiological procedures: A focus group study,” inProc. IEEE Int. Conf. Robot Hum. Interact. Commun. (RO- MAN), 2025

    2025

  4. [4]

    Using a robot to personalise health education for children with diabetes type 1: A pilot study,

    O. A. Blanson Henkemans, B. P. B. Bierman, J. Janssen, M. A. Neerincx, R. Looije, H. van der Bosch, and J. A. M. van der Giessen, “Using a robot to personalise health education for children with diabetes type 1: A pilot study,”Patient Educ. Couns., vol. 92, no. 2, pp. 174–181, 2013

    2013

  5. [5]

    Mitigating Children’s pain and anxiety during blood draw using social robots,

    M. H. J. Smakman, K. Smit, L. Buser, T. Monshouwer, N. van Putten, T. Trip, C. Schoof, D. F. Preciado, E. A. Konijn, E. M. van der Roest, and W. M. Tiel Groenestege, “Mitigating Children’s pain and anxiety during blood draw using social robots,”Electronics, vol. 10, no. 10, p. 1211, 2021

    2021

  6. [6]

    The effect of a social robot on Children’s pain and anxiety during blood draw,

    M. De Haas, K. Smit, D. F. Preciado Vanegas, E. van der Roest, M. Smakman, and W. Tiel Groenestege, “The effect of a social robot on Children’s pain and anxiety during blood draw,” inProc. ACM Interact. Des. Child. Conf. (IDC), 2024, pp. 776–780

    2024

  7. [7]

    Au- tonomous robot-administered anxiety screening in pediatric oncology,

    E. Nichols, N. Pérez Higueras, M. Orozco, G. Pérez, G. Álvarez- Benito, J. G. Amores-Carredano, L. Merino, and R. Gomez, “Au- tonomous robot-administered anxiety screening in pediatric oncology,” inProc. IEEE Int. Conf. Robot Hum. Interact. Commun. (RO-MAN), 2026, to appear

    2026

  8. [8]

    Empathic robots for long-term interaction,

    I. Leite, G. Castellano, A. Pereira, C. Martinho, and A. Paiva, “Empathic robots for long-term interaction,”Int. J. Soc. Robot., vol. 6, no. 3, pp. 329–341, 2014

    2014

Show all 28 references
  1. [9]

    Detecting user engagement with a robot companion using task and social interaction-based features,

    G. Castellano, A. Pereira, I. Leite, A. Paiva, and P. W. McOwan, “Detecting user engagement with a robot companion using task and social interaction-based features,” inProc. Int. Conf. Multimodal Interfaces (ICMI-MLMI), 2009, pp. 119–126

    2009

  2. [10]

    Socially assistive robots as mental health interventions for children: A scoping review,

    K. Kabaci ´nska, T. J. Prescott, and J. M. Robillard, “Socially assistive robots as mental health interventions for children: A scoping review,” Int. J. Soc. Robot., vol. 13, no. 5, pp. 919–935, 2020

    2020

  3. [11]

    Comparing the benefits of pseudonymi- sation and anonymisation under the GDPR,

    M. Hintze and K. El Emam, “Comparing the benefits of pseudonymi- sation and anonymisation under the GDPR,”J. Data Prot. Priv., vol. 2, no. 2, pp. 145–158, 2018

    2018

  4. [12]

    Confounding by indication in clinical research,

    D. N. Kyriacou and R. J. Lewis, “Confounding by indication in clinical research,”JAMA, vol. 316, no. 17, pp. 1818–1819, 2016

    2016

  5. [13]

    Reimagining social robots as recommender systems: Foundations, framework, and applications,

    J. Huang, F. I. Do ˘gan, and H. Gunes, “Reimagining social robots as recommender systems: Foundations, framework, and applications,” in Proc. ACM/IEEE Int. Conf. Hum.-Robot Interact. (HRI), 2026, pp. 406–416

    2026

  6. [14]

    Are robots ready for administering health status surveys?: First results from an HRI study with subjects with Parkinson’s disease,

    P. Briggs, M. Scheutz, and L. Tickle-Degnen, “Are robots ready for administering health status surveys?: First results from an HRI study with subjects with Parkinson’s disease,” inProc. ACM/IEEE Int. Conf. Hum.-Robot Interact. (HRI), 2015, pp. 327–334

    2015

  7. [15]

    A social robot for autonomous health data acquisition among hospitalized patients: An exploratory field study,

    D. van der Putte, R. Boumans, M. Neerincx, M. O. Rikkert, and M. de Mul, “A social robot for autonomous health data acquisition among hospitalized patients: An exploratory field study,” inProc. ACM/IEEE Int. Conf. Hum.-Robot Interact. (HRI), 2019, pp. 658–659

    2019

  8. [16]

    Can robots help in the evaluation of mental wellbeing in children? An empirical study,

    N. I. Abbasi, M. Spitale, J. Anderson, T. Ford, P. B. Jones, and H. Gunes, “Can robots help in the evaluation of mental wellbeing in children? An empirical study,” inProc. IEEE Int. Conf. Robot Hum. Interact. Commun. (RO-MAN), 2022, pp. 1459–1466

    2022

  9. [17]

    A cloud-based robot system for long- term interaction: Principles, implementation, lessons learned,

    F. Kaptein, B. Kiefer, A. Cully, O. Celiktutan, B. Bierman, R. Rijgersberg-Peters, J. Broekens, W. van Vught, M. van Bekkum, Y . Demiris, and M. A. Neerincx, “A cloud-based robot system for long- term interaction: Principles, implementation, lessons learned,”ACM Trans. Hum.-Ro...

    2021

  10. [18]

    A lon- gitudinal study of child wellbeing assessment via online interactions with a social robot,

    N. I. Abbasi, G. Laban, T. Ford, P. B. Jones, and H. Gunes, “A lon- gitudinal study of child wellbeing assessment via online interactions with a social robot,”ACM Trans. Hum.-Robot Interact., vol. 14, no. 3, pp. 1–35, 2025

    2025

  11. [19]

    Exploring the use of social robots for depression screening among older adults: An expanded feasibility study,

    B. S. de Araujo, M. Fantinato, M. Cachioni, M. S. Yassuda, R. C. de Melo, S. M. Peres, and P. C. K. Hung, “Exploring the use of social robots for depression screening among older adults: An expanded feasibility study,”IEEE Access, vol. 14, pp. 22 977–22 997, 2026

    2026

  12. [20]

    A GPT-reinforced social robot for patient communication: a pilot study,

    J.-W. J. R. van ’t Klooster, M. Capasso, D. van Gorssel, E. Vrolijk, G. Rettagliata, D. Gerritsen, M. Hegeman, E. Tauro, E. G. Caiani, and H. E. V onkeman, “A GPT-reinforced social robot for patient communication: a pilot study,”Front. Digit. Health, vol. 7, 2026

    2026

  13. [21]

    A social robot for explaining medical tests and procedures: An exploratory study in the wild,

    R. Boumans, R. Melis, T. Bosse, and S. Thill, “A social robot for explaining medical tests and procedures: An exploratory study in the wild,” inCompanion ACM/IEEE Int. Conf. Hum.-Robot Interact. (HRI), 2023, pp. 263–267

    2023

  14. [22]

    Robotising psychometrics: Validating wellbeing assessment tools in child-robot interactions,

    N. I. Abbasi, G. Laban, T. Ford, P. B. Jones, and H. Gunes, “Robotising psychometrics: Validating wellbeing assessment tools in child-robot interactions,” inProc. IEEE Int. Conf. Robot Hum. Interact. Commun. (RO-MAN), 2024, pp. 1651–1658

    2024

  15. [23]

    Symptom monitoring with patient-reported outcomes during routine cancer treatment: A randomized controlled trial,

    E. Basch, A. M. Deal, M. G. Kris, H. I. Scher, C. A. Hudis, P. Sabbatini, L. Rogak, A. V . Bennett, A. C. Dueck, T. M. Atkinson, J. F. Chou, D. Dulko, L. Sit, A. Barz, P. Novotny, M. Fruscione, J. A. Sloan, and D. Schrag, “Symptom monitoring with patient-reported outcomes duri...

    2016

  16. [24]

    Child-robot collaborative problem-solving and the importance of child’s voluntary interaction: A developmental perspective,

    V . Charisi, E. Gomez, G. Mier, L. Merino, and R. Gomez, “Child-robot collaborative problem-solving and the importance of child’s voluntary interaction: A developmental perspective,”Front. Robot. AI, vol. 7, p. 15, 2020

    2020

  17. [25]

    Robot-led vision language model wellbeing assessment of children,

    N. I. Abbasi, F. I. Do ˘gan, G. Laban, J. Anderson, T. Ford, P. B. Jones, and H. Gunes, “Robot-led vision language model wellbeing assessment of children,” inProc. IEEE Int. Conf. Robot Hum. Interact. Commun. (RO-MAN), 2025, pp. 59–64

    2025

  18. [26]

    Personalizing activity selection in assistive social robots from explicit and implicit user feedback,

    M. Maroto-Gómez, M. Malfaz, J. C. Castillo, Á. Castro-González, and M. Á. Salichs, “Personalizing activity selection in assistive social robots from explicit and implicit user feedback,”Int. J. Soc. Robot., vol. 17, no. 10, pp. 1999–2017, 2025

    1999

  19. [27]

    Long-term personal- ization of an in-home socially assistive robot for children with autism spectrum disorders,

    C. Clabaugh, K. Mahajan, S. Jain, R. Pakkar, D. Becerra, Z. Shi, E. Deng, R. Lee, G. Ragusa, and M. Matari ´c, “Long-term personal- ization of an in-home socially assistive robot for children with autism spectrum disorders,”Front. Robot. AI, vol. 6, p. 110, 2019

    2019

  20. [28]

    Recommendations as treatments: Debiasing learning and evaluation,

    T. Schnabel, A. Swaminathan, A. Singh, N. Chandak, and T. Joachims, “Recommendations as treatments: Debiasing learning and evaluation,” inProc. Int. Conf. Mach. Learn. (ICML), ser. Proc. Mach. Learn. Res. (PMLR), vol. 48, 2016, pp. 1670–1679

    2016

Pith tools

Reviewed July 11, 2026 · model on record in the stance chip above.