Pith. sign in

REVIEW 4 major objections 4 minor 43 references

Latent Behavior Diffusion for Sequential Reaction Generation in Dyadic Setting

T0 review · 4 major / 4 minor · reviewed 2026-08-15 · deepseek-v4-flash

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

Pith's one-line read A latent diffusion model generates more diverse, synchronized listener reactions in dyadic conversations, beating prior methods on the REACT2024 benchmark.

desk verdict Incremental but competently built latent diffusion for fMARG; the headline 'SOTA appropriateness' is contradicted by its own FRDist column, but the diversity/synchrony gains look real. read the letter →

arxiv 2505.07901 v1 pith:EX5VANZL submitted 2025-05-12 cs.LG cs.AI

Minh-Duc Nguyen , Hyung-Jeong Yang , Soo-Hyung Kim , Ji-Eun Shin , Seung-Won Kim This is my paper · ORCID
classification cs.LGcs.AI
keywords latentdiffusiondyadicinteractionfacialreactiongenerationmultipleappropriatereactionsnon-autoregressiveactionunitsREACT2024diversityandsynchrony
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 tries to establish that a two-stage latent diffusion model—a context-aware autoencoder that compresses listener reaction sequences into a compact latent code, plus a diffusion generator that samples in that latent space—can produce multiple appropriate facial reactions to a speaker's behavior. The authors argue that existing deterministic and Gaussian-prior models either collapse to a single reaction or suffer posterior collapse, while their latent diffusion with a Transformer-based autoencoder prior captures the one-to-many mapping between speaker behavior and plausible listener reactions. If true, the reported REACT2024 results (best FRCorr, FRDiv, FRVar, FRDvs, FRSyn, and FID among compared methods) would mean that latent-space diffusion is a practical recipe for generating diverse, synchronized, and contextually appropriate non-verbal reactions.

What carries the argument

The central mechanism is the Latent Behavior Diffusion Model, made of two stages: a context-aware autoencoder trained with VQ-VAE-style losses to map fixed-length listener reaction sequences into low-dimensional latent codes, and a conditional latent diffusion generator whose denoising network is a residual MLP conditioned on a semantic code extracted from the speaker behavior by the same encoder. The generator learns to denoise in latent space with a composite loss combining MAE in latent space and MSE in reconstructed space, and is sampled non-autoregressively using a fourth-order PLMS (Pseudo Linear Multi-Step) sampler.

What would settle it

Re-running Trans-VAE, BeLFusion, Dam et al., and Liu et al. under exactly the same splits, metric code, and compute budget and finding that one of them matches or beats the reported FRDiv, FRDvs, FRSyn, or FID values would directly falsify the claim of state-of-the-art performance.

Watch

Extended reading notes

Core claim

The paper claims its Latent Behavior Diffusion Model outperforms recent methods in appropriateness, diversity, and synchrony of generated facial reactions on REACT2024, with reported values of FRCorr 0.37, FRDist 89.40, FRDiv 0.1211, FRVar 0.0653, FRDvs 0.1505, FRSyn 43.48, and FID 50.95. The key discovery is that running the diffusion process in the low-dimensional latent space of a context-aware autoencoder avoids the posterior collapse seen in Gaussian-prior VAE approaches and the diversity limits of one-dimensional discrete codebooks, while the non-autoregressive reverse sampler (fourth-order PLMS with 50 steps) keeps the generated reactions both diverse and synchronized with the speaker.

Load-bearing premise

The claimed gains rest on the comparisons being fair: identical REACT2024 train and test splits, identical implementations of the metrics, and comparable training budgets across every method, none of which the paper documents.

Editorial extensions

If this is right

  • The two-stage latent diffusion recipe can be adapted to other one-to-many sequence generation tasks, such as gesture or body-motion synthesis, where diverse plausible outputs are required from the same context.
  • The improved FRVar and FRDiv indicate the model produces a wider spread of distinct reactions from the same speaker input, directly addressing the one-to-many mapping problem.
  • The lower FRSyn suggests the generated reactions stay better synchronized with the speaker's behavior across time, which matters for natural human-like interaction.
  • The lower FID relative to ground-truth rendering implies the predicted 3DMM parameters, when rendered to images, look more realistic than the ground-truth renderings used in the comparison.
  • The ablation showing 50 denoising steps as the best balance implies the method is computationally practical and does not need very long sampling chains.

Reading between the lines

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

  • An implication the authors leave implicit is that their results, if the comparison is fair, weaken the case for discrete tokenization in this task: continuous latent diffusion can match or beat finite-scalar and discrete-latent baselines on diversity and synchrony.
  • A testable extension is whether the model generalizes to unseen listeners without fine-tuning; the paper does not measure identity preservation beyond image-level FID on rendered faces.
  • The two-stage design implies that the autoencoder's reconstruction fidelity bounds the final reaction quality; one could probe this by varying the latent compression ratio and tracking FRDist and FRSyn.
  • If the claimed performance holds, the 50-step PLMS sampler is short enough to suggest near-real-time interactive agents that generate varied listener feedback in video calls or social robots, though runtime is not reported.
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 / 4 minor

Summary. The paper proposes a two-stage generative model for dyadic facial reaction generation (fMARG). A context-aware autoencoder compresses listener reaction sequences into a latent space, and a conditional latent diffusion model with a PLMS sampler generates multiple appropriate reactions from speaker behavior. The method is evaluated on the REACT2024 dataset against four baselines (Trans-VAE, BeLFusion, Dam et al., Liu et al.) using six reaction-level metrics plus FID. The authors report the best results on FRCorr, FRDiv, FRVar, FRDvs, and FRSyn, and on FID, but their FRDist value (89.40) is worse than Dam et al. (84.94) and Liu et al. (88.32). The paper claims in the abstract and in Section 3 that the method 'significantly outperforms recent methods' and achieves state-of-the-art appropriateness, diversity, and synchrony.

Significance. If the empirical claims were fully supported, the work would be a useful contribution to the fMARG literature: it demonstrates that a non-autoregressive latent diffusion approach with a transformer-based autoencoder prior and PLMS sampling can produce diverse and synchronized reactions, and it includes ablations over denoising steps, sampler types, and PLMS convergence orders. The authors are also to be credited for evaluating against two challenge baselines and for reporting FID in addition to the official metrics. However, the headline claim of state-of-the-art appropriateness is not supported by the reported numbers, and the absence of error bars, statistical tests, a precise evaluation protocol, and code makes the significance claim difficult to assess reproducibly. The central idea is plausible, but the paper currently overstates what its own Table 1 shows.

major comments (4)
  1. [Abstract, Section 3, and Table 1] The claim that the method achieves state-of-the-art performance in appropriateness is internally contradicted by the paper's own results. FRDist is a lower-is-better appropriateness metric, and Table 1 reports Ours = 89.40, Dam et al. = 84.94, and Liu et al. = 88.32. Thus on one of the two appropriateness metrics the proposed model is third among the four compared methods, and the abstract's 'significantly outperforms recent methods' is not supported. The text in Section 4.3 acknowledges a 'trade-off' and calls the gap 'within a fair margin,' but without error bars or statistical tests there is no basis for calling the margin fair. Please either provide a corrected or extended comparison (e.g., rerunning baselines under identical conditions with multiple seeds) or substantially soften the appropriateness claim.
  2. [Section 4.1 and Section 4.3, Tables 1-3] No error bars, confidence intervals, statistical significance tests, or number of random seeds are reported anywhere, so the word 'significantly' in the abstract is not justified by the evidence. This matters particularly for the FRDist comparison, where the reported gap between Ours and Dam et al. is large (89.40 vs. 84.94), and for the ablation table, where several configurations differ by only 0.001-0.002 in FRVar or FRDiv. Please report mean +/- standard deviation over at least three runs and apply a paired test (e.g., Wilcoxon or paired bootstrap) for the headline comparisons.
  3. [Section 4.1, Comparison Methods] The comparison omits the methods in references [27] and [28], which are highly relevant prior fMARG models, and [28] is from the same group as the current paper. Since the paper's contribution is framed as superiority over 'recent methods' and both references are cited, the omission needs explicit justification. If those methods are excluded for a specific reason (e.g., different input features or evaluation protocol), state the reason; otherwise include them in Table 1 or clearly position the current work as an incremental study relative to [28].
  4. [Section 3.2, Eq. (4)] The description of the autoencoder is ambiguous about whether the latent space is discrete or continuous. The text mentions a Vector Quantized technique and a codebook, then states the model avoids a standard Gaussian or a 1-D codebook, and later the diffusion operates on a continuous latent. Please clarify the exact training objective for the codebook (including the codebook size and commitment coefficient) and specify how the discrete and continuous branches interact, since this affects the reproducibility of the whole method.
minor comments (4)
  1. [Table 1 caption] There is a typo in the caption: 'Comparision' should be 'Comparison', and the table is preceded by 'T able 1'.
  2. [References] References [25] and [26] do not include publication venues or years; please complete the bibliographic information.
  3. [Section 3.1, Eq. (2)] The notation in Eq. (2) mixes a set definition with inequality symbols; please rewrite it to make clear that the M reactions are distinct sampled outputs rather than a set of pairwise-inequivalent elements.
  4. [Section 3.3, Eq. (7)] The loss notation uses a transposed summation symbol 'TX' and tildes in inconsistent places; please clean up the mathematical typesetting for readability.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the model's training objectives and evaluation metrics are independent, and no fitted parameter is relabeled as a prediction.

full rationale

The paper's derivation chain is empirically self-contained. The proposed Latent Behavior Diffusion model is trained in two stages: a context-aware autoencoder is trained with reconstruction, codebook, and commitment losses (Eq. 4), and the latent diffusion generator is trained with an MAE loss in latent space plus an MSE reconstruction loss (Eq. 7). Neither objective contains the evaluation metrics that support the central claim; appropriateness (FRCorr, FRDist), diversity (FRVar, FRDiv, FRDvs), synchrony (FRSyn), and image quality (FID) are all computed externally on generated outputs. No fitted parameter is renamed as a prediction, and no metric is optimized directly so as to force the reported result by construction. The comparison in Table 1 is a direct empirical comparison against prior methods, so the stated superiority claim rests on reported numbers rather than on a self-referential argument. The paper includes self-citations (refs 27 and 28) and omits those same-group methods from Table 1, but these references are used only in a literature-review sentence listing works that generate facial responses; they are not invoked as evidence for the model's performance. The internal tension that the model is best on FRCorr but third among five methods on FRDist is a correctness or presentation issue about whether 'state-of-the-art in appropriateness' is fully supported, not a circularity issue. Therefore, no specific circular step can be identified, and the circularity score is 0.

Assumptions & free parameters 5 free parameters · 4 assumptions · 0 invented entities

The paper's contribution is an empirical architecture; no new physical entities or theoretical constants are postulated. The assumptions above are typical for behavior-generation benchmarks but are not independently verified, and the most important hyperparameters (codebook size, window, denoise steps, PLMS order) are chosen by hand without sensitivity beyond the reported ablation.

free parameters (5)
  • VQ codebook size / latent dimension v
    Not reported in the paper; capacity of the autoencoder and diversity of latent codes depend on these choices, which are set implicitly and never ablated.
  • Window size w = 50 frames
    Set in Section 4.1; determines temporal context and directly influences all metrics, but no sensitivity analysis is provided.
  • Denoising steps T = 50
    Selected after ablation in Section 4.4 as the best balance; affects the appropriateness/diversity trade-off.
  • PLMS convergence order k = 4
    Chosen from ablation in Section 4.4; the effect is minimal but the final result depends on this choice.
  • Training hyperparameters = lr 1e-3, wd 5e-4, batch 32, epochs 1000/200/200
    Hand-selected in Section 4.1; not derived and affects the SOTA comparison.
assumptions (4)
  • domain assumption REACT2024 annotations (15 AUs, valence/arousal, 8 expressions, 3DMM) are accurate and sufficient to represent speaker behavior and listener reactions.
    Stated in Section 4.1; the model uses only these annotations, so errors or insufficiency in them would limit the task.
  • domain assumption The evaluation metrics from Song et al. (FRDist, FRCorr, FRVar, FRDiv, FRDvs, FRSyn) validly measure appropriateness, diversity, and synchrony.
    Section 4.2 adopts these metrics without independent validation; the central claim rests on their validity.
  • domain assumption Fixed 50-frame windows and 30-second clips preserve the temporal context needed for reaction generation.
    The model defines Y and C over windows in Section 3.1; longer-range dependencies beyond the window are ignored.
  • domain assumption PIRender-generated images from 3DMM parameters are realistic enough that FID is a meaningful quality metric.
    Used in Table 2 to compare image-level realism; depends on PIRender's rendering quality.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Latent Behavior Diffusion for Sequential Reaction Generation in Dyadic Setting." pith.science (2026). https://pith.science/paper/EX5VANZL

@misc{pith2026250507901,
  author       = {Pith},
  title        = {Pith review of: Latent Behavior Diffusion for Sequential Reaction Generation in Dyadic Setting},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/EX5VANZL}},
  note         = {Machine review of arXiv:2505.07901}
}
read the original abstract

The dyadic reaction generation task involves synthesizing responsive facial reactions that align closely with the behaviors of a conversational partner, enhancing the naturalness and effectiveness of human-like interaction simulations. This paper introduces a novel approach, the Latent Behavior Diffusion Model, comprising a context-aware autoencoder and a diffusion-based conditional generator that addresses the challenge of generating diverse and contextually relevant facial reactions from input speaker behaviors. The autoencoder compresses high-dimensional input features, capturing dynamic patterns in listener reactions while condensing complex input data into a concise latent representation, facilitating more expressive and contextually appropriate reaction synthesis. The diffusion-based conditional generator operates on the latent space generated by the autoencoder to predict realistic facial reactions in a non-autoregressive manner. This approach allows for generating diverse facial reactions that reflect subtle variations in conversational cues and emotional states. Experimental results demonstrate the effectiveness of our approach in achieving superior performance in dyadic reaction synthesis tasks compared to existing methods.

Figures

Figures reproduced from arXiv: 2505.07901 by the authors.

Figure 1
Figure 1. Overview of our proposed Latent Diffusion Model (LDM) for generating multi￾ple reactions. During the training phase, the autoencoder (AE) is first trained to encode the time series of listener reactions through a reconstruction task. Concurrently, the LDM is trained to predict future targets based on speaker behaviors C. During the sampling phase, the latent representation of the time series is first generated by th… view at source ↗
Figure 2
Figure 2. Conditional Behavior Decoder architecture in Latent space. semantically meaningful zsem. Here, our conditional behavior decoder takes the high-level semantic subcode zsem and the low-level stochastic subcode zT . In sampling process, our approach by reversing the generative process of Pseudo Linear Multi-step (PLMS) [30] to infer zT . Diffusion Models. Diffusion Models [31] are probabilistic generative models design… view at source ↗
Figure 3
Figure 3. The qualitative results of the generated facial frames of multiple listeners. Our comparison between other baselines by 3D rendering translation. The bottom of the figure shows our model-generated variants of reaction that are expressed from Speaker ground truth. The time t in second. that PIRender [43] generated from 3DMM parameters by Df , results are shown in [PITH_FULL_IMAGE:figures/full_fig_p011_3.png] view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: Generated listener facial reactions from different speakers [PITH_FULL_IMAGE:figures/full_fig_p012_4.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

43 extracted references · 27 canonical work pages

  1. [27]

    Multiple facial reaction generation using gaussian mixture of models and multimodal bottleneck transformer

    Dang-Khanh Nguyen, Prabesh Paudel, Seung-Won Kim, Ji-Eun Shin, Soo-Hyung Kim, and Hyung-Jeong Yang. Multiple facial reaction generation using gaussian mixture of models and multimodal bottleneck transformer. In2024 IEEE 18th International Conference on Automatic Face and Gesture Recognition (FG), pages 1–5. IEEE, 2024

    work page 2024

  2. [28]

    Vector quantized diffusion models for multiple ap- propriate reactions generation

    Minh-Duc Nguyen, Hyung-Jeong Yang, Ngoc-Huynh Ho, Soo-Hyung Kim, Seung- won Kim, and Ji-Eun Shin. Vector quantized diffusion models for multiple ap- propriate reactions generation. In2024 IEEE 18th International Conference on Automatic Face and Gesture Recognition (FG), pages 1–5. IEEE, 2024

    work page 2024

  3. [1]

    the MIT Press, 1974

    Albert Mehrabian and James A Russell.An approach to environmental psychology. the MIT Press, 1974

    work page 1974

  4. [2]

    The sor (stimulus-organism- response) paradigm in online learning: an empirical study of students’ knowledge hiding perceptions

    Xuesong Zhai, Minjuan Wang, and Usman Ghani. The sor (stimulus-organism- response) paradigm in online learning: an empirical study of students’ knowledge hiding perceptions. InCross Reality (XR) and Immersive Learning Environments (ILEs) in Education, pages 48–63. Routledge, 2023

    work page 2023

  5. [3]

    The affec- tive facial recognition task: The influence of cognitive styles and exposure times

    Shuna Peng, Yang Dong, Weisha Wang, Jieyi Hu, and Weiyang Dong. The affec- tive facial recognition task: The influence of cognitive styles and exposure times. Journal of Visual Communication and Image Representation, 65:102674, 2019

    work page 2019

  6. [4]

    Comparison of spatio-temporal models for human motion and pose forecasting in face-to-face interaction scenarios supplementary material

    German Barquero, Johnny Núnez, Zhen Xu, Sergio Escalera, Wei-Wei Tu, Isabelle Guyon, and Cristina Palmero. Comparison of spatio-temporal models for human motion and pose forecasting in face-to-face interaction scenarios supplementary material. 2022

    work page 2022

  7. [5]

    Dyadgan: Generating facial expressions in dyadic interactions

    Yuchi Huang and Saad M Khan. Dyadgan: Generating facial expressions in dyadic interactions. InProceedings of the IEEE Conference on Computer Vision and Pattern Recognition Workshops, pages 11–18, 2017

    work page 2017

  8. [6]

    Chalearn lap challenges on self-reported personality recognition and non-verbal behavior forecasting during social dyadic interactions: Dataset, design, and results

    Cristina Palmero, German Barquero, Julio CS Jacques Junior, Albert Clapés, Johnny Núnez, David Curto, Sorina Smeureanu, Javier Selva, Zejian Zhang, David Saeteros, et al. Chalearn lap challenges on self-reported personality recognition and non-verbal behavior forecasting during social dyadic interactions: Dataset, design, and results. InUnderstanding Soci...

    work page 2022

Show all 43 references
  1. [7]

    Personality recognition by modelling person-specific cognitive pro- cesses using graph representation

    Zilong Shao, Siyang Song, Shashank Jaiswal, Linlin Shen, Michel Valstar, and Hatice Gunes. Personality recognition by modelling person-specific cognitive pro- cesses using graph representation. Inproceedings of the 29th ACM international conference on multimedia, pages 357–366...

    2021

  2. [8]

    Learning person-specific cognition from facial reactions for automatic personality recognition.IEEE Transactions on Affective Computing, 14(4):3048– 3065, 2022

    Siyang Song, Zilong Shao, Shashank Jaiswal, Linlin Shen, Michel Valstar, and Hat- ice Gunes. Learning person-specific cognition from facial reactions for automatic personality recognition.IEEE Transactions on Affective Computing, 14(4):3048– 3065, 2022

    2022

  3. [9]

    Re- sponsive listening head generation: a benchmark dataset and baseline

    Mohan Zhou, Yalong Bai, Wei Zhang, Ting Yao, Tiejun Zhao, and Tao Mei. Re- sponsive listening head generation: a benchmark dataset and baseline. InEuropean Conference on Computer Vision, pages 124–142. Springer, 2022

    2022

  4. [10]

    Learning to listen: Modeling non-deterministic dyadic facial motion

    Evonne Ng, Hanbyul Joo, Liwen Hu, Hao Li, Trevor Darrell, Angjoo Kanazawa, and Shiry Ginosar. Learning to listen: Modeling non-deterministic dyadic facial motion. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 20395–20405, 2022

    2022

  5. [11]

    Multiple appropriate facial reaction generation in dyadic interaction settings: What, why and how?arXiv preprint arXiv:2302.06514, 2023

    Siyang Song, Micol Spitale, Yiming Luo, Batuhan Bal, and Hatice Gunes. Multiple appropriate facial reaction generation in dyadic interaction settings: What, why and how?arXiv preprint arXiv:2302.06514, 2023

    2023 arXiv

  6. [12]

    High-resolution image synthesis with latent diffusion models

    Robin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser, and Björn Ommer. High-resolution image synthesis with latent diffusion models. InPro- ceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 10684–10695, 2022

    2022

  7. [13]

    React 2024: the second multiple appropriate facial reaction generation challenge

    Siyang Song, Micol Spitale, Cheng Luo, Cristina Palmero, German Barquero, Hengde Zhu, Sergio Escalera, Michel Valstar, Tobias Baur, Fabien Ringeval, et al. React 2024: the second multiple appropriate facial reaction generation challenge. arXiv preprint arXiv:2401.05166, 2024

    2024 arXiv

  8. [14]

    Responsive listening behavior.Computer animation and virtual worlds, 19(5):579–589, 2008

    Marco Gillies, Xueni Pan, Mel Slater, and John Shawe-Taylor. Responsive listening behavior.Computer animation and virtual worlds, 19(5):579–589, 2008

    2008

  9. [15]

    To react or not to react: End-to-end visual pose forecasting for personalized avatar during dyadic conversations

    Chaitanya Ahuja, Shugao Ma, Louis-Philippe Morency, and Yaser Sheikh. To react or not to react: End-to-end visual pose forecasting for personalized avatar during dyadic conversations. In2019 International conference on multimodal interaction, pages 74–84, 2019

    2019

  10. [16]

    Predicting head pose in dyadic conversation

    David Greenwood, Stephen Laycock, and Iain Matthews. Predicting head pose in dyadic conversation. InIntelligent Virtual Agents: 17th International Conference, IVA 2017, Stockholm, Sweden, August 27-30, 2017, Proceedings 17, pages 160–169. Springer, 2017

    2017

  11. [17]

    Affective faces for goal-driven dyadic communication.arXiv preprint arXiv:2301.10939, 2023

    Scott Geng, Revant Teotia, Purva Tendulkar, Sachit Menon, and Carl Von- drick. Affective faces for goal-driven dyadic communication.arXiv preprint arXiv:2301.10939, 2023

    2023 arXiv

  12. [18]

    Generative adversarial net- works.Communications of the ACM, 63(11):139–144, 2020

    Ian Goodfellow, Jean Pouget-Abadie, Mehdi Mirza, Bing Xu, David Warde-Farley, Sherjil Ozair, Aaron Courville, and Yoshua Bengio. Generative adversarial net- works.Communications of the ACM, 63(11):139–144, 2020

    2020

  13. [19]

    Context-aware human behaviour forecasting in dyadic interactions

    Nguyen Tan Viet Tuyen and Oya Celiktutan. Context-aware human behaviour forecasting in dyadic interactions. InUnderstanding Social Behavior in Dyadic and Small Group Interactions, pages 88–106. PMLR, 2022

    2022

  14. [20]

    Long short-term memory.Supervised sequence labelling with recurrent neural networks, pages 37–45, 2012

    Alex Graves and Alex Graves. Long short-term memory.Supervised sequence labelling with recurrent neural networks, pages 37–45, 2012

    2012

  15. [21]

    A morphable model for the synthesis of 3d faces

    Volker Blanz and Thomas Vetter. A morphable model for the synthesis of 3d faces. InSeminal Graphics Papers: Pushing the Boundaries, Volume 2, pages 157–164. 2023

    2023

  16. [22]

    Neural discrete representation learning

    Aaron Van Den Oord, Oriol Vinyals, et al. Neural discrete representation learning. Advances in neural information processing systems, 30, 2017

    2017

  17. [23]

    Emo- tional listener portrait: Neural listener head generation with emotion

    Luchuan Song, Guojun Yin, Zhenchao Jin, Xiaoyi Dong, and Chenliang Xu. Emo- tional listener portrait: Neural listener head generation with emotion. InPro- Latent Behavior Diffusion for Sequential Reaction Generation 15 ceedings of the IEEE/CVF International Conference on Compu...

    2023

  18. [24]

    React2023: the first multi-modal multiple appropriate facial reaction generation challenge.arXiv preprint arXiv:2306.06583, 2023

    Siyang Song, Micol Spitale, Cheng Luo, German Barquero, Cristina Palmero, Ser- gio Escalera, Michel Valstar, Tobias Baur, Fabien Ringeval, Elisabeth Andre, et al. React2023: the first multi-modal multiple appropriate facial reaction generation challenge.arXiv preprint arXiv:23...

    2023 arXiv

  19. [25]

    Finite scalar quantization as facial tokenizer for dyadic reaction generation

    Quang Tien Dam, Tri Tung Nguyen Nguyen, Dinh Tuan Tran, and Joo-Ho Lee. Finite scalar quantization as facial tokenizer for dyadic reaction generation

  20. [26]

    One-to-many appropriate reaction mapping mod- eling with discrete latent variable

    Zhenjie Liu, Cong Liang, Jiahe Wang, Haofan Zhang, Yadong Liu, Caichao Zhang, Jialin Gui, and Shangfei Wang. One-to-many appropriate reaction mapping mod- eling with discrete latent variable

  21. [29]

    Diffusion autoencoders: Toward a meaningful and decodable rep- resentation

    Konpat Preechakul, Nattanat Chatthee, Suttisak Wizadwongsa, and Supasorn Suwajanakorn. Diffusion autoencoders: Toward a meaningful and decodable rep- resentation. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 10619–10629, 2022

    2022

  22. [30]

    Pseudo numerical methods for diffusion models on manifolds.arXiv preprint arXiv:2202.09778, 2022

    Luping Liu, Yi Ren, Zhijie Lin, and Zhou Zhao. Pseudo numerical methods for diffusion models on manifolds.arXiv preprint arXiv:2202.09778, 2022

    2022 arXiv

  23. [31]

    Denoising diffusion probabilistic mod- els.Advances in neural information processing systems, 33:6840–6851, 2020

    Jonathan Ho, Ajay Jain, and Pieter Abbeel. Denoising diffusion probabilistic mod- els.Advances in neural information processing systems, 33:6840–6851, 2020

    2020

  24. [32]

    The noxi database: multimodal recordings of mediated novice-expert interactions

    Angelo Cafaro, Johannes Wagner, Tobias Baur, Soumia Dermouche, Mercedes Tor- res Torres, Catherine Pelachaud, Elisabeth André, and Michel Valstar. The noxi database: multimodal recordings of mediated novice-expert interactions. InPro- ceedings of the 19th ACM International Con...

    2017

  25. [33]

    Intro- ducing the recola multimodal corpus of remote collaborative and affective inter- actions

    Fabien Ringeval, Andreas Sonderegger, Juergen Sauer, and Denis Lalanne. Intro- ducing the recola multimodal corpus of remote collaborative and affective inter- actions. In2013 10th IEEE international conference and workshops on automatic face and gesture recognition (FG), page...

    2013

  26. [34]

    Learn- ing multi-dimensional edge feature-based au relation graph for facial action unit recognition.arXiv preprint arXiv:2205.01782, 2022

    Cheng Luo, Siyang Song, Weicheng Xie, Linlin Shen, and Hatice Gunes. Learn- ing multi-dimensional edge feature-based au relation graph for facial action unit recognition.arXiv preprint arXiv:2205.01782, 2022

    2022 arXiv

  27. [35]

    Gratis: Deep learning graph representation with task-specific topology and multi-dimensional edge fea- tures.arXiv preprint arXiv:2211.12482, 2022

    Siyang Song, Yuxin Song, Cheng Luo, Zhiyuan Song, Selim Kuzucu, Xi Jia, Zhi- jiang Guo, Weicheng Xie, Linlin Shen, and Hatice Gunes. Gratis: Deep learning graph representation with task-specific topology and multi-dimensional edge fea- tures.arXiv preprint arXiv:2211.12482, 2022

    2022 arXiv

  28. [36]

    Estimation of continuous valence and arousal levels from faces in natural- istic conditions.Nature Machine Intelligence, 3(1):42–50, 2021

    Antoine Toisoul, Jean Kossaifi, Adrian Bulat, Georgios Tzimiropoulos, and Maja Pantic. Estimation of continuous valence and arousal levels from faces in natural- istic conditions.Nature Machine Intelligence, 3(1):42–50, 2021

    2021

  29. [37]

    Teach: Temporal action composition for 3d humans

    Nikos Athanasiou, Mathis Petrovich, Michael J Black, and Gül Varol. Teach: Temporal action composition for 3d humans. In2022 International Conference on 3D Vision (3DV), pages 414–423. IEEE, 2022. 16 Nguyen et al

    2022

  30. [38]

    Belfusion: Latent diffusion for behavior-driven human motion prediction

    German Barquero, Sergio Escalera, and Cristina Palmero. Belfusion: Latent diffusion for behavior-driven human motion prediction. InProceedings of the IEEE/CVF International Conference on Computer Vision, 2023

    2023

  31. [39]

    Finite scalar quantization: Vq-vae made simple.arXiv preprint arXiv:2309.15505, 2023

    Fabian Mentzer, David Minnen, Eirikur Agustsson, and Michael Tschannen. Finite scalar quantization: Vq-vae made simple.arXiv preprint arXiv:2309.15505, 2023

    2023 arXiv

  32. [40]

    Pytorch: An imperative style, high-performance deep learning library.Advances in neural information processing systems, 32, 2019

    Adam Paszke, Sam Gross, Francisco Massa, Adam Lerer, James Bradbury, Gre- gory Chanan, Trevor Killeen, Zeming Lin, Natalia Gimelshein, Luca Antiga, et al. Pytorch: An imperative style, high-performance deep learning library.Advances in neural information processing systems, 32, 2019

    2019

  33. [41]

    Attention is all you need

    Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N Gomez, Łukasz Kaiser, and Illia Polosukhin. Attention is all you need. Advances in neural information processing systems, 30, 2017

    2017

  34. [42]

    Gans trained by a two time-scale update rule converge to a local nash equilibrium.Advances in neural information processing systems, 30, 2017

    Martin Heusel, Hubert Ramsauer, Thomas Unterthiner, Bernhard Nessler, and Sepp Hochreiter. Gans trained by a two time-scale update rule converge to a local nash equilibrium.Advances in neural information processing systems, 30, 2017

    2017

  35. [43]

    Pirenderer: Control- lable portrait image generation via semantic neural rendering

    Yurui Ren, Ge Li, Yuanqi Chen, Thomas H Li, and Shan Liu. Pirenderer: Control- lable portrait image generation via semantic neural rendering. InProceedings of the IEEE/CVF international conference on computer vision, pages 13759–13768, 2021

    2021

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.