pith. sign in

arxiv: 2605.25748 · v1 · pith:FZFZOTLMnew · submitted 2026-05-25 · 💻 cs.AI

Agent-Centric Social Trajectory Prediction: A Free Energy Principle Perspective

Yanping Wu , Ji Zhang , Hao Chen , Edmond S.L. Ho , Chongfeng Wei This is my paper

Pith reviewed 2026-06-29 21:20 UTC · model grok-4.3

classification 💻 cs.AI
keywords trajectory predictionfree energy principleagent-centric modelingsocial interactiondiffusion modelsbelief inferencepartial observabilitycognitive alignment
0
0 comments X

The pith

FEP-Diff grounds trajectory prediction in the free energy principle to produce cognitively plausible forecasts from local observations alone.

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

The paper proposes FEP-Diff as an agent-centric framework that applies the Free Energy Principle to predict agent trajectories in social scenes. Current methods often assume complete global views and omit cognitive constraints, leading to failures when information is partial. The new approach uses a dual-branch encoder to pull motion and interaction signals from local data, then optimizes multimodal latent beliefs through a free-energy objective plus a social consistency term on the neighborhood graph. A residual diffusion generator conditioned on those beliefs produces the final trajectories. Tests across five benchmarks show consistent gains over prior methods precisely when observability is restricted.

Core claim

FEP-Diff is an agent-centric trajectory prediction framework grounded in the Free Energy Principle that extracts ego-motion and social cues via a dual-branch spatiotemporal encoder, infers multimodal latent belief distributions through a goal-conditioned learner optimized by a free-energy objective with social consistency constraint on the local neighborhood graph, and generates precise diverse futures via a residual diffusion trajectory generator with token-level proxy conditioning.

What carries the argument

Goal-conditioned belief learner that infers and optimizes latent beliefs via free-energy objective plus social consistency constraint on the neighborhood graph.

If this is right

  • Predictions remain accurate even when global state information is unavailable.
  • Social consistency among neighboring agents is enforced during belief inference.
  • Generated trajectories exhibit greater diversity while respecting cognitive constraints.
  • The framework supports deployment in settings with realistic partial observability.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same belief-optimization structure could be tested on raw sensor streams from moving vehicles to check whether social consistency still improves accuracy.
  • If the free-energy term is removed while keeping the encoder and generator, performance under restricted views should drop according to the paper's logic.
  • The approach suggests that neuroscience-derived belief updating may transfer to other multi-agent forecasting tasks such as team sports or swarm robotics.

Load-bearing premise

Optimizing latent beliefs with a free-energy objective and social consistency constraint on the local graph produces predictions that are both more accurate and more cognitively aligned than methods without these components.

What would settle it

A controlled test on any of the five public benchmarks where FEP-Diff fails to exceed the accuracy of prior methods when each agent receives only its local neighborhood observations.

Figures

Figures reproduced from arXiv: 2605.25748 by Chongfeng Wei, Edmond S.L. Ho, Hao Chen, Ji Zhang, Yanping Wu.

Figure 1
Figure 1. Figure 1: Comparison of trajectory prediction paradigms. (a) [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Overview of FEP-Diff, comprising three modules: an [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Qualitative comparison of stochastic (𝐾 = 20, bottom) and deterministic (𝐾 = 1, top) trajectory predictions across five benchmarks (ETH, Hotel, UNIV, ZARA1, ZARA2). For the stochastic setting, the best-of-20 sample is shown for each method. 5.2 Experimental Results 5.2.1 Stochastic Trajectory Prediction Performance [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Sensitivity analysis of loss weights on ETH. [PITH_FULL_IMAGE:figures/full_fig_p009_4.png] view at source ↗
read the original abstract

Trajectory prediction methods have demonstrated remarkable capabilities in capturing complex motion patterns. However, existing methods rely on global state assumptions, suffer from insufficient belief inference under partial observability, and lack cognitive behavioral constraints in prediction. These limitations severely compromise both deployment feasibility and physical plausibility in real-world settings. In this work, we propose FEP-Diff, an agent-centric trajectory prediction framework grounded in the Free Energy Principle, aimed at achieving cognitively plausible predictions under realistic constraints. Specifically, a dual-branch spatiotemporal encoder extracts ego-motion dynamics and social interaction cues from local observations. Building upon this, a goal-conditioned belief learner infers multimodal latent belief distributions optimized via a free-energy objective, with a social consistency constraint on the local neighborhood graph to promote cognitive alignment among neighboring agents. Finally, a residual diffusion trajectory generator is conditioned on the learned belief representations with token-level proxy conditioning, producing precise and diverse future predictions. Extensive experiments on five public benchmarks demonstrate that FEP-Diff consistently outperforms state-of-the-art methods under restricted observability. Code: https://anonymous.4open.science/r/FEP-Diff-8876.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

2 major / 2 minor

Summary. The paper proposes FEP-Diff, an agent-centric trajectory prediction model grounded in the Free Energy Principle. It consists of a dual-branch spatiotemporal encoder extracting ego-motion and social cues from local observations, a goal-conditioned belief learner whose multimodal latent beliefs are optimized via a free-energy objective together with a social consistency constraint on the local neighborhood graph, and a residual diffusion trajectory generator conditioned on the learned beliefs via token-level proxy conditioning. The central claim is that this yields cognitively plausible predictions that outperform state-of-the-art methods on five public benchmarks under restricted observability.

Significance. If the FEP grounding and performance claims hold after verification, the work would offer a principled active-inference approach to multi-agent prediction under partial observability, potentially improving robustness and cognitive alignment compared with purely data-driven baselines. The combination of variational belief optimization with diffusion-based generation is a non-trivial technical contribution that could influence both trajectory forecasting and cognitive modeling literatures.

major comments (2)
  1. [Abstract] Abstract: the social consistency constraint is described as added 'with' the free-energy objective 'to promote cognitive alignment'. If this term is an auxiliary regularizer on the neighborhood graph rather than a term arising inside the variational free-energy functional (e.g., from a joint generative model over neighboring agents), then measured gains cannot be attributed to the FEP component; this is load-bearing for the paper's central claim that FEP-Diff is 'grounded in the Free Energy Principle'.
  2. [Abstract] Abstract (and any experimental section): the claim of consistent outperformance on five benchmarks under restricted observability is stated without reference to specific metrics, baselines, error bars, ablation results, or observability protocols. Without these details the magnitude and robustness of the reported gains cannot be evaluated against the stated FEP-based mechanism.
minor comments (2)
  1. The anonymous code link should be replaced with a permanent repository upon acceptance to support reproducibility.
  2. Notation for the free-energy objective and the precise mathematical form of the social consistency constraint should be introduced with equation numbers in the methods section for clarity.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive feedback. We address each major comment below with clarifications and proposed revisions where appropriate.

read point-by-point responses

  1. Referee: [Abstract] Abstract: the social consistency constraint is described as added 'with' the free-energy objective 'to promote cognitive alignment'. If this term is an auxiliary regularizer on the neighborhood graph rather than a term arising inside the variational free-energy functional (e.g., from a joint generative model over neighboring agents), then measured gains cannot be attributed to the FEP component; this is load-bearing for the paper's central claim that FEP-Diff is 'grounded in the Free Energy Principle'.

    Authors: The referee is correct that the manuscript describes the social consistency constraint as added 'with' the free-energy objective. In the current formulation, the goal-conditioned belief learner is optimized via the free-energy objective, while the social consistency constraint functions as an auxiliary regularizer applied to the local neighborhood graph. This term is motivated by FEP principles of social cognition and alignment but is not derived as an intrinsic component of the variational free-energy functional (e.g., via a joint generative model). We acknowledge that this distinction limits the extent to which performance gains can be attributed exclusively to the FEP component. We will revise the abstract and the method section to explicitly clarify the auxiliary nature of the constraint and its relationship to the FEP grounding. revision: yes

  2. Referee: [Abstract] Abstract (and any experimental section): the claim of consistent outperformance on five benchmarks under restricted observability is stated without reference to specific metrics, baselines, error bars, ablation results, or observability protocols. Without these details the magnitude and robustness of the reported gains cannot be evaluated against the stated FEP-based mechanism.

    Authors: Abstracts are concise summaries and conventionally omit exhaustive quantitative details. The full manuscript's experimental section reports results on five benchmarks with specific metrics, baseline comparisons, error bars, ablation studies, and explicit observability protocols under restricted settings. To address the concern, we will strengthen the experimental section by adding explicit analysis linking performance gains to the FEP-based belief learner and social consistency components. We will also consider adding one or two key quantitative highlights to the abstract if space allows under the venue constraints. revision: partial

Circularity Check

0 steps flagged

No significant circularity; derivation applies external FEP without reducing to self-inputs

full rationale

The paper applies the Free Energy Principle as an external optimization objective to a belief learner and adds a social consistency constraint on neighborhood graphs. No equations or claims reduce a 'prediction' to a fitted parameter by construction, nor does any load-bearing step rely on self-citation chains or imported uniqueness theorems. The central claims rest on empirical benchmark comparisons under restricted observability, which are falsifiable outside the model's fitted values. The social consistency term is presented as an addition 'to promote cognitive alignment' rather than derived inside the variational free-energy functional, but this is a modeling choice, not a definitional loop. No self-definitional, fitted-input-renamed-as-prediction, or ansatz-smuggled patterns appear in the provided text.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Only the abstract is available; no concrete free parameters, axioms, or invented entities can be extracted or verified from the given information.

pith-pipeline@v0.9.1-grok · 5730 in / 1094 out tokens · 39985 ms · 2026-06-29T21:20:19.922853+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

59 extracted references · 6 canonical work pages · 4 internal anchors

  1. [1]

    Ijaz Ahmed, Miswar Akhtar Syed, Muhammad Maaruf, and Muhammad Khalid

  2. [2]

    Computing 107, 1 (2025), 2

    Distributed computing in multi-agent systems: a survey of decentralized machine learning approaches. Computing 107, 1 (2025), 2

    2025

  3. [3]

    Alexandre Alahi, Kratarth Goel, Vignesh Ramanathan, Alexandre Robicquet, Li Fei-Fei, and Silvio Savarese. 2016. Social lstm: Human trajectory prediction in crowded spaces. In Proceedings of the IEEE conference on computer vision and pattern recognition. 961–971

    2016

  4. [4]

    Inhwan Bae, Junoh Lee, and Hae-Gon Jeon. 2024. Can language beat numerical regression? language-based multimodal trajectory prediction. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition . 753–766

    2024

  5. [5]

    Inhwan Bae, Jean Oh, and Hae-Gon Jeon. 2023. Eigentrajectory: Low-rank de- scriptors for multi-modal trajectory forecasting. In Proceedings of the IEEE/CVF International Conference on Computer Vision . 10017–10029

    2023

  6. [6]

    Inhwan Bae, Jin-Hwi Park, and Hae-Gon Jeon. 2022. Learning pedestrian group representations for multi-modal trajectory prediction. In European Conference on Computer Vision . Springer, 270–289

    2022

  7. [7]

    Inhwan Bae, Jin-Hwi Park, and Hae-Gon Jeon. 2022. Non-probability sam- pling network for stochastic human trajectory prediction. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition . 6477–6487

    2022

  8. [8]

    Inhwan Bae, Young-Jae Park, and Hae-Gon Jeon. 2024. Singulartrajectory: Uni- versal trajectory predictor using diffusion model. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition . 17890–17901

    2024

  9. [9]

    Mohammadhossein Bahari, Saeed Saadatnejad, Amirhossein Askari Farsangi, Seyed-Mohsen Moosavi-Dezfooli, and Alexandre Alahi. 2025. Certified human trajectory prediction. In Proceedings of the Computer Vision and Pattern Recogni- tion Conference. 12301–12311

    2025

  10. [10]

    Guillem Capellera, Antonio Rubio, Luis Ferraz, and Antonio Agudo. 2025. Uni- fied uncertainty-aware diffusion for multi-agent trajectory modeling. In Proceed- ings of the Computer Vision and Pattern Recognition Conference . 22476–22486

    2025

  11. [11]

    Kai Chen, Xiaodong Zhao, Yujie Huang, Guoyu Fang, Xiao Song, Ruiping Wang, and Ziyuan Wang. 2025. Socialmoif: Multi-order intention fusion for pedestrian trajectory prediction. In Proceedings of the Computer Vision and Pattern Recogni- tion Conference. 22465–22475

    2025

  12. [12]

    Yujiao Cheng, Liting Sun, Changliu Liu, and Masayoshi Tomizuka. 2020. To- wards efficient human-robot collaboration with robust plan recognition and tra- jectory prediction. IEEE Robotics and Automation Letters 5, 2 (2020), 2602–2609

    2020

  13. [13]

    Patrick Dendorfer, Sven Elflein, and Laura Leal-Taixé. 2021. Mg-gan: A multi- generator model preventing out-of-distribution samples in pedestrian trajectory prediction. In Proceedings of the IEEE/CVF international conference on computer vision. 13158–13167

    2021

  14. [14]

    Yepeng Ding, Ahmed Twabi, Junwei Yu, Lingfeng Zhang, Tohru Kondo, and Hi- royuki Sato. 2025. Decentralized Multi-Agent System with Trust-Aware Com- munication. In 2025 IEEE International Symposium on Parallel and Distributed Processing with Applications (ISPA). IEEE, 1439–1445

    2025

  15. [15]

    Karl Friston. 2009. The free-energy principle: a rough guide to the brain? Trends in cognitive sciences 13, 7 (2009), 293–301

    2009

  16. [16]

    Karl Friston. 2010. The free-energy principle: a unified brain theory? Nature reviews neuroscience 11, 2 (2010), 127–138

    2010

  17. [17]

    Karl Friston, Thomas FitzGerald, Francesco Rigoli, Philipp Schwartenbeck, and Giovanni Pezzulo. 2017. Active inference: a process theory. Neural computation 29, 1 (2017), 1–49

    2017

  18. [18]

    Karl Friston, Michael Levin, Biswa Sengupta, and Giovanni Pezzulo. 2015. Know- ing one’s place: a free-energy approach to pattern regulation. Journal of the Royal Society Interface 12, 105 (2015)

    2015

  19. [19]

    Yuxiang Fu, Qi Yan, Lele Wang, Ke Li, and Renjie Liao. 2025. Moflow: One-step flow matching for human trajectory forecasting via implicit maximum likeli- hood estimation based distillation. In Proceedings of the Computer Vision and Pattern Recognition Conference. 17282–17293

    2025

  20. [20]

    Zheng Fu, Kun Jiang, Chuchu Xie, Yuhang Xu, Jin Huang, and Diange Yang

  21. [21]

    IEEE Transactions on Intelligent Vehicles (2024), 1–33

    Summary and Reflections on Pedestrian Trajectory Prediction in the Field of Autonomous Driving. IEEE Transactions on Intelligent Vehicles (2024), 1–33. doi:10.1109/TIV.2024.3399327

    work page doi:10.1109/tiv.2024.3399327 2024

  22. [22]

    Tianpei Gu, Guangyi Chen, Junlong Li, Chunze Lin, Yongming Rao, Jie Zhou, and Jiwen Lu. 2022. Stochastic trajectory prediction via motion indeterminacy diffusion. In Proceedings of the IEEE/CVF conference on computer vision and pat- tern recognition. 17113–17122

    2022

  23. [23]

    Agrim Gupta, Justin Johnson, Li Fei-Fei, Silvio Savarese, and Alexandre Alahi

  24. [24]

    In Proceedings of the IEEE conference on computer vision and pattern recognition

    Social gan: Socially acceptable trajectories with generative adversarial networks. In Proceedings of the IEEE conference on computer vision and pattern recognition. 2255–2264

  25. [25]

    Tiankai Hang, Shuyang Gu, Chen Li, Jianmin Bao, Dong Chen, Han Hu, Xin Geng, and Baining Guo. 2023. Efficient diffusion training via min-snr weight- ing strategy. In Proceedings of the IEEE/CVF international conference on computer vision. 7441–7451

    2023

  26. [26]

    Jonathan Ho, Ajay Jain, and Pieter Abbeel. 2020. Denoising diffusion probabilis- tic models. Advances in neural information processing systems 33 (2020), 6840– 6851

    2020

  27. [27]

    Yanjun Huang, Jiatong Du, Ziru Yang, Zewei Zhou, Lin Zhang, and Hong Chen

  28. [28]

    IEEE transactions on intelligent vehicles 7, 3 (2022), 652–674

    A survey on trajectory-prediction methods for autonomous driving. IEEE transactions on intelligent vehicles 7, 3 (2022), 652–674

    2022

  29. [29]

    Durk P Kingma, Tim Salimans, Rafal Jozefowicz, Xi Chen, Ilya Sutskever, and Max Welling. 2016. Improved variational inference with inverse autoregressive flow. Advances in neural information processing systems 29 (2016)

    2016

  30. [30]

    Diederik P Kingma and Max Welling. 2013. Auto-encoding variational bayes. arXiv preprint arXiv:1312.6114 (2013)

    work page internal anchor Pith review Pith/arXiv arXiv 2013

  31. [31]

    Alon Lerner, Yiorgos Chrysanthou, and Dani Lischinski. 2007. Crowds by exam- ple. In Computer graphics forum, Vol. 26. Wiley Online Library, 655–664

    2007

  32. [32]

    Haicheng Liao, Yongkang Li, Zhenning Li, Chengyue Wang, Zhiyong Cui, Shengbo Eben Li, and Chengzhong Xu. 2024. A Cognitive-Based Trajectory Prediction Approach for Autonomous Driving. IEEE Transactions on Intelligent Vehicles 9, 4 (2024), 4632–4643. doi:10.1109/TIV.2024.3376074

    work page doi:10.1109/tiv.2024.3376074 2024

  33. [33]

    Yiwei Lyu, Wenhao Luo, and John M Dolan. 2023. Risk-aware safe control for decentralized multi-agent systems via dynamic responsibility allocation. In 2023 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) . IEEE, 1–8

    2023

  34. [34]

    Dipankar Maity and Panagiotis Tsiotras. 2021. Multiagent consensus subject to communication and privacy constraints. IEEE Transactions on Control of Network Systems 9, 2 (2021), 943–955

    2021

  35. [35]

    Weibo Mao, Chenxin Xu, Qi Zhu, Siheng Chen, and Yanfeng Wang. 2023. Leapfrog diffusion model for stochastic trajectory prediction. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition . 5517–5526. , , Yanping Wu et al

    2023

  36. [36]

    Pietro Mazzaglia, Tim Verbelen, Ozan Catal, and Bart Dhoedt. 2022. The free energy principle for perception and action: A deep learning perspective. Entropy 24, 2 (2022), 301

    2022

  37. [37]

    Abduallah Mohamed, Kun Qian, Mohamed Elhoseiny, and Christian Claudel

  38. [38]

    In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition

    Social-stgcnn: A social spatio-temporal graph convolutional neural net- work for human trajectory prediction. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition . 14424–14432

  39. [39]

    Shayegan Omidshafiei, Jason Pazis, Christopher Amato, Jonathan P How, and John Vian. 2017. Deep decentralized multi-task multi-agent reinforcement learn- ing under partial observability. In International conference on machine learning . PMLR, 2681–2690

    2017

  40. [40]

    Stefano Pellegrini, Andreas Ess, Konrad Schindler, and Luc Van Gool. 2009. You’ll never walk alone: Modeling social behavior for multi-target tracking. In 2009 IEEE 12th international conference on computer vision . IEEE, 261–268

    2009

  41. [41]

    Giovanni Pezzulo, Thomas Parr, and Karl Friston. 2024. Active inference as a theory of sentient behavior. Biological Psychology 186 (2024), 108741

    2024

  42. [42]

    Tim Salzmann, Boris Ivanovic, Punarjay Chakravarty, and Marco Pavone. 2020. Trajectron++: Dynamically-feasible trajectory forecasting with heterogeneous data. In European conference on computer vision . Springer, 683–700

    2020

  43. [43]

    Allahkaram Shafiei, Hozefa Jesawada, Karl Friston, and Giovanni Russo. 2025. Distributionally robust free energy principle for decision-making. Nature Com- munications (2025)

    2025

  44. [44]

    Liushuai Shi, Le Wang, Chengjiang Long, Sanping Zhou, Mo Zhou, Zhenxing Niu, and Gang Hua. 2021. SGCN: Sparse graph convolution network for pedes- trian trajectory prediction. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition . 8994–9003

    2021

  45. [45]

    Apoorv Singh. 2023. Trajectory-prediction with vision: A survey. In Proceedings of the IEEE/CVF International Conference on Computer Vision . 3318–3323

    2023

  46. [46]

    Jiaming Song, Chenlin Meng, and Stefano Ermon. 2020. Denoising diffusion implicit models. arXiv preprint arXiv:2010.02502 (2020)

    work page internal anchor Pith review Pith/arXiv arXiv 2020

  47. [47]

    Pinhao Song, Pengteng Li, Erwin Aertbeliën, and Renaud Detry. 2024. Robot trajectron: Trajectory prediction-based shared control for robot manipulation. In 2024 IEEE International Conference on Robotics and Automation (ICRA) . IEEE, 5585–5591

    2024

  48. [48]

    Yang Song, Jascha Sohl-Dickstein, Diederik P Kingma, Abhishek Kumar, Stefano Ermon, and Ben Poole. 2020. Score-based generative modeling through stochas- tic differential equations. arXiv preprint arXiv:2011.13456 (2020)

    work page internal anchor Pith review Pith/arXiv arXiv 2020

  49. [49]

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

    2017

  50. [50]

    Petar Veličković, Guillem Cucurull, Arantxa Casanova, Adriana Romero, Pietro Lio, and Yoshua Bengio. 2017. Graph attention networks. arXiv preprint arXiv:1710.10903 (2017)

    work page internal anchor Pith review Pith/arXiv arXiv 2017

  51. [51]

    Chenxin Xu, Maosen Li, Zhenyang Ni, Ya Zhang, and Siheng Chen. 2022. Group- net: Multiscale hypergraph neural networks for trajectory prediction with rela- tional reasoning. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. 6498–6507

    2022

  52. [52]

    Chenxin Xu, Robby T Tan, Yuhong Tan, Siheng Chen, Yu Guang Wang, Xinchao Wang, and Yanfeng Wang. 2023. Eqmotion: Equivariant multi-agent motion pre- diction with invariant interaction reasoning. In Proceedings of the IEEE/CVF con- ference on computer vision and pattern recognition . 1410–1420

    2023

  53. [53]

    Pei Xu, Jean-Bernard Hayet, and Ioannis Karamouzas. 2022. Socialvae: Human trajectory prediction using timewise latents. In European Conference on Com- puter Vision. Springer, 511–528

    2022

  54. [54]

    Changzhi Yang, Huihui Pan, Jue Wang, and Yuanduo Hong. 2025. TrajDiff: Tra- jectory Prediction With Diffusion Probabilistic Models. IEEE Transactions on Image Processing 34 (2025), 8257–8270

    2025

  55. [55]

    Songru Yang, Zhenwei Shi, and Zhengxia Zou. 2025. Unified Multi-Agent Trajec- tory Modeling with Masked Trajectory Diffusion. In Proceedings of the IEEE/CVF International Conference on Computer Vision . 27563–27574

    2025

  56. [56]

    Ye Yuan, Xinshuo Weng, Yanglan Ou, and Kris M Kitani. 2021. Agentformer: Agent-aware transformers for socio-temporal multi-agent forecasting. In Pro- ceedings of the IEEE/CVF international conference on computer vision. 9813–9823

    2021

  57. [57]

    Yinyan Zhang, Shuai Li, and Jian Weng. 2022. Distributed k-winners-take-all network: An optimization perspective. IEEE Transactions on Cybernetics 53, 8 (2022), 5069–5081

    2022

  58. [58]

    Zhengquan Zhang and Feng Xu. 2024. An overview of the free energy principle and related research. Neural Computation 36, 5 (2024), 963–1021

    2024

  59. [59]

    Yuanshao Zhu, Yongchao Ye, Shiyao Zhang, Xiangyu Zhao, and James Yu. 2023. Difftraj: Generating gps trajectory with diffusion probabilistic model. Advances in Neural Information Processing Systems 36 (2023), 65168–65188. Received 20 February 2007; revised 12 March 2009; accepted 5 June 2009

    2023