{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,6,15]],"date-time":"2026-06-15T14:30:43Z","timestamp":1781533843683,"version":"3.54.5"},"reference-count":39,"publisher":"MDPI AG","issue":"5","license":[{"start":{"date-parts":[[2026,4,29]],"date-time":"2026-04-29T00:00:00Z","timestamp":1777420800000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"funder":[{"DOI":"10.13039\/501100012190","name":"Ministry of Science and Higher Education of the Russian Federation","doi-asserted-by":"publisher","award":["075-15-2024-544"],"award-info":[{"award-number":["075-15-2024-544"]}],"id":[{"id":"10.13039\/501100012190","id-type":"DOI","asserted-by":"publisher"}]},{"award":["075-15-2024-544"],"award-info":[{"award-number":["075-15-2024-544"]}],"id":[{"id":"https:\/\/ror.org\/00ghqgy32","id-type":"ROR","asserted-by":"publisher"}]}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["MAKE"],"abstract":"<jats:p>In this paper, we propose a novel approach, COMPAS (COMPose Actions and Slots), which leverages the strengths of state-of-the-art object-centric approaches for modeling the dynamics of an environment. Our method encodes the environment\u2019s state into symbol-like, object-centric representations, known as slots, where each slot corresponds to an individual object. This approach offers a structured and interpretable way to model complex environments by combining slots with action representations for accurate next-state prediction. The primary contribution of our work is an efficient world model with a dynamics predictor capable of predicting accurate trajectories in action-dependent environments. Additionally, our slot extractor module enhances the predictive capabilities by extracting deterministic slots that remain consistent both within a single trajectory and across episodes. Unlike slots sampled from a trainable distribution, deterministic slots are generated from a single trainable parameter together with slot positional embeddings. This design improves the consistency across episodes, which in turn leads to more accurate dynamics prediction. We present a comprehensive evaluation of our approach in various environments, demonstrating that our proposed method outperforms competing models in environments with discrete and continuous action spaces.<\/jats:p>","DOI":"10.3390\/make8050117","type":"journal-article","created":{"date-parts":[[2026,4,29]],"date-time":"2026-04-29T09:46:11Z","timestamp":1777455971000},"page":"117","update-policy":"https:\/\/doi.org\/10.3390\/mdpi_crossmark_policy","source":"Crossref","is-referenced-by-count":0,"title":["COMPAS: Compose Actions and Slots in Object-Centric World Models"],"prefix":"10.3390","volume":"8","author":[{"ORCID":"https:\/\/orcid.org\/0009-0007-3280-351X","authenticated-orcid":false,"given":"Vitaliy","family":"Vorobyov","sequence":"first","affiliation":[{"name":"Moscow Independent Research Institute of Artificial Intelligence, Moscow 105064, Russia"},{"name":"Federal Research Center \u201cComputer Science and Control\u201d, Russian Academy of Sciences, Moscow 119333, Russia"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0009-0009-0033-9294","authenticated-orcid":false,"given":"Leonid","family":"Ugadiarov","sequence":"additional","affiliation":[{"name":"Moscow Independent Research Institute of Artificial Intelligence, Moscow 105064, Russia"},{"name":"Cognitive AI Systems Lab, Moscow 123317, Russia"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0009-0002-1321-713X","authenticated-orcid":false,"given":"Vladimir","family":"Frolov","sequence":"additional","affiliation":[{"name":"Moscow Independent Research Institute of Artificial Intelligence, Moscow 105064, Russia"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Alexey","family":"Kovalev","sequence":"additional","affiliation":[{"name":"Moscow Independent Research Institute of Artificial Intelligence, Moscow 105064, Russia"},{"name":"Cognitive AI Systems Lab, Moscow 123317, Russia"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-6071-9557","authenticated-orcid":false,"given":"Aleksandr","family":"Panov","sequence":"additional","affiliation":[{"name":"Federal Research Center \u201cComputer Science and Control\u201d, Russian Academy of Sciences, Moscow 119333, Russia"},{"name":"Cognitive AI Systems Lab, Moscow 123317, Russia"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"1968","published-online":{"date-parts":[[2026,4,29]]},"reference":[{"key":"ref_1","unstructured":"Lin, B., Bouneffouf, D., and Rish, I. (2023). A Survey on Compositional Generalization in Applications. arXiv."},{"key":"ref_2","unstructured":"Greff, K., Van Steenkiste, S., and Schmidhuber, J. (2020). On the binding problem in artificial neural networks. arXiv."},{"key":"ref_3","unstructured":"Locatello, F., Weissenborn, D., Unterthiner, T., Mahendran, A., Heigold, G., Uszkoreit, J., Dosovitskiy, A., and Kipf, T. (2020). Object-Centric Learning with Slot Attention. arXiv."},{"key":"ref_4","unstructured":"Wu, Z., Dvornik, N., Greff, K., Kipf, T., and Garg, A. (2023). SlotFormer: Unsupervised Visual Dynamics Simulation with Object-Centric Models. arXiv."},{"key":"ref_5","unstructured":"Singh, G., Deng, F., and Ahn, S. (2022, January 25\u201329). Illiterate DALL-E Learns to Compose. Proceedings of the 10th International Conference on Learning Representations, ICLR 2022, Virtual."},{"key":"ref_6","unstructured":"Seitzer, M., Horn, M., Zadaianchuk, A., Zietlow, D., Xiao, T., Simon-Gabriel, C.J., He, T., Zhang, Z., Sch\u00f6lkopf, B., and Brox, T. (2023). Bridging the Gap to Real-World Object-Centric Learning. Proceedings of the The Eleventh International Conference on Learning Representations, OpenReview."},{"key":"ref_7","unstructured":"Wu, P., Escontrela, A., Hafner, D., Abbeel, P., and Goldberg, K. (2023). Daydreamer: World models for physical robot learning. Proceedings of the Conference on Robot Learning, PMLR."},{"key":"ref_8","doi-asserted-by":"crossref","unstructured":"Burgard, W., Hebert, M., and Bennewitz, M. (2016). World modeling. Springer Handbook of Robotics, Springer International Publishing.","DOI":"10.1007\/978-3-319-32552-1_45"},{"key":"ref_9","unstructured":"Ha, D., and Schmidhuber, J. (2018). World models. arXiv."},{"key":"ref_10","unstructured":"Hafner, D., Lillicrap, T.P., Norouzi, M., and Ba, J. (2020). Mastering Atari with Discrete World Models. arXiv."},{"key":"ref_11","unstructured":"Micheli, V., Alonso, E., and Fleuret, F. (2023, January 1\u20135). Transformers are Sample-Efficient World Models. Proceedings of the Eleventh International Conference on Learning Representations, ICLR, Kigali, Rwanda."},{"key":"ref_12","unstructured":"Guyon, I., Luxburg, U.V., Bengio, S., Wallach, H., Fergus, R., Vishwanathan, S., and Garnett, R. (2017). A simple neural network module for relational reasoning. Proceedings of the Advances in Neural Information Processing Systems, Curran Associates, Inc."},{"key":"ref_13","doi-asserted-by":"crossref","first-page":"2806","DOI":"10.1109\/TPAMI.2020.3045007","article-title":"A Review on Deep Learning Techniques for Video Prediction","volume":"44","author":"Oprea","year":"2022","journal-title":"IEEE Trans. Pattern Anal. Mach. Intell."},{"key":"ref_14","unstructured":"Oh, A.H., Agarwal, A., Belgrave, D., and Cho, K. (2022). Simple Unsupervised Object-Centric Learning for Complex and Naturalistic Videos. Proceedings of the Advances in Neural Information Processing Systems, Curran Associates, Inc."},{"key":"ref_15","unstructured":"Kipf, T., Elsayed, G.F., Mahendran, A., Stone, A., Sabour, S., Heigold, G., Jonschkowski, R., Dosovitskiy, A., and Greff, K. (2022, January 25\u201329). Conditional Object-Centric Learning from Video. Proceedings of the International Conference on Learning Representations (ICLR), Virtual."},{"key":"ref_16","doi-asserted-by":"crossref","unstructured":"Elsayed, G.F., Mahendran, A., van Steenkiste, S., Greff, K., Mozer, M.C., and Kipf, T. (2022). SAVi++: Towards end-to-end object-centric learning from real-world videos. Proceedings of the Advances in Neural Information Processing Systems (NeurIPS), Curran Associates, Inc.","DOI":"10.52202\/068431-2098"},{"key":"ref_17","unstructured":"Zadaianchuk, A., Seitzer, M., and Martius, G. (2023). Object-Centric Learning for Real-World Videos by Predicting Temporal Feature Similarities. Proceedings of the Thirty-Seventh Conference on Neural Information Processing Systems (NeurIPS 2023), Curran Associates, Inc."},{"key":"ref_18","doi-asserted-by":"crossref","unstructured":"Villar-Corrales, A., Wahdan, I., and Behnke, S. (2023). Object-Centric Video Prediction via Decoupling of Object Dynamics and Interactions. Proceedings of the Internation Conference on Image Processing (ICIP), Kuala Lumpur, Malaysia, 8\u201311 October 2023, IEEE.","DOI":"10.1109\/ICIP49359.2023.10222810"},{"key":"ref_19","doi-asserted-by":"crossref","unstructured":"Chang, M., Griffiths, T., and Levine, S. (2022). Object Representations as Fixed Points: Training Iterative Refinement Algorithms with Implicit Differentiation. Proceedings of the Advances in Neural Information Processing Systems, Curran Associates, Inc.","DOI":"10.52202\/068431-2369"},{"key":"ref_20","unstructured":"Jia, B., Liu, Y., and Huang, S. (2023, January 1\u20135). Improving Object-centric Learning with Query Optimization. Proceedings of the Eleventh International Conference on Learning Representations, ICLR, Kigali, Rwanda."},{"key":"ref_21","unstructured":"Singh, G., Kim, Y., and Ahn, S. (2023, January 1\u20135). Neural Systematic Binder. Proceedings of the Eleventh International Conference on Learning Representations, ICLR, Kigali, Rwanda."},{"key":"ref_22","unstructured":"Kirilenko, D., Vorobyov, V., Kovalev, A., and Panov, A. (2024, January 7\u201311). Object-Centric Learning with Slot Mixture Module. Proceedings of the Twelfth International Conference on Learning Representations, Vienna, Austria."},{"key":"ref_23","unstructured":"Patil, V., Radler, A., Klotz, D., and Hochreiter, S. (2024). Simplified priors for Object-Centric Learning. arXiv."},{"key":"ref_24","unstructured":"Van Den Oord, A., Vinyals, O., and Kavukcuoglu, K. (2017, January 4\u20139). Neural discrete representation learning. Proceedings of the 31st International Conference on Neural Information Processing Systems (NIPS 2017), Long Beach, CA, USA."},{"key":"ref_25","doi-asserted-by":"crossref","unstructured":"Caron, M., Touvron, H., Misra, I., J\u00e9gou, H., Mairal, J., Bojanowski, P., and Joulin, A. (2021). Emerging Properties in Self-Supervised Vision Transformers. Proceedings of the International Conference on Computer Vision (ICCV), Montreal, QC, Canada, 10\u201317 October 2021, IEEE.","DOI":"10.1109\/ICCV48922.2021.00951"},{"key":"ref_26","unstructured":"Kosiorek, A., Kim, H., Teh, Y.W., and Posner, I. (2018). Sequential attend, infer, repeat: Generative modelling of moving objects. Proceedings of the Advances in Neural Information Processing Systems, Curran Associates, Inc."},{"key":"ref_27","unstructured":"Jiang, J., Janghorbani, S., de Melo, G., and Ahn, S. (2020, January 26\u201330). SCALOR: Generative World Models with Scalable Object Representations. Proceedings of the 8th International Conference on Learning Representations, ICLR, Addis Ababa, Ethiopia."},{"key":"ref_28","doi-asserted-by":"crossref","unstructured":"Crawford, E., and Pineau, J. (2020). Exploiting Spatial Invariance for Scalable Unsupervised Object Tracking. Proceedings of the AAAI Conference on Artificial Intelligence, AAAI Press.","DOI":"10.1609\/aaai.v34i04.5777"},{"key":"ref_29","unstructured":"Kossen, J., Stelzner, K., Hussing, M., Voelcker, C., and Kersting, K. (2019). Structured Object-Aware Physics Prediction for Video Modeling and Planning. arXiv."},{"key":"ref_30","doi-asserted-by":"crossref","first-page":"61","DOI":"10.1109\/TNN.2008.2005605","article-title":"The Graph Neural Network Model","volume":"20","author":"Scarselli","year":"2009","journal-title":"IEEE Trans. Neural Netw."},{"key":"ref_31","unstructured":"Lin, Z., Wu, Y.F., Peri, S., Fu, B., Jiang, J., and Ahn, S. (2020). Improving generative imagination in object-centric world models. Proceedings of the International Conference on Machine Learning, PMLR."},{"key":"ref_32","unstructured":"Kipf, T.N., van der Pol, E., and Welling, M. (2020, January 26\u201330). Contrastive Learning of Structured World Models. Proceedings of the 8th International Conference on Learning Representations, ICLR, Addis Ababa, Ethiopia."},{"key":"ref_33","unstructured":"Biza, O., van der Pol, E., and Kipf, T. (2021, January 24). The Impact of Negative Sampling on Contrastive Structured World Models. Proceedings of the ICML Workshop: Self-Supervised Learning for Reasoning and Perception, Online."},{"key":"ref_34","unstructured":"Biza, O., Platt, R., van de Meent, J.W., Wong, L.L.S., and Kipf, T. (2022, January 25\u201329). Binding Actions to Objects in World Models. Proceedings of the ICLR 2022 Workshop on Objects, Structure and Causality, Online."},{"key":"ref_35","unstructured":"Vaswani, A., Shazeer, N., Parmar, N., Uszkoreit, J., Jones, L., Gomez, A.N., Kaiser, \u0141., and Polosukhin, I. (2017). Attention is all you need. Proceedings of the 31st International Conference on Neural Information Processing Systems, NeurIPS, Curran Associates, Inc."},{"key":"ref_36","doi-asserted-by":"crossref","first-page":"248","DOI":"10.1109\/TG.2018.2877047","article-title":"ViZDoom Competitions: Playing Doom from Pixels","volume":"11","author":"Wydmuch","year":"2019","journal-title":"IEEE Trans. Games"},{"key":"ref_37","unstructured":"Ahmed, O., Tr\u00e4uble, F., Goyal, A., Neitz, A., Bengio, Y., Sch\u00f6lkopf, B., W\u00fcthrich, M., and Bauer, S. (2020). Causalworld: A robotic manipulation benchmark for causal structure and transfer learning. arXiv."},{"key":"ref_38","unstructured":"Zhu, Y., Wong, J., Mandlekar, A., Mart\u00edn-Mart\u00edn, R., Joshi, A., Nasiriany, S., and Zhu, Y. (2022). robosuite: A Modular Simulation Framework and Benchmark for Robot Learning. arXiv."},{"key":"ref_39","unstructured":"Zhou, G., Pan, H., LeCun, Y., and Pinto, L. (2024). DINO-WM: World Models on Pre-trained Visual Features enable Zero-shot Planning. arXiv."}],"container-title":["Machine Learning and Knowledge Extraction"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/www.mdpi.com\/2504-4990\/8\/5\/117\/pdf","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,4,29]],"date-time":"2026-04-29T10:53:08Z","timestamp":1777459988000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.mdpi.com\/2504-4990\/8\/5\/117"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,4,29]]},"references-count":39,"journal-issue":{"issue":"5","published-online":{"date-parts":[[2026,5]]}},"alternative-id":["make8050117"],"URL":"https:\/\/doi.org\/10.3390\/make8050117","relation":{},"ISSN":["2504-4990"],"issn-type":[{"value":"2504-4990","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,4,29]]}}}