Natural Building Blocks for Structured World Models: Theory, Evidence, and Scaling
VERSES AI Research Lab(VERSES AI研究实验室);University of Tübingen(图宾根大学);ELLIS Institute, Tübingen(图宾根ELLIS研究所);MPI for Intelligent Systems, Tübingen(图宾根智能系统研究所)
Lancelot Da Costa, Sanjeev Namjoshi, Mohammed Abbas Ansari, Bernhard Schölkopf
备注:
13 pages, 3 figures, under review for World Modeling Workshop 2026
The field of world modeling is fragmented, with researchers developing bespoke architectures that rarely build upon each other. We propose a framework that specifies the natural building blocks for structured world models based on the fundamental stochastic processes that any world model must capture: discrete processes (logic, symbols) and continuous processes (physics, dynamics); the world model is then defined by the hierarchical composition of these building blocks. We examine Hidden Markov Models (HMMs) and switching linear dynamical systems (sLDS) as natural building blocks for discrete and continuous modeling--which become partially-observable Markov decision processes (POMDPs) and controlled sLDS when augmented with actions. This modular approach supports both passive modeling (generation, forecasting) and active control (planning, decision-making) within the same architecture. We avoid the combinatorial explosion of traditional structure learning by largely fixing the causal architecture and searching over only four depth parameters. We review practical expressiveness through multimodal generative modeling (passive) and planning from pixels (active), with performance competitive to neural approaches while maintaining interpretability. The core outstanding challenge is scalable joint structure-parameter learning; current methods finesse this by cleverly growing structure and parameters incrementally, but are limited in their scalability. If solved, these natural building blocks could provide foundational infrastructure for world modeling, analogous to how standardized layers enabled progress in deep learning.
Block-Structured Optimization for Subgraph Detection in Interdependent Networks
Hefei University of Technology(合肥工业大学);University at Albany – SUNY(纽约州立大学奥尔巴尼分校);The University of Texas at Dallas(德克萨斯大学达拉斯分校);Mininglamp Academy of Sciences, Mininglamp Technologies(明略科技集团明略科学院)
Efficient Transformer-based Large Scale Language Representations using Hardware-friendly Block Structured Pruning
University of Connecticut(康涅狄格大学);Northeastern University(东北大学);Syracuse University(雪城大学);Microsoft Corporation(微软公司);Stevens Institute of Technology(史蒂文斯理工学院)
Learning Diverse-Structured Networks for Adversarial Robustness
Hong Kong Baptist University(香港浸会大学);University of Wisconsin-Madison(威斯康星大学麦迪逊分校);RIKEN(理化学研究所);University of Sydney(悉尼大学);Tencent AI Lab(腾讯人工智能实验室);University of Tokyo(东京大学)
A Randomized Nonmonotone Block Proximal Gradient Method for a Class of Structured Nonlinear Programming
Simon Fraser University(西蒙弗雷泽大学);Microsoft Research(微软研究院)
我们提出了一个随机非单调块邻近梯度(RNBPG)方法,用于最小化光滑(可能非凸)函数和块可分(可能非凸非光滑)函数之和。在每次迭代中,该方法根据任何规定的概率分布随机选取一个块,并通常解决几个相关的邻近子问题,通常具有封闭形式的解决方案,直到目标值取得一定的进展。与通常的随机块坐标下降法[23,20]相比,我们的方法具有非单调的味道,并使用可变步长,可以部分利用目标函数的光滑分量的局部曲率信息。我们证明了该方法解序列的任何聚点都是问题的一个稳定点{ \it almost surely},并且该方法能够以很高的概率找到一个近似的稳定点.我们还建立了一个次线性的收敛速率的方法在最小期望平方范数的某些邻近梯度的迭代。当所考虑的问题是凸的,我们证明了RNBPG产生的期望目标值收敛到问题的最优值。在一些假设下,我们进一步建立了一个次线性和线性收敛速率的期望目标值产生的单调版本的RNBPG。最后,我们进行了一些初步的实验,以测试性能的RNBPG上的$\ell_1$正则化最小二乘问题和机器学习中的对偶SVM问题。计算结果表明,我们的方法大大优于随机块坐标{ \it下降}方法与固定或可变步长。