发表机构
The Chinese University of Hong Kong (CUHK)（香港中文大学（CUHK））
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
该研究提出正交JEPA框架，通过正交预测分解改进潜在世界模型，在多类任务实验中评估了其表示质量、预测、规划及长时程稳定性。
AI 中文摘要
世界模型构建潜在状态，以支持对底层系统的预测、规划与推理。联合嵌入预测架构（JEPAs）提供了一种直接学习此类状态的方式，其通过在表示空间中预测目标，而非重构观测的每个细节。然而，标准JEPAs将所有可预测内容组织为单个目标嵌入与一条预测通路，在复杂系统中，这种整体状态会将冗余容量分配给主导信号，同时为占比更小的预测结构提供微弱或冲突的梯度。我们提出正交JEPA（\textbackslash method），一种基于正交预测分解的潜在世界建模框架：学习基矩阵将每个目标状态分解为多个分量，专用预测分支从共享上下文表示中估计每个分量；预测回归保留状态合成所需的分量幅度，正交性目标抑制重复方向，分量活性正则化维持投影目标的变异性，在线方差正则化抑制按坐标的编码器坍缩。预测分量被合成为完整潜在状态，可供读出器、解码器、规划器或自回归展开使用，当目标为时间上的未来、空间上的隐藏或同一系统的其他部分观测时，该预测状态机制同样适用。在受控视觉、单细胞转录组学、纵向健康记录、连续控制及分子动力学上开展的实验，评估了表示质量、预测、规划及长时程稳定性。
英文摘要
World models construct latent states that support prediction, planning, and reasoning about an underlying system. Joint-embedding predictive architectures (JEPAs) offer a direct way to learn such states by predicting targets in representation space instead of reconstructing every detail of the observation. Standard JEPAs, however, organize all predictable content through one target embedding and one prediction pathway. In complex systems, this monolithic state can allocate redundant capacity to dominant signals while providing weak or conflicting gradients to less dominant predictive structure. We introduce \method, a latent world-modeling framework based on orthogonal predictive factorization. Learned basis matrices analyze each target state into multiple components, and a dedicated prediction branch estimates each component from a shared context representation. Predictive regression preserves the factor magnitudes required for state synthesis, an orthogonality objective discourages repeated directions, factor-activity regularization maintains variation in projected targets, and online variance regularization discourages coordinate-wise encoder collapse. Predicted components are synthesized into a complete latent state that can be used by a readout, decoder, planner, or autoregressive rollout. The same predictive-state mechanism applies when the target is temporally future, spatially hidden, or another partial observation of the same system. Experiments on controlled vision, single-cell transcriptomics, longitudinal health records, continuous control, and molecular dynamics evaluate representation quality, forecasting, planning, and long-horizon stability.
CommentsThis manuscript has been superseded by a substantially revised version, arXiv:2609.20800. The simultaneous existence of both records has caused significant confusion regarding the relationship and development of the work. To prevent further misunderstanding and maintain a clear scientific record, we request withdrawal of arXiv:2608.20065
