Naga: Vedic Encoding for Deep State Space Models
cs.LG, cs.AI, cs.SY, eess.SY
Submitted: 2025-11-17
Updated: 2026-10-12
License: http://creativecommons.org/licenses/by/4.0/
The gist: This paper presents Naga, a deep State Space Model (SSM) encoding approach inspired by structural concepts from Vedic mathematics.
Terminology
Abstract
This paper presents Naga, a deep State Space Model (SSM) encoding approach inspired by structural concepts from Vedic mathematics. The proposed method introduces a bidirectional representation for time series by jointly processing forward and time-reversed input sequences. These representations are then combined through an element-wise (Hadamard) interaction, resulting in a Vedic-inspired encoding that enhances the model's ability to capture temporal dependencies across distant time steps. We evaluate Naga on multiple long-term time series forecasting (LTSF) benchmarks, including ETTh1, ETTh2, ETTm1, ETTm2, Weather, Traffic, and ILI. The experimental results show that Naga outperforms 28 current state of the art models and demonstrates improved efficiency compared to existing deep SSM-based approaches. The findings suggest that incorporating structured, Vedic-inspired decomposition can provide an interpretable and computationally efficient alternative for long-range sequence modeling.
Sources
- Layer Normalization
- Transformers are SSMs: Generalized Models and Efficient Algorithms Through Structured State Space Duality
- Long-term Forecasting with TiDE: Time-series Dense Encoder
- PatchMixer: A Patch-Mixing Architecture for Long-Term Time Series Forecasting
- Mamba: Linear-Time Sequence Modeling with Selective State Spaces
- Efficiently Modeling Long Sequences with Structured State Spaces
- Reformer: The Efficient Transformer
- Long-term series forecasting with Query Selector -- efficient model of sparse attention
- PAniC-3D: Stylized Single-view 3D Reconstruction from Portraits of Anime Characters
- Revisiting Long-term Time Series Forecasting: An Investigation on Linear Mapping
- SegRNN: Segment Recurrent Neural Network for Long-Term Time Series Forecasting
- Mixture-of-Linear-Experts for Long-term Time Series Forecasting
- A Time Series is Worth 64 Words: Long-term Forecasting with Transformers
- Advancing Intelligent Sequence Modeling: Evolution, Trade-offs, and Applications of State-Space Architectures from S4 to Mamba
- Physics-Inspired Deep Learning Anti-Aliasing Framework in Efficient Channel State Feedback
- Are Transformers Effective for Time Series Forecasting?
- First De-Trend then Attend: Rethinking Attention for Time-Series Forecasting
- Disentangled Parameter-Efficient Linear Model for Long-Term Time Series Forecasting
- FiLM: Frequency improved Legendre Memory Model for Long-term Time Series Forecasting
Related papers
- Polynomial-Augmented Neural Networks (PANNs) with Weak Orthogonality Constraints for Enhanced Function and PDE Approximation
- AIRL-S: Unifying Reinforcement Learning and Search-Based Test-Time Scaling via Adversarial Inverse Reinforcement Learning
- Transformers as Bayesian In-Context Experimenters: Smoothness-Adaptive Efficient ATE Estimation
- Convergence issues in Relational Concept Analysis based on AOC-posets
- Beliefs Beyond Posteriors: The Predictive Cost of ELBO Binding
- Understanding Diffusion Models via Ratio-Based Function Approximation with SignReLU Networks