The recent empirical success of Mamba and other selective state space models (SSMs) has renewed interest in non-attention architectures for sequence modeling, yet their theoretical foundations remain underexplored. We present a first-step analysis of generalization and learning dynamics for a simplified but representative Mamba block: a single-layer, single-head selective SSM with input-dependent gating, followed by a two-layer MLP trained via gradient descent (GD). Our study adopts a structured data model with tokens that include both class-relevant and class-irrelevant patterns under token-level noise and examines two canonical regimes: majority-voting and locality-structured data sequences. We prove that the model achieves guaranteed generalization by establishing non-asymptotic sample complexity and convergence rate bounds, which improve as the effective signal increases and the noise decreases. Furthermore, we show that the gating vector aligns with class-relevant features while ignoring irrelevant ones, thereby formalizing a feature-selection role similar to attention but realized through selective recurrence. Numerical experiments on synthetic data justify our theoretical results. Overall, our results provide principled insight into when and why Mamba-style selective SSMs learn efficiently, offering a theoretical counterpoint to Transformer-centric explanations.
DG-Mamba: Robust and Efficient Dynamic Graph Structure Learning with Selective State Space Models
Beihang University（北京航空航天大学）；National Superior College for Engineers, Beihang University（北京航空航天大学国家卓越工程师学院）；Guangxi Normal University（广西师范大学）；The Third Research Institute of Ministry of Public Security（公安部第三研究所）
ButterMamba: Butterworth-Enhanced Spatial-Temporal Mamba for Efficient Traffic Flow Prediction
Institute of Physical Science and Information Technology, Anhui University（安徽大学物理与信息技术学院）；School of Innovation and Entrepreneurship, Shandong University（山东大学创新创业学院）；School of Computer Science and Technology, Anhui University（安徽大学计算机科学与技术学院）
MSEMG: Surface Electromyography Denoising with a Mamba-based Efficient Network
National Yang Ming Chiao Tung University（国立阳明交通大学）；National Taiwan University（国立台湾大学）；University of Palermo（巴勒莫大学）；Academic Sinica（中央研究院）；Research Center for Information Technology Innovation（资讯科技创新研究中心）
FEMBA on the Edge: Physiologically-Aware Pre-Training, Quantization, and Deployment of a Bidirectional Mamba EEG Foundation Model on an Ultra-low Power Microcontroller
Integrated Systems Laboratory, ETH Zürich（苏黎世联邦理工学院集成系统实验室）；DEI, University of Bologna（博洛尼亚大学电气、电子与信息工程系）
Mamba Can Learn Low-Dimensional Targets In-Context via Test-Time Feature Learning
KAIST AI（韩国科学技术院人工智能研究所）；University of Tokyo（东京大学）；RIKEN AIP（日本研究机构人工智能计划）；The Institute of Statistical Mathematics（统计数学研究所）；University of Tokyo/RIKEN AIP（东京大学/日本研究机构人工智能计划）
A Mamba Foundation Model for Time Series Forecasting
Tsinghua University（清华大学）；Naval University of Engineering（海军工程大学）；Beijing Institute of Technology（北京理工大学）；China Meteorological Administration（中国气象局）；National Supercomputing Center（国家超级计算中心）；Zhejiang Lab（之江实验室）
Automatic Noise Filtering with Dynamic Sparse Training in Deep Reinforcement Learning
Eindhoven University of Technology（埃因霍温理工大学）；University of Alberta（阿尔伯塔大学）；University of Twente（特文特大学）；Alberta Machine Intelligence Institute (Amii)（阿尔伯塔机器智能研究所（Amii））；University of Luxembourg（卢森堡大学）
Improving information retrieval from electronic health records using dynamic and multi-collaborative filtering
University of Illinois at Chicago（伊利诺伊大学芝加哥分校）；CSCI Consulting Inc.（CSCI咨询公司）；The Ohio State University（俄亥俄州立大学）；Regenstrief Institute（雷根斯特里夫研究所）；Indiana University School of Medicine（印第安纳大学医学院）
