State Space Model
PulseAugur coverage of State Space Model — every cluster mentioning State Space Model across labs, papers, and developer communities, ranked by signal.
3 day(s) with sentiment data
-
New IR275K benchmark targets infrared super-resolution challenges
Researchers have introduced IR275K, a new benchmark dataset designed to evaluate multi-frame super-resolution (MFSR) techniques specifically for infrared remote sensing applications. This benchmark addresses the unique …
-
New research details information omission in air-gapped LLM agents
A new research paper, "Where Facts Go Missing," introduces a taxonomy and attribution methodology for information omission in air-gapped LLM agent pipelines. The study identifies that a significant portion of omissions,…
-
HERMES model advances traffic conflict prediction using graph neural networks
Researchers have developed HERMES, a novel graph neural network designed for predicting traffic conflicts at signalized intersections. This model represents vehicles and pedestrians as heterogeneous nodes and their inte…
-
DevOps engineer shares 10 Claude prompt patterns for efficient automation
A DevOps engineer shares ten prompt engineering patterns learned while using Claude for complex automation tasks. The advice emphasizes providing context, showing examples, and specifying constraints to improve AI outpu…
-
FuseMamba-VD introduces efficient dual-branch architecture for video violence detection
Researchers have introduced FuseMamba-VD, a novel dual-branch architecture for efficient video violence detection. This model combines a State Space Model (SSM) backbone with a gating mechanism to fuse spatial and tempo…
-
MambaLIE: State Space Model enhances low-light images efficiently
Researchers have developed MambaLIE, a novel method for low-light image enhancement that utilizes a State Space Model (SSM). This approach aims to overcome the limitations of existing Convolutional Neural Networks (CNNs…
-
ICML 2026 shifts focus to AI theory, science, and embodied intelligence
The International Conference on Machine Learning (ICML) 2026 is seeing a near doubling of submissions, yet maintaining a strict acceptance rate of 26.56%, indicating a significant recalibration of academic review standa…
-
SUMO framework unifies visual object tracking and motion segmentation
Researchers have introduced SUMO, a novel framework designed to unify visual object tracking (VOT) and moving object segmentation (MOS). This zero-shot, training-free system integrates nonlinear dynamics with vision-bas…
-
New research explores advanced video generation and manipulation with diffusion models
Researchers are exploring advanced techniques to improve video generation and manipulation using diffusion models. One approach involves integrating State Space Models (SSMs) with video diffusion models to enhance effic…
-
New SSM adapters outperform LoRA for long-context fine-tuning
Researchers have developed a new parameter-efficient fine-tuning (PEFT) method called Hankel Reduced order Model (HRM) adapters, which utilize state space models (SSMs) for long-context fine-tuning. Unlike traditional P…
-
New research explores diffusion and state space models beyond autoregressive AI
Two new arXiv papers explore advanced modeling techniques beyond traditional autoregressive language models. The first paper surveys Diffusion Models, Code World Models, and State Space Models for code intelligence, sug…
-
Infrawise tool prevents AI-written code from causing costly cloud errors
Infrawise has developed a tool that analyzes code and infrastructure to prevent costly AI-generated errors, such as inefficient database scans. The system first scans code repositories to identify database client calls …
-
StateSMix compressor uses Mamba SSMs and n-grams for online lossless compression
Researchers have developed StateSMix, a novel lossless compression algorithm that utilizes Mamba-style State Space Models (SSMs) combined with sparse n-gram context mixing. This system trains token-by-token on the data …
-
ViM-Q enables efficient Vision Mamba model inference on FPGAs
Researchers have developed ViM-Q, a novel algorithm-hardware co-design specifically for accelerating Vision Mamba (ViM) model inference on FPGAs. This approach tackles challenges in quantizing dynamic activation outlier…
-
NAKUL-Med model enhances medical signal analysis with dynamic kernels and spectral context
Researchers have developed NAKUL-Med, a novel spectral-graph state space model designed to enhance the analysis of multi-channel medical signals. This model addresses limitations in existing state space models by incorp…
-
SSMProbe framework reveals importance of token order in visual representations
Researchers have developed SSMProbe, a new framework for analyzing visual representations in AI models. This method utilizes State Space Models (SSMs) to account for the critical role of token order, challenging the tra…
-
Caracal architecture uses Fourier transforms for efficient long-sequence modeling
Researchers have introduced Caracal, a new architecture designed to improve the scalability of large language models for processing long sequences. Caracal replaces the computationally expensive attention mechanism with…
-
L2RU introduces stable state-space models for machine learning and control
Researchers have introduced L2RU, a new class of structured state-space models (SSMs) designed to ensure input-output stability and robustness. This architecture integrates deep learning expressiveness with dynamical sy…
-
New Mamba model variant enhances memory retention and bilinear computation
Researchers have introduced Bilinear Input Modulation (BIM) to enhance Selective State Space Models (SSMs), specifically Mamba, by incorporating state-input products. This augmentation allows for improved memory retenti…
-
Mamba model offers Transformer-level performance with faster inference and longer context
Mamba, a new State Space Model (SSM), presents an alternative to the dominant Transformer architecture in AI. It aims to match Transformer performance and scaling laws while efficiently handling extremely long sequences…