State Space Models
PulseAugur coverage of State Space Models — every cluster mentioning State Space Models across labs, papers, and developer communities, ranked by signal.
9 day(s) with sentiment data
-
New nonlinear interventional state space models identify neural dynamics
Researchers have developed a new framework for identifying neural dynamics using nonlinear interventional state space models (iSSMs). This advancement extends previous iSSM capabilities beyond linear dynamics to a broad…
-
New research merges diffusion and state-space models for advanced time-series prediction · 2 sources tracked
Two new research papers introduce novel approaches to combine diffusion models with state-space models for improved time-series forecasting and spatiotemporal prediction. The first paper, "Diffusion-Driven State Space M…
-
AWS and EDB Postgres AI enable semantic layers for agentic AI
AWS has detailed how to construct a semantic layer for agentic AI applications using EDB Postgres AI and Amazon Bedrock AgentCore. This approach integrates business definitions and data governance directly within the op…
-
AWS enables low-latency voice AI with State Space Models
AWS is enabling the development of ultra-low-latency voice AI applications through its infrastructure, partnering with companies like Cartesia. Cartesia specializes in real-time voice AI, utilizing State Space Models (S…
-
Dynamical Memory Transformer offers constant cost attention for long sequences
Researchers have introduced the Dynamical Memory Transformer (DMT), a novel hybrid architecture designed to improve how models handle long sequences. Unlike standard transformers that store an explicit history, leading …
-
New research questions state tracking robustness in recurrent AI models
A new paper from arXiv explores the limitations of state tracking in recurrent neural network architectures, arguing that error control dynamics are as crucial as theoretical expressivity. The research demonstrates that…
-
Deep learning models struggle with financial forecasting benchmarks, study finds
A new study published on arXiv evaluates 16 deep learning architectures for financial time series forecasting, finding that many popular models like generic Transformers and State Space Models underperform simpler bench…
-
New survey paper unifies Time Series Anomaly Detection research
A new survey paper published on arXiv provides a comprehensive overview of the evolving field of Time Series Anomaly Detection (TSAD). It categorizes recent advancements, including the integration of diffusion models, s…
-
AI models show position bias in long-context question answering · arXiv research
A new research paper titled "Mixing Matters: Evidence Position Bias Across Sequence Mixers in Long-Context Question Answering" explores how different AI architectures handle information presented at various positions wi…
-
Liquid AI unveils hybrid LLM architectures for on-device deployment
Liquid AI has developed new LLM architectures, LFM one and LFM two, designed for on-device deployment with improved memory, latency, and context length compared to traditional transformers. These models, particularly LF…
-
Transformers criticized for limiting AI progress; new architectures emerge
The current transformer architecture, which powers most AI models, lacks persistent memory and continual learning capabilities, hindering progress towards superintelligence. While retrieval-augmented generation (RAG) of…
-
State Space Models Mimic Brain Activity and Scale to Cognitive Functions
Researchers have proposed that State Space Models (SSMs) can naturally generate neural behaviors observed in the brain, such as time cells, oscillatory activity, and ramping activity. These behaviors emerge from a struc…
-
New MaRK framework dynamically conditions State Space Models
Researchers have introduced MaRK (Markov-adapted Recurrent Kernels), a novel framework for dynamically conditioning State Space Models (SSMs) in iterative generation tasks. Unlike previous methods that modulated inputs …
-
New Exponentially Weighted Signature Model Extends State-Space Models
Researchers have introduced the exponentially weighted signature (EWS), a novel continuous-time model that extends state-space models (SSMs) by computing iterated integrals of a path with increments weighted by a learna…
-
New theory frames State-Space Model training as optimal control
Researchers have developed a new theoretical framework for understanding the training dynamics of State-Space Models (SSMs). By formulating continuous-time SSM parameter optimization as an ensemble optimal control probl…
-
New research improves State Space Models with depth recurrence and input reshaping
A new research paper introduces two methods to enhance State Space Models (SSMs), making them more efficient and performant. The first method, depth recurrence, reduces the memory footprint of SSMs without sacrificing p…
-
New LipSSM architecture enhances DNN robustness with cascaded state-space models
Researchers have introduced LipSSM, a novel architecture for cascaded state-space models (SSMs) designed to enhance the robustness of deep neural networks (DNNs). This new model builds upon the LipKernel concept, which …
-
New Fusion Method Merges Dissimilar Vision Models
Researchers have developed a novel method called Riemannian--Lorentz Parameter Fusion (RLPF) to merge independently trained vision models, even when their architectures differ. This technique addresses the challenges of…
-
State Space Models Enhance Long-Context Language Model Efficiency
Researchers have developed a new method for demonstration selection in language models, which aims to reduce computational costs associated with long-context scenarios. The approach utilizes state space models (SSMs) to…
-
Gating mechanisms hinder State Space Models' in-context learning, research finds
A new research paper published on arXiv explores the role of gating mechanisms in State Space Models (SSMs), which are emerging as an alternative to Transformers for sequence modeling. The study reveals that these gatin…