paper-with-me

Papers

Recency-weighted Markovian inference

2017-11-08 · Kristjan Kalm

We describe a Markov latent state space (MLSS) model, where the latent state distribution is a decaying mixture over multiple past states. We present a simple sampling algorithm that allows to approximate such high-order MLSS with fixed time and memory costs.

📄 PDF Abstract BibTeX arXiv:1711.03038

Code (0)

등록된 구현이 없습니다.

Similar Papers 제목 키워드 기반

Stochastic Stability of a Recency Weighted Sampling Dynamic

2020-09-27 · Alexander Aurell, Gustav Karreskog

We introduce and study a model of long-run convention formation for rare interactions. Players in this model form beliefs by observing a recency-weighted sample of past interactions, to which they noisily best respond. W…

Prompt Compression in Production Task Orchestration: A Pre-Registered Randomized Trial

2026-03-06 · Warren Johnson, Charles Lee arxiv

The economics of prompt compression depend not only on reducing input tokens but on how compression changes output length, which is typically priced several times higher. We evaluate this in a pre-registered six-arm rand…

Markovian Score Climbing: Variational Inference with KL(p||q)

2020-03-23 · NeurIPS 2020 12 · Christian A. Naesseth, Fredrik Lindsten, David Blei

Modern variational inference (VI) uses stochastic gradients to avoid intractable expectations, enabling large-scale probabilistic inference in complex models. VI posits a family of approximating distributions q and then …

Variational Inference

Recency-Weighted Temporally-Segmented Ensemble for Time-Series Modeling

2024-03-04 · Pål V. Johnsen, Eivind Bøhn, Sølve Eidnes, Filippo Remonato 외

Time-series modeling in process industries faces the challenge of dealing with complex, multi-faceted, and evolving data characteristics. Conventional single model approaches often struggle to capture the interplay of di…

Time Series

TempoFit: Plug-and-Play Layer-Wise Temporal KV Memory for Long-Horizon Vision-Language-Action Manipulation

2026-03-08 · Jun Sun, Boyu Yang, Jiahao Zhang, Ning Ma 외 arxiv

Pretrained Vision-Language-Action (VLA) policies have achieved strong single-step manipulation, but their inference remains largely memoryless, which is brittle in non-Markovian long-horizon settings with occlusion, stat…