#autoencoder
7 resultsDeep Learning for Accelerated Long-Horizon Forecasting of Multicomponent Multiphase Microstructure Evolution in High-Entropy Alloys
Hamidreza Razavi, Nele Moelans
The paper introduces a surrogate model combining autoencoders, graph convolutional networks, and LSTM to rapidly predict long‑term microstructure evolution in multicomponent high‑e…
A Lightweight Foundation Model for Collider Physics with Multi-Domain Adaptation
Liangyu Wu, Qibin Liu, Alexander Yue +1
The paper introduces NEXUS, a lightweight autoencoder foundation model with ~3 M parameters that is pretrained on Large Hadron Collider track data and fine‑tuned for collider tasks…
System-Aware Adaptive CSI Feedback via RL-Guided Autoencoder Switching in Multi-User MIMO System
Maryam Ansarifard, Mohit K. Sharma, George Exarchakos +1
The paper introduces a reinforcement‑learning controller that selects among multiple pretrained autoencoders with different compression ratios to adaptively compress CSI feedback i…
ZipL-Dialog: Memory-Efficient Long-Form Spoken Dialog Synthesis via Latent Flow Matching
Jihwan Kim, Nam Soo Kim
The paper introduces ZipL-Dialog, a method that compresses mel-spectrograms into a low‑dimensional latent space and applies flow‑matching there, enabling memory‑efficient synthesis…
Mapping recrystallization trajectories in GaAs using latent space diffraction analysis
Ellis Rae Kennedy, Kwanghwi Je, Erik Thiede
The paper presents a latent‑space method using a convolutional autoencoder and unsupervised clustering to map recrystallization pathways from in‑situ 4D‑STEM diffraction data of io…
Learning reduced-order latent linear models for Kalman filtering of nonlinear systems
Manas Mejari, Milad Banitalebi Dehkordi, Dario Piga
The paper introduces an end-to-end learning framework that jointly trains an autoencoder and a reduced-order linear model to perform Kalman filtering directly in a low-dimensional…
VisCo: Leveraging Large Language Models as Intrinsic Encoders for Visual Token Compression
Yupeng Zheng, Kai Zou, Bin Liu +1
VisCo introduces a training-efficient self-compression framework that reuses a pretrained vision-language model as an intrinsic autoencoder to compress visual tokens into a small s…