Results 91 to 100 of about 27,364 (268)

Matryoshka Quantization

open access: yesCoRR
Quantizing model weights is critical for reducing the communication and inference costs of large models. However, quantizing models -- especially to low precisions like int4 or int2 -- requires a trade-off in model quality; int2, in particular, is known to severely degrade model quality. Consequently, practitioners are often forced to maintain multiple
Pranav Ajit Nair   +4 more
openaire   +3 more sources

Machine‐Learning‐Enhanced Printed Vertical Magnetoresistive Sensors for Transparent, Flexible, Multimodal Interactive Magnetoelectronics

open access: yesAdvanced Science, EarlyView.
This study presents printed magnetoresistive sensors with a vertically aligned architecture that enables high optical transparency and mechanical flexibility. By integrating deep learning for the analysis of complex spatiotemporal signal patterns, the system further achieves intelligent multimodal interaction capabilities.
Rui Xu   +11 more
wiley   +1 more source

Optimal Number of Choices in Rating Contexts

open access: yesBig Data and Cognitive Computing, 2019
In many settings, people must give numerical scores to entities from a small discrete set—for instance, rating physical attractiveness from 1–5 on dating sites, or papers from 1–10 for conference reviewing.
Sam Ganzfried, Farzana Beente Yusuf
doaj   +1 more source

Branched Quantization

open access: yesPhysical Review Letters, 2012
We propose a method for quantization of Lagrangians for which the Hamiltonian, as a function of momentum, is a branched function with cusps. Appropriate boundary conditions, which we identify, insure unitary time evolution. In special cases a dual (canonical) transformation maps the problem into a problem of quantum mechanics on singular spatial ...
Shapere, Alfred, Wilczek, Frank
openaire   +6 more sources

Dual‐Mode Nanoporous SiO2 Memristors with Coexisting Volatile and Nonvolatile Dynamics for Reservoir Computing

open access: yesAdvanced Science, EarlyView.
A nanoporous SiO2 memristor enabling reconfigurable volatile and non‐volatile switching within a single device is demonstrated. The dual‐mode functionality supports both physical reservoir dynamics and synaptic weight storage, allowing unified hardware implementation of reservoir computing for temporal information processing, including image and ...
Bohao Ding   +5 more
wiley   +1 more source

AI‐Assisted Digital Single‐Molecule Activity Tracker for Decoupling Intrinsic Heterogeneity from Photo‐Oxidative Damage in High‐Photon‐Flux Enzymology

open access: yesAdvanced Science, EarlyView.
Employing a digital single‐molecule activity tracker (dSMAT), this research demonstrates that high‐photon‐flux irradiation drives progressive oxidative scarring in polymerases. Unlike simple thermal denaturation, real‐time kinetic tracking dynamically visualizes enzymes degrading into multiple impaired subpopulations.
Anran Zheng   +11 more
wiley   +1 more source

Molecularly Engineered Wing‐Shaped Azobenzene Memristors for Logic‐in‐Memory and Edge Visual Intelligence

open access: yesAdvanced Science, EarlyView.
Rational engineering of terminal substituents in symmetric azobenzene‐based molecules enables precise control over conformationally coupled charge‐transfer processes. This design yields tunable nonvolatile resistive memory behaviors, ranging from write‐once‐read‐many‐times (WORM) to rewritable switching.
Yanze Liu   +11 more
wiley   +1 more source

Chiral Plasmonic Printing of Discrete Hierarchical Chiral Au@Chiral Ag Semi‐Core‐Shell Nanorods With Chiral Plasmon‐Enhanced Photocatalysis

open access: yesAdvanced Science, EarlyView.
Chiral plasmonic printing uses CPL as a chiral stimulus to guide asymmetric Ag deposition on chiral Au NRs seeds. Under CPL, hot carriers reduce Ag+, and the process is controlled by both the CPL handedness and the chiral SPR. L‐dc Au generates more hot electrons under LCP than RCP, while D‐dc Au shows the opposite, revealing chiral photon‐selective ...
Yaxin Cao   +6 more
wiley   +1 more source

A Unified Flash Memory Platform for Mode‐Adaptive and Robust AI Computation

open access: yesAdvanced Science, EarlyView.
A unified AND‐type flash memory platform enables both transistor‐mode and capacitor‐mode computing‐in‐memory operations within the same device structure. By selectively switching the sensing mode through peripheral reconfiguration, the platform provides adaptable trade‐offs between computational accuracy, robustness, and energy efficiency for AI ...
Dayeon Yu   +6 more
wiley   +1 more source

QuantNAS for Super Resolution: Searching for Efficient Quantization-Friendly Architectures Against Quantization Noise

open access: yesIEEE Access
This work aims to develop an automated procedure for discovering new, efficient solutions that can be effectively quantized in mixed-precision mode with minimal degradation.
Egor Shvetsov   +6 more
doaj   +1 more source

Home - About - Disclaimer - Privacy