On-device research index

arXiv research

A locally-built, LLM-digested index of recent arXiv papers in quant finance, geometry/topology, and statistical ML — keyword search served straight from SQLite on this machine.

168,742 papers · 148 categories

Trend · papers per month

count % of corpus

4 results for “training-resistant”

OGAN: Disrupting Deepfakes with an Adversarial Attack that Survives Trainingcs.CV

OGAN attacks autoencoders to prevent deepfake creation.

problem Preventing deepfake creation using adversarial attacks.

method Training-resistant adversarial attack on face-swapping autoencoders.

result OGAN attacks are training-resistant and transfer across different models and faces.

adversarial attacks deepfakes face-swapping autoencoders training-resistant

2020-06-17abs ↗pdf ↗

Semi-Implicit Generative Modelstat.ML

To combine explicit and implicit generative models, we introduce semi-implicit generator (SIG) as a flexible hierarchical model that can be trained in the maximum likelihood framework. Both theoretically and experimentally, we demonstrate that SIG can generate high quality samples especially when dealing with multi-mod…

2019-05-29abs ↗pdf ↗

Kolmogorov-Smirnov GANstat.ML

KSGAN uses KS distance for deep generative modeling.

problem Deep generative modeling challenges, especially for multivariate distributions.

method Formulates adversarial training as minimization of KS distance, using quantile function as critic.

result KSGAN trained distributions closely match target distributions.

Generative Adversarial Networks Kolmogorov-Smirnov distance Deep learning Multivariate distributions Quantile function

2024-06-28abs ↗pdf ↗

TxSim:Modeling Training of Deep Neural Networks on Resistive Crossbar Systemscs.LG

TxSim models DNN training on resistive crossbars, improving accuracy and feasibility.

problem Non-idealities in resistive crossbar systems degrade DNN training accuracy.

method TxSim is a fast, customizable framework that models all training operations, reducing simulation time with minimal accuracy loss.

result TxSim reveals substantial (3%-10%) accuracy degradation due to non-idealities in large-scale DNNs.

DNN training resistive crossbars non-idealities accuracy degradation simulation speed

2020-02-25abs ↗pdf ↗