[HTML][HTML] Self-attentive speaker embeddings for text-independent speaker verification.

Y Zhu, T Ko, D Snyder, B Mak, D Povey - Interspeech, 2018 - pianshen.com
data from 2004 to 2010 along with Mixer 6 for a total of about 63k recordings from 4.4k speakers.
The four data augmentation … increase the amount of training data and to improve the ro- …

[HTML][HTML] Open sourcing a deep learning solution for detecting nsfw images

J Mahadeokar, G Pesavento - Retrieved August, 2016 - csdndocs.com
… flipped for data augmentation, and randomly cropped to 224x224 pixels, and were then fed
to the network. For training residual networks, we used scale augmentation as described in …

[PDF][PDF] Exploring the affordances and effectiveness of a digital game in the Chinese dual language immersion classroom

F Poole, J Clarke-Midura, S Ji - Journal of Technology and Chinese …, 2022 - tclt.us
… this study two sixth-grade Chinese dual language immersion classes play a digital game in
eight sessions across four weeks. Audio data … Although the game in this study was designed …

[PDF][PDF] FLOOD MITIGATION AND LAND RECLAMATION IN HILLY RIVERSA CASE STUDY OF TOORSA RIVER, BHUTAN

RR Patra, GN Paudyal, H Enggrob… - DHI Water and …, 2006 - academia.edu
… contours and available ground survey data, with the survey data taking precedence over
the … to collate and manipulate the available ground survey data into 2D surfaces quickly and …

[PDF][PDF] A review of research on vehicle re-identification methods with unsupervised learning

Y Xu, X Guo, L Rong - J. Front. Comput. Sci. Technol, 2023 - researchgate.net
… deviation and insufficient information of data samples, the former is further divided into three
categories of style transfer, multi-view generation, and data augmentation. For the labeling …

[PDF][PDF] Review of classical recommendation algorithms

C Zhou, J Shen, Y Li, X Guo - Computer Science and Application, 2019 - pdf.hanspub.org
Recommender systems are effective tools of information filtering that are prevalent due to
continuous popularization of the Internet, personalization trends, and changing habits of …

[PDF][PDF] " Between Similitude and Dissimilitude": Supporting Family Reminiscence with Non-Photorealistic Visualizations of Past Places

J Leu - (No Title), 2016 - core.ac.uk
data of aerial survey photographs and maps as the main source of reference, because these
data … Also this type of survey data and map are more likely to be archived by public entity or …

任务型对话系统中对话状态追踪技术研究

黄伟 - 2021 - ir.lzu.edu.cn
… In the second stage, the model uses text classification methods or machine reading …
Aiming at the problem of data sparseness, a data augmentation method based on slot value …

基于数据增强和多任务学习的突发公共卫生事件谣言识别研究*

曾子明, 张瑜 - 数据分析与知识发现, 2023 - manu44.magtech.com.cn
data augmentation and multi-task learning, aiming to address the issue of unbalanced data
and insufficient labeled data … [Methods] Firstly, we extracted the text features of public health …

CCL23-Eval 任务4 系统报告: 基于深度学习的空间语义理解(System Report for CCL23-Eval Task4: Spatial Semantic Understanding Based on Deep Learning.)

CK Tan, XN Hu, XP Qiu - Proceedings of the 22nd Chinese …, 2023 - aclanthology.org
… Additionally, we model the Role Classification as a large-scale text classification task. There
… In our approach to data augmentation, we treat all fields at the same level as sentence id, …