Filters








1,008 Hits in 4.9 sec

Disaster mapping from satellites: damage detection with crowdsourced point labels [article]

Danil Kuzin, Olga Isupova, Brooke D. Simmons, Steven Reece
2021 arXiv   pre-print
This work presents methods for aggregating potentially inconsistent damage marks to train a neural network damage detector.  ...  However, a combination of crowdsourcing and recent advances in deep learning reduces the effort needed to just a few hours in real time.  ...  Damage Detection In this section we demonstrate the viability of our approach for building a training dataset for damage detection from point crowdsourced marks by training a neural net on this dataset  ... 
arXiv:2111.03693v1 fatcat:6f6gnf6ejzh6tmsptwtioflcra

SOMOS: The Samsung Open MOS Dataset for the Evaluation of Neural Text-to-Speech Synthesis [article]

Georgia Maniati, Alexandra Vioni, Nikolaos Ellinas, Karolos Nikitaras, Konstantinos Klapsas, June Sig Sung, Gunu Jho, Aimilios Chalamandaris, Pirros Tsiakoulis
2022 arXiv   pre-print
We collect MOS naturalness evaluations on 3 English Amazon Mechanical Turk locales and share practices leading to reliable crowdsourced annotations for this task.  ...  It can be employed to train automatic MOS prediction systems focused on the assessment of modern synthesizers, and can stimulate advancements in acoustic model evaluation.  ...  honestly, which has been demonstrated to improve quality of crowdsourced responses [35] .  ... 
arXiv:2204.03040v1 fatcat:qpbs4bpn4zhr3ch7epfdsztbuq

How to Tell Ancient Signs Apart? Recognizing and Visualizing Maya Glyphs with CNNs

Gülcan Can, Jean-Marc Odobez, Daniel Gatica-Perez
2018 ACM Journal on Computing and Cultural Heritage  
In this context, this paper assesses three different Convolutional Neural Network (CNN) architectures along with three learning approaches to train them for hieroglyph classification, which is a very challenging  ...  Sketch-a-Net), and the recent Residual Networks. The sketch-specific model trained from scratch outperforms other models and training strategies.  ...  As such, it does not require modifying the CNN model to visualize the activation maps, and it can be applied to any type of neural networks, even to pretrained ones without the need of re-training.  ... 
doi:10.1145/3230670 fatcat:djw46etqgrgadh6ielgtdd662u

Learning From Graph Neighborhoods Using LSTMs [article]

Rakshit Agrawal, Luca de Alfaro, Vassilis Polychronopoulos
2016 arXiv   pre-print
The approach is based on a multi-level architecture built from Long Short-Term Memory neural nets (LSTMs); the LSTMs learn how to summarize the neighborhood from data.  ...  We demonstrate the effectiveness of the proposed technique on a synthetic example and on real-world data related to crowdsourced grading, Bitcoin transactions, and Wikipedia edit reversions.  ...  instance, both for neural nets and for LSTMs.  ... 
arXiv:1611.06882v1 fatcat:7g64abtvvzhptlmoafpdtytjyi

Deep learning from crowds [article]

Filipe Rodrigues, Francisco Pereira
2017 arXiv   pre-print
Then, a novel general-purpose crowd layer is proposed, which allows us to train deep neural networks end-to-end, directly from the noisy labels of multiple annotators, using only backpropagation.  ...  Recently, crowdsourcing has established itself as an efficient and cost-effective solution for labeling large sets of data in a scalable manner, but it often requires aggregating labels from multiple noisy  ...  Acknowledgments The research leading to these results has received funding from the People Programme (Marie Curie Actions) of the European Union's Seventh Framework Programme (FP7/2007(FP7/ -2013 under  ... 
arXiv:1709.01779v2 fatcat:zlhdbbum6rfb3blllk5fxsb4ce

Moderating with the Mob: Evaluating the Efficacy of Real-Time Crowdsourced Fact-Checking

William Godel, Zeve Sanderson, Kevin Aslett, Jonathan Nagler, Richard Bonneau, Nathaniel Persily, Joshua Tucker
2021 Journal of Online Trust and Safety  
Research on the "wisdom of the crowds" suggests one possible solution: aggregating the evaluations of ordinary users to assess the veracity of information.  ...  Although we find that machine learning-based models using the crowd perform better at identifying false news than simple aggregation rules, our results suggest that neither approach is able to perform  ...  We evaluate three algorithms -elastic net, random forest, and neural networks (NNs) 29 -using these features.  ... 
doi:10.54501/jots.v1i1.15 fatcat:xafm52rvtzhsfcx5rwbqzkw2ti

Deep Learning Relevance: Creating Relevant Information (as Opposed to Retrieving it) [article]

Christina Lioma and Birger Larsen and Casper Petersen and Jakob Grue Simonsen
2016 arXiv   pre-print
Given a query, we train a Recurrent Neural Network (RNN) on existing relevant information to that query.  ...  We design a crowdsourcing experiment to assess how relevant the "deep learned" document is, compared to existing relevant documents.  ...  SYNTHESISING RELEVANT INFORMA-TION WITH RECURRENT NEURAL NET-WORKS Given a query and a set of relevant documents to that query, we ask whether a new synthetic document can be created automatically, which  ... 
arXiv:1606.07660v2 fatcat:ymqxcsfwnvcsdfzneagojaneuq

Neural Networks Assist Crowd Predictions in Discerning the Veracity of Emotional Expressions [article]

Zhenyue Qin and Tom Gedeon and Sabrina Caldwell
2018 arXiv   pre-print
Constraining data to best performers can further increase the result up to 92%. Neural networks can achieve an accuracy to 99.69% by aggregating participants' answers.  ...  Furthermore, neural networks that are trained with one emotion data can also produce high accuracies on discerning the veracity of other emotion types: our crowdsourced transfer of emotion learning is  ...  Acknowledgments The authors are grateful to Aaron Manson for access to the data, and particularly acknowledge his efforts in data collection.  ... 
arXiv:1808.05359v1 fatcat:x7ghgqr4fffazce2bpbb3gc47e

MultiScene: A Large-scale Dataset and Benchmark for Multi-scene Recognition in Single Aerial Images [article]

Yuansheng Hua, Lichao Mou, Pu Jin, Xiao Xiang Zhu
2021 arXiv   pre-print
To facilitate progress, we make our dataset and trained models available on https://gitlab.lrz.de/ai4eo/reasoning/multiscene.  ...  Considering that manually labeling such images is extremely arduous, we resort to low-cost annotations from crowdsourcing platforms, e.g., OpenStreetMap (OSM).  ...  For the latter, we leverage the same test set but train deep neural networks on the other 93,000 images with only crowdsourced annotations. Evaluation.  ... 
arXiv:2104.02846v3 fatcat:vlsv27qebzbgvoai4z454qrrfe

Analytics and Evolving Landscape of Machine Learning for Emergency Response [chapter]

Rajendra Akerkar Minsung Hong
2019 Zenodo  
The purpose of this chapter is to discuss a hybrid crowdsourcing and real-time ma- chine learning approaches to rapidly process large volumes of data for emergency response in a time-sensitive manner.We  ...  In particular, this chapter also focuses on crowdsourcing approaches with machine learning to achieve better understanding and decision support during a disaster, and we discusses the issues on the approaches  ...  , HURDAT2 b -Labeling Neural Networks to Predict Earthquakes in Chile [20] Earthquake Prediction Artificial Neural Net- work Earthquake of magni- tude equal or larger to 3.0 Chile's National  ... 
doi:10.5281/zenodo.5106014 fatcat:ui7kmryflbaljkim4u2ync2dqi

Reply & Supply: Efficient crowdsourcing when workers do more than answer questions

Thomas C. McAndrew, Elizaveta A. Guseva, James P. Bagrow, Zhong-Ke Gao
2017 PLoS ONE  
experience and creativity to provide new and unexpected information to the crowdsourcer.  ...  Here we study how to perform efficient crowdsourcing with such growing question sets.  ...  However, for sufficiently large numbers of workers, the average response is always going to be the primary concern, particularly in most crowdsourcing tasks which need to aggregate multiple worker responses  ... 
doi:10.1371/journal.pone.0182662 pmid:28806413 pmcid:PMC5555646 fatcat:4tqv6xdpxbfwjfphsxptrd6a6y

PoCoNet: Better Speech Enhancement with Frequency-Positional Embeddings, Semi-Supervised Conversational Data, and Biased Loss [article]

Umut Isik, Ritwik Giri, Neerad Phansalkar, Jean-Marc Valin, Karim Helwani, Arvindh Krishnaswamy
2020 arXiv   pre-print
We introduce several innovations that lead to better large neural networks for speech enhancement.  ...  use cases beyond what is encountered in training data.  ...  Architecture For the neural model N , we start with a fully-convolutional 2D U-Net architecture with self-attention layers and 4-layer DenseNet blocks at each level, similar to [17] .  ... 
arXiv:2008.04470v1 fatcat:d3xh4xvzcrbh3l6ltl7f2cnynu

PoCoNet: Better Speech Enhancement with Frequency-Positional Embeddings, Semi-Supervised Conversational Data, and Biased Loss

Umut Isik, Ritwik Giri, Neerad Phansalkar, Jean-Marc Valin, Karim Helwani, Arvindh Krishnaswamy
2020 Interspeech 2020  
We introduce several innovations that lead to better large neural networks for speech enhancement.  ...  use cases beyond what is encountered in training data.  ...  Architecture For the neural model N , we start with a fully-convolutional 2D U-Net architecture with self-attention layers and 4-layer DenseNet blocks at each level, similar to [17] .  ... 
doi:10.21437/interspeech.2020-3027 dblp:conf/interspeech/IsikGPVHK20 fatcat:2pbpgh6mojhv3f5bihpmy42vre

Transparent Machine Education of Neural Networks for Swarm Shepherding Using Curriculum Design [article]

Alexander Gee, Hussein Abbass
2019 arXiv   pre-print
The shepherd needs to deal with complex and dynamic environments and make decisions in order to direct the swarm from one location to another.  ...  Swarm control is a difficult problem due to the need to guide a large number of agents simultaneously.  ...  , to be presented as training samples to the supervised learning AIES neural nets.  ... 
arXiv:1903.09297v1 fatcat:ks7rtgwntfg7pp53rmodidhzxm

VRBagged-Net: Ensemble Based Deep Learning Model for Disaster Event Classification

Muhammad Hanif, Muhammad Atif Tahir, Muhammad Rafi
2021 Electronics  
The framework utilizes the deep learning models Visual Geometry Group (VGG) and Residual Network (ResNet), along with the technique of Bootstrap aggregating (Bagging).  ...  Various disaster-based datasets were selected for the validation of the VRBagged-Net framework.  ...  An appropriate response to flooding requires a timely and accurate flow of information from the affected area to the responsible organizations.  ... 
doi:10.3390/electronics10121411 fatcat:bqma5u5bj5cbxicyltepf7k4km
« Previous Showing results 1 — 15 out of 1,008 results