Filters








106,686 Hits in 6.1 sec

Diversity and Consistency: Exploring Visual Question-Answer Pair Generation

Sen Yang, Qingyu Zhou, Dawei Feng, Yang Liu, Chao Li, Yunbo Cao, Dongsheng Li
2021 Findings of the Association for Computational Linguistics: EMNLP 2021   unpublished
It requires not only generating diverse question-answer pairs but also keeping the consistency of them.  ...  Moreover, this task can be used to improve visual question generation and visual question answering.  ...  In future works, we will explore generating consistent deep question-answer pairs. A Autoregressive Generation We adopt an autoregressive generation for sequences in this paper.  ... 
doi:10.18653/v1/2021.findings-emnlp.91 fatcat:73upzahaurhgvovdel6tkf2l5e

Automated generation of consistent, diverse and structurally realistic graph models

Oszkár Semeráth, Aren A. Babikian, Boqi Chen, Chuning Li, Kristóf Marussy, Gábor Szárnyas, Dániel Varró
2021 Journal of Software and Systems Modeling  
AbstractIn this paper, we present a novel technique to automatically synthesize consistent, diverse and structurally realistic domain-specific graph models.  ...  A graph model is (1) consistent if it is metamodel-compliant and it satisfies the well-formedness constraints of the domain; (2) it is diverse if local neighborhoods of nodes are highly different; and  ...  Acknowledgements We would like to thank all three reviewers for their detailed and insightful feedback.  ... 
doi:10.1007/s10270-021-00884-z fatcat:rx2t264z6vfuliqpwavu4b3tki

Consistent and Clear Reporting of Results from Diverse Modeling Techniques: The A3 Method

Scott Fortmann-Roe
2015 Journal of Statistical Software  
Here, a general method and an R package, A3, are presented to support the assessment and communication of the quality of a model fit along with metrics of variable importance.  ...  The presented method is accurate, robust, and adaptable to a wide range of predictive modeling algorithms. The method is described along with case studies and a usage guide.  ...  The second question is answered by the A3 package allowing inferences based solely on the predictive accuracy of models.  ... 
doi:10.18637/jss.v066.i07 fatcat:joqi46g3gjbttnzfjabwrppjgu

C3VQG: Category Consistent Cyclic Visual Question Generation [article]

Shagun Uppal, Anish Madan, Sarthak Bhagat, Yi Yu, Rajiv Ratn Shah
2021 arXiv   pre-print
Visual Question Generation (VQG) is the task of generating natural questions based on an image.  ...  In this paper, we try to exploit the different visual cues and concepts in an image to generate questions using a variational autoencoder (VAE) without ground-truth answers.  ...  Visual Question Generation (VQG) VQG is the task of developing visual understanding from images using cues from ground-truth answers and/or answer categories in order to generate relevant question.  ... 
arXiv:2005.07771v5 fatcat:numf777pq5dsdbbsvjapx44gya

Cycle-Consistency for Robust Visual Question Answering

Meet Shah, Xinlei Chen, Marcus Rohrbach, Devi Parikh
2019 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)  
In addition, our approach outperforms state-of-the-art approaches on the standard VQA and Visual Question Generation tasks on the challenging VQA v2.0 dataset.  ...  Specifically, we train a model to not only answer a question, but also generate a question conditioned on the answer, such that the answer predicted for the generated question is the same as the ground  ...  correct given a question and image pair (Q, I).  ... 
doi:10.1109/cvpr.2019.00681 dblp:conf/cvpr/ShahCRP19 fatcat:6zte5bkxknfcrn4e6eofaxwkpu

Sunny and Dark Outside?! Improving Answer Consistency in VQA through Entailed Question Generation [article]

Arijit Ray, Karan Sikka, Ajay Divakaran, Stefan Lee, Giedrius Burachas
2019 arXiv   pre-print
CTM automatically generates entailed (or similar-intent) questions for a source QA pair and fine-tunes the VQA model if the VQA's answer to the entailed question is consistent with the source QA pair.  ...  For a given observable fact in an image (e.g. the balloon's color), we generate a set of logically consistent question-answer (QA) pairs (e.g. Is the balloon red?)  ...  The views, opinions and/or findings expressed are those of the authors' and should not be interpreted as representing the official views/policies of the DoD / U.S. Govt.  ... 
arXiv:1909.04696v1 fatcat:tci3pyvm7rdonimhvdvq6yax54

Cycle-Consistency for Robust Visual Question Answering [article]

Meet Shah, Xinlei Chen, Marcus Rohrbach, Devi Parikh
2019 arXiv   pre-print
In addition, our approach outperforms state-of-the-art approaches on the standard VQA and Visual Question Generation tasks on the challenging VQA v2.0 dataset.  ...  Specifically, we train a model to not only answer a question, but also generate a question conditioned on the answer, such that the answer predicted for the generated question is the same as the ground  ...  given a question and image pair (Q, I).  ... 
arXiv:1902.05660v1 fatcat:4qkbpcns3bhfzd7xnejby5soba

Sunny and Dark Outside?! Improving Answer Consistency in VQA through Entailed Question Generation

Arijit Ray, Karan Sikka, Ajay Divakaran, Stefan Lee, Giedrius Burachas
2019 Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP)  
CTM automatically generates entailed (or similar-intent) questions for a source QA pair and fine-tunes the VQA model if the VQA's answer to the entailed question is consistent with the source QA pair.  ...  For a given observable fact in an image (e.g. the balloon's color), we generate a set of logically consistent question-answer (QA) pairs (e.g. Is the balloon red?)  ...  The views, opinions and/or findings expressed are those of the authors' and should not be interpreted as representing the official views/policies of the DoD / U.S. Govt.  ... 
doi:10.18653/v1/d19-1596 dblp:conf/emnlp/RaySDLB19 fatcat:frwhkvzperdjjgpqzvc272tgxm

Consistent Multiple Sequence Decoding [article]

Bicheng Xu, Leonid Sigal
2020 arXiv   pre-print
In this paper, we introduce a consistent multiple sequence decoding architecture, which is while relatively simple, is general and allows for consistent and simultaneous decoding of an arbitrary number  ...  More importantly, we illustrate that the decoded sentences, for the same regions, are more consistent (improvement of 9.5%), while across images and regions maintain diversity.  ...  For example, it is a critical component in a range of visual-lingual architectures, for tasks such as image captioning [19, 25, 32, 36] and question answering [2, 3, 23] , as well as in generative models  ... 
arXiv:2004.00760v2 fatcat:gb6d3cm55rc3zjjaqkwk7giwhu

ManyModalQA: Modality Disambiguation and QA over Diverse Inputs [article]

Darryl Hannan, Akshay Jain, Mohit Bansal
2020 arXiv   pre-print
We collect our data by scraping Wikipedia and then utilize crowdsourcing to collect question-answer pairs.  ...  We present a new multimodal question answering challenge, ManyModalQA, in which an agent must answer a question by considering three distinct modalities: text, images, and tables.  ...  This work was supported by DARPA MCS Grant #N66001-19-2-4031 ARO-YIP Award #W911NF-18-1-0336, an NSF PhD Fellowship, and faculty awards from Google and Facebook.  ... 
arXiv:2001.08034v1 fatcat:dhnwxiwffjbs3asu66amq7ge6a

Learning to Sketch with Shortcut Cycle Consistency [article]

Jifei Song, Kaiyue Pang, Yi-Zhe Song, Tao Xiang, Timothy Hospedales
2018 arXiv   pre-print
This means that even if photo-sketch pairs are available, they only provide weak supervision signal to learn a translation model.  ...  Compared with existing unsupervised approaches based on cycle consistency (i.e., D(E(D(E(photo)))) -> photo), we introduce a shortcut consistency enforced at the encoder bottleneck (e.g., D(E(photo)) -  ...  We can quickly answer this question by sketching a few line strokes.  ... 
arXiv:1805.00247v1 fatcat:obpsspzy7na7fh6phh2zzfcy7q

ManyModalQA: Modality Disambiguation and QA over Diverse Inputs

Darryl Hannan, Akshay Jain, Mohit Bansal
2020 PROCEEDINGS OF THE THIRTIETH AAAI CONFERENCE ON ARTIFICIAL INTELLIGENCE AND THE TWENTY-EIGHTH INNOVATIVE APPLICATIONS OF ARTIFICIAL INTELLIGENCE CONFERENCE  
We collect our data by scraping Wikipedia and then utilize crowdsourcing to collect question-answer pairs.  ...  We present a new multimodal question answering challenge, ManyModalQA, in which an agent must answer a question by considering three distinct modalities: text, images, and tables.  ...  This work was supported by DARPA MCS Grant #N66001-19-2-4031 ARO-YIP Award #W911NF-18-1-0336, an NSF PhD Fellowship, and faculty awards from Google and Facebook.  ... 
doi:10.1609/aaai.v34i05.6294 fatcat:ssdj5moa5vhrfasqmvypx4mcpu

Learning to Sketch with Shortcut Cycle Consistency

Jifei Song, Kaiyue Pang, Yi-Zhe Song, Tao Xiang, Timothy M. Hospedales
2018 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition  
We can quickly answer this question by sketching a few line strokes.  ...  For decades, researchers in computer vision have dedicated themselves to answering this question, by injecting intelligence and supervision into the machine with the hope of seeing better.  ... 
doi:10.1109/cvpr.2018.00090 dblp:conf/cvpr/SongPSXH18 fatcat:rc73xztlvrejzpnzb3ugiff62i

Nation Formation and Genetic Diversity

Klaus Desmet, Michel Le Breton, Ignacio Ortuño Ortín, Shlomo Weber
2006 Social Science Research Network  
By using data on genetic distances, we examine the stability of the current map of Europe and identify the regions prone to secession and the countries that are more likely to merge.  ...  This tradeoff induces agents' preferences over different geographical conÞgurations, thus determining the likelihood of secession and uniÞcation.  ...  Each question has q different possible answers and we denote by x i,j = ( x 1 i,j , x 2 i,j , ...x q i,j ) the vector of relative answers to question i in nation j .  ... 
doi:10.2139/ssrn.949720 fatcat:a3vw2eok75cxnluuhcojpbb7yu

Improving Generative Visual Dialog by Answering Diverse Questions

Vishvak Murahari, Prithvijit Chattopadhyay, Dhruv Batra, Devi Parikh, Abhishek Das
2019 Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP)  
i.e. be exposed to more visual concepts to talk about, and varied questions to answer.  ...  Prior work on training generative Visual Dialog models with reinforcement learning (Das et al., 2017b) has explored a Q-BOT-A-BOT image-guessing game and shown that this 'self-talk' approach can lead to  ...  Acknowledgements We thank Nirbhay Modhe and Viraj Prabhu for the PyTorch implementation (Modhe et al., 2018) of Das et al. (2017b) that we built on, and Jiasen Lu for helpful discussions.  ... 
doi:10.18653/v1/d19-1152 dblp:conf/emnlp/MurahariCBPD19 fatcat:35dt2n4qa5govetclxcjw63kqe
« Previous Showing results 1 — 15 out of 106,686 results