Filters








550 Hits in 5.6 sec

Weakly Supervised Pre-Training for Multi-Hop Retriever [article]

Yeon Seonwoo, Sang-Woo Lee, Ji-Hoon Kim, Jung-Woo Ha, Alice Oh
2021 arXiv   pre-print
To address the issue, we propose a new method for weakly supervised multi-hop retriever pre-training without human efforts.  ...  as weak supervision for pre-training, and 3) a pre-training model structure based on dense encoders.  ...  sue, we proposed a weakly supervised pre-training We use the same reader model as PathRetriever. method for multi-hop retriever, LOUVRE.  ... 
arXiv:2106.09983v1 fatcat:mhnlxpyp2rbsdisjbkfax7jpkq

Simple yet Effective Bridge Reasoning for Open-Domain Multi-Hop Question Answering [article]

Wenhan Xiong, Mo Yu, Xiaoxiao Guo, Hong Wang, Shiyu Chang, Murray Campbell, William Yang Wang
2019 arXiv   pre-print
This model, the bridge reasoner, is trained with a weakly supervised signal and produces the candidate answer passages for the passage reader to extract the answer.  ...  However, IR-based approaches are insufficient for multi-hop questions, as the topic of the second or further hops is not explicitly covered by the question.  ...  While IR works reasonably well for simple questions 1 , it often fails to retrieve the correct answer paragraph for multi-hop questions.  ... 
arXiv:1909.07597v2 fatcat:ofduwzd7tzgjtnllzvdjvh3hzu

LEPUS: Prompt-based Unsupervised Multi-hop Reranking for Open-domain QA [article]

Muhammad Khalifa, Lajanugen Logeswaran, Moontae Lee, Honglak Lee, Lu Wang
2022 arXiv   pre-print
We study unsupervised multi-hop reranking for multi-hop QA (MQA) with open-domain questions.  ...  Finally, we show that when integrated with a reader module, LEPUS can obtain competitive multi-hop QA performance, e.g., outperforming fully-supervised QA systems.  ...  Related Work Multi-hop Question Answering. The majority of approaches for multi-hop question answering rely on two main components: a retriever and a reader.  ... 
arXiv:2205.12650v1 fatcat:mzcozv3vifh2fb4cowpldvtgty

Differentiable Reasoning over a Virtual Knowledge Base [article]

Bhuwan Dhingra, Manzil Zaheer, Vidhisha Balachandran, Graham Neubig, Ruslan Salakhutdinov, William W. Cohen
2020 arXiv   pre-print
We consider the task of answering complex multi-hop questions using a corpus as a virtual knowledge base (KB).  ...  DrKIT is also very efficient, processing 10-100x more queries per second than existing multi-hop systems.  ...  E Z 0 Z 1 DIFFERENTIABLE MULTI-HOP REASONING We assume a weakly supervised setting where during training we only know the final answer entities a ∈ E for a T -hop question.  ... 
arXiv:2002.10640v1 fatcat:2fcnpovwjjfdvfl36nvhoj3u54

TriBERT: Full-body Human-centric Audio-visual Representation Learning for Visual Sound Separation [article]

Tanzila Rahman, Mengyu Yang, Leonid Sigal
2021 arXiv   pre-print
We pre-train our model on the large MUSIC21 dataset and demonstrate improved performance in audio-visual sound source separation on that dataset as well as other datasets through fine-tuning.  ...  The recent success of transformer models in language, such as BERT, has motivated the use of such architectures for multi-modal feature learning and tasks.  ...  Acknowledgments: This work was funded in part by the Vector Institute for AI, Canada CIFAR AI Chair, NSERC Canada Research Chair (CRC) and an NSERC Discovery and Discovery Accelerator Supplement Grants  ... 
arXiv:2110.13412v1 fatcat:oejb6j7hebaiflohlib76r2pae

PullNet: Open Domain Question Answering with Iterative Retrieval on Knowledge Bases and Text [article]

Haitian Sun, Tania Bedrax-Weiss, William W. Cohen
2019 arXiv   pre-print
This retrieve-and-reason process allows us to answer multi-hop questions using large KBs and corpora. PullNet is weakly supervised, requiring question-answer pairs but not gold inference paths.  ...  ., "multi-hop") reasoning.  ...  The "pull" classifier is weakly supervised , using question-answer pairs for supervision.  ... 
arXiv:1904.09537v1 fatcat:idytaora4vhnpl4f5iyju4tg44

A Unified Continuous Learning Framework for Multi-modal Knowledge Discovery and Pre-training [article]

Zhihao Fan, Zhongyu Wei, Jingjing Chen, Siyuan Wang, Zejun Li, Jiarong Xu, Xuanjing Huang
2022 arXiv   pre-print
Multi-modal pre-training and knowledge discovery are two important research topics in multi-modal machine learning.  ...  For knowledge discovery, a pre-trained model is used to identify cross-modal links on the graph.  ...  For model pre-training, we take graph structure as the knowledge for guidance. In practice, two-hop neighbors are utilized for the node representation learning.  ... 
arXiv:2206.05555v1 fatcat:wick43nxbrf5vdqb67kthnk7fq

PullNet: Open Domain Question Answering with Iterative Retrieval on Knowledge Bases and Text

Haitian Sun, Tania Bedrax-Weiss, William Cohen
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)  
This retrieve-and-reason process allows us to answer multi-hop questions using large KBs and corpora. PullNet is weakly supervised, requiring question-answer pairs but not gold inference paths.  ...  ., "multi-hop") reasoning. We describe PullNet, an integrated framework for (1) learning what to retrieve and (2) reasoning with this heterogeneous information to find the best answer.  ...  The "pull" classifier is weakly supervised, using question-answer pairs.  ... 
doi:10.18653/v1/d19-1242 dblp:conf/emnlp/SunBC19 fatcat:5n4ooeogarhh5b4j6dzi3nhb54

Exploring Phrase Grounding without Training: Contextualisation and Extension to Text-Based Image Retrieval

Letitia Parcalabescu, Anette Frank
2020 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW)  
We show that our extensions benefit the model and establish a harder, but fairer baseline for (weakly) supervised models.  ...  We also perform a stress test to assess the further applicability of such a system for creating a sentence retrieval system requiring no training nor annotated data.  ...  It is pre-trained in a multi-task fashion (masked multi-modal modelling and multimodal alignment prediction) on even larger training data and is applicable to many downstream tasks beyond image retrieval  ... 
doi:10.1109/cvprw50498.2020.00489 dblp:conf/cvpr/ParcalabescuF20 fatcat:7mktuhqifbhkjdk7dk3z5hfzda

Hypergraph Transformer: Weakly-supervised Multi-hop Reasoning for Knowledge-based Visual Question Answering [article]

Yu-Jung Heo, Eun-Sol Kim, Woo Suk Choi, Byoung-Tak Zhang
2022 arXiv   pre-print
Answering complex questions that require multi-hop reasoning under weak supervision is considered as a challenging problem since i) no supervision is given to the reasoning process and ii) high-order semantics  ...  of multi-hop knowledge facts need to be captured.  ...  Acknowledgements We would like to thank Woo Young Kang, Kyoung-Woon On, Seonil Son, Gi-Cheon Kang, Christina Baek, Junseok Park, Min Whoo Lee, Hwiyeol Jo and Sang-Woo Lee for their helpful comments and  ... 
arXiv:2204.10448v1 fatcat:pkmftox2iba43bwfrgiwmp6rcu

Hyperlink-induced Pre-training for Passage Retrieval in Open-domain Question Answering [article]

Jiawei Zhou, Xiaoguang Li, Lifeng Shang, Lan Luo, Ke Zhan, Enrui Hu, Xinyu Zhang, Hao Jiang, Zhao Cao, Fan Yu, Xin Jiang, Qun Liu (+1 others)
2022 arXiv   pre-print
To alleviate the data scarcity problem in training question answering systems, recent works propose additional intermediate pre-training for dense passage retrieval (DPR).  ...  We demonstrate that the hyperlink-based structures of dual-link and co-mention can provide effective relevance signals for large-scale pre-training that better facilitate downstream passage retrieval.  ...  For example, entity-level variants such as "Robert and Richard Sherman" vs.  ... 
arXiv:2203.06942v2 fatcat:faho7inxgzhidb3knzialwdoqe

Improving Embedded Knowledge Graph Multi-hop Question Answering by introducing Relational Chain Reasoning [article]

Weiqiang Jin, Hang Yu, Xi Tao, Ruiping Yin
2022 arXiv   pre-print
Comprehensive ablation experiments also verify the effectiveness of our method for multi-hop KGQA tasks.  ...  As a complex branchtask of KBQA, multi-hop KGQA requires reasoning over multi-hop relational chains preserved in KG to arrive at the right answer.Despite the successes made in recent years, the existing  ...  The improved method, called Pull-Net, where the "pull" classifier is weakly supervised that only utilizes QA pairs for supervision.  ... 
arXiv:2110.12679v2 fatcat:xob6jgsqgvcsbd5rfe4b5uszfy

A Survey of Knowledge-Enhanced Text Generation [article]

Wenhao Yu, Chenguang Zhu, Zaitang Li, Zhiting Hu, Qingyun Wang, Heng Ji, Meng Jiang
2022 arXiv   pre-print
The main content includes two parts: (i) general methods and architectures for integrating knowledge into text generation; (ii) specific techniques and applications according to different forms of knowledge  ...  Table 8 compares different grounded text enhanced methods from three dimensions: retrieval supervision, pre-training of the retriever, and number of stages.  ...  M3 is commonly used for multi-hop path reasoning and M4 is used for multi-hop information aggregation, except that CCM [148] only aggregates one-hop neighbors.  ... 
arXiv:2010.04389v3 fatcat:vzdtlz4j65g2va7gwkbmzyxkhq

MotifClass: Weakly Supervised Text Classification with Higher-order Metadata Information [article]

Yu Zhang, Shweta Garg, Yu Meng, Xiusi Chen, Jiawei Han
2022 arXiv   pre-print
We study the problem of weakly supervised text classification, which aims to classify text documents into a set of pre-defined categories with category surface names only and without any annotated training  ...  In this paper, we explore the potential of using metadata to help weakly supervised text classification.  ...  Xu and Dheeraj Mekala for their help with the experimental setup and anonymous reviewers for their valuable and insightful feedback.  ... 
arXiv:2111.04022v3 fatcat:cuupcigyzfhj7hjiciwonbpjzq

Deep CNN Framework for Audio Event Recognition using Weakly Labeled Web Data [article]

Anurag Kumar, Bhiksha Raj
2017 arXiv   pre-print
kind conventionally required for training detectors or classifiers.  ...  Training from these web data, however, poses several challenges, the most important being the availability of labels : labels, if any, that may be obtained for the data are generally weak, and not of the  ...  Multi-label Problem: Often several audio events are simultaneously present in an audio recording. Hence, we design our network for multi-label training and prediction.  ... 
arXiv:1707.02530v2 fatcat:xlug6lxqtnfpvoctkpqvg6hytm
« Previous Showing results 1 — 15 out of 550 results