12,841 Hits in 5.0 sec

Semi-Supervised Clustering with Contrastive Learning for Discovering New Intents [article]

Feng Wei, Zhenbo Chen, Zhenghong Hao, Fengxin Yang, Hua Wei, Bing Han, Sheng Guo
2022 arXiv   pre-print
In this paper, we propose Deep Contrastive Semi-supervised Clustering (DCSC), which aims to cluster text samples in a semi-supervised way and provide grouped intents to operation staff.  ...  Considering that most scenarios have few intents known already and most intents waiting to be discovered, we focus on semi-supervised text clustering and try to make the proposed method benefit from labeled  ...  CONCLUSION In this paper, we propose Deep Contrastive Semi-supervised Clustering (DCSC), which is for discovering new intents from raw user queries.  ... 
arXiv:2201.07604v1 fatcat:l2p3ck2fjzaqbjn7aohdn6isry

Semi-Supervised Class Discovery [article]

Jeremy Nixon, Jeremiah Liu, David Berthelot
2020 arXiv   pre-print
We show that our class discovery system can be successfully applied to vision and language, and we demonstrate the value of semi-supervised learning in automatically discovering novel classes.  ...  We apply a new heuristic, class learnability, for deciding whether a class is worthy of addition to the training dataset.  ...  In contrast with conventional semi-supervised learning, we consider X u ∈ OOD := (x l+1 ∈ OOD, ..., x l+u ∈ OOD) for which the data is not necessarily in the same domain as the X l , and a correct existing  ... 
arXiv:2002.03480v2 fatcat:fg6gjs75drhmbdmk5bkhu2s5wu

Discovering New Intents with Deep Aligned Clustering [article]

Hanlei Zhang, Hua Xu, Ting-En Lin, Rui Lyu
2021 arXiv   pre-print
They also have difficulties in providing high-quality supervised signals to learn clustering-friendly features for grouping unlabeled intents.  ...  In this work, we propose an effective method, Deep Aligned Clustering, to discover new intents with the aid of the limited known intent data.  ...  ., China Joint Research Center for Industrial Intelligence and Internet of Things.  ... 
arXiv:2012.08987v7 fatcat:w444n65osjdldjds6qhkqresge

Automatic Discovery of Novel Intents Domains from Text Utterances [article]

Nikhita Vedula, Rahul Gupta, Aman Alok, Mukund Sridhar
2020 arXiv   pre-print
It learns discriminative deep features to group together utterances and discover multiple latent intent categories within them in an unsupervised manner.  ...  Most existing research formulates this as a supervised classification problem with a closed-world assumption, i.e. the domains or intents to be identified are pre-defined or known beforehand.  ...  We then transfer knowledge learned from intents seen during training to the unlabeled data containing novel intents.  ... 
arXiv:2006.01208v1 fatcat:muaiqdi425hw5bjiibv5u7ddkq

Discovering New Intents via Constrained Deep Adaptive Clustering with Cluster Refinement [article]

Ting-En Lin, Hua Xu, Hanlei Zhang
2019 arXiv   pre-print
Identifying new user intents is an essential task in the dialogue system.  ...  Moreover, we refine the clusters by forcing the model to learn from the high confidence assignments.  ...  Introduction Discovering new user intents that have not been met is an important task in the dialogue system.  ... 
arXiv:1911.08891v1 fatcat:uvaea6pvofd7rdjlfjxojf6ylq

Discovering New Intents via Constrained Deep Adaptive Clustering with Cluster Refinement

Ting-En Lin, Hua Xu, Hanlei Zhang
Identifying new user intents is an essential task in the dialogue system.  ...  Moreover, we refine the clusters by forcing the model to learn from the high confidence assignments.  ...  Introduction Discovering new user intents that have not been met is an important task in the dialogue system.  ... 
doi:10.1609/aaai.v34i05.6353 fatcat:hscnyqovwbdrpppmixj6xbozae

Semi-Supervised Learning Approach to Discover Enterprise User Insights from Feedback and Support [article]

Xin Deng, Ross Smith, Genevieve Quintin
2020 arXiv   pre-print
In this paper, we proposed and developed an innovative Semi-Supervised Learning approach by utilizing Deep Learning and Topic Modeling to have a better understanding of the user voice.This approach combines  ...  a BERT-based multiclassification algorithm through supervised learning combined with a novel Probabilistic and Semantic Hybrid Topic Inference (PSHTI) Model through unsupervised learning, aiming at automating  ...  We have a special callout to the Microsoft feedback team for enabling us to leverage their already established feedback repository and the Outlook product group for graciously letting us implement the  ... 
arXiv:2007.09303v3 fatcat:4khr5mtobzgw7cergjc7xktnne

Social media intention mining for sustainable information systems: categories, taxonomy, datasets and challenges

Ayesha Rashid, Muhammad Shoaib Farooq, Adnan Abid, Tariq Umer, Ali Kashif Bashir, Yousaf Bin Zikria
2021 Complex & Intelligent Systems  
This area has been consistently getting pertinence with an increasing trend for online purchasing. Noticeable research work has been accomplished in this area for the last two decades.  ...  The analysis reveals that there exist eight prominent categories of intention. Furthermore, a taxonomy of the approaches and techniques used for intention mining have been discussed in this article.  ...  distribution Framework for extract- ing intention from Why-type questions 1 1 1 1 1 1 1 0 7 [106] Intentional clusters 2015 News headlines Semi-supervised Clustering by input method 1  ... 
doi:10.1007/s40747-021-00342-9 fatcat:ak3y4ao2sbffjd5b3rbttidvjy

CSS-LM: A Contrastive Framework for Semi-supervised Fine-tuning of Pre-trained Language Models [article]

Yusheng Su, Xu Han, Yankai Lin, Zhengyan Zhang, Zhiyuan Liu, Peng Li, Jie Zhou, Maosong Sun
2021 arXiv   pre-print
To address this issue, we introduce a novel framework (named "CSS-LM") to improve the fine-tuning phase of PLMs via contrastive semi-supervised learning.  ...  We then perform contrastive semi-supervised learning on both the retrieved unlabeled and original labeled instances to help PLMs capture crucial task-related semantic features.  ...  CSS-LM learns semantic features by contrastive semi-supervised learning.  ... 
arXiv:2102.03752v3 fatcat:bv6kzaqcwfh7rcdxb4at4qsdi4

Open-World Semi-Supervised Learning [article]

Kaidi Cao, Maria Brbic, Jure Leskovec
2022 arXiv   pre-print
Here, we introduce a novel open-world semi-supervised learning setting that formalizes the notion that novel classes may appear in the unlabeled test data.  ...  A fundamental limitation of applying semi-supervised learning in real-world settings is the assumption that unlabeled test data contains only classes previously encountered in the labeled training data  ...  ACKNOWLEDGEMENTS The authors thank Kexin Huang, Hongyu Ren, Yusuf Roohani, Camilo Ruiz, Pranay Reddy Samala, Tailin Wu and Michael Zhang for their feedback on our manuscript.  ... 
arXiv:2102.03526v3 fatcat:dv5eqh255bgexkqiaqp3eayi64

Dialog Intent Induction with Deep Multi-View Clustering [article]

Hugh Perkins, Yi Yang
2020 arXiv   pre-print
multi-view clustering techniques for inducing the dialog intent.  ...  In particular, we propose alternating-view k-means (AV-KMEANS) for joint multi-view representation learning and clustering analysis.  ...  We thank Willie Chang for his post-publication bug reports.  ... 
arXiv:1908.11487v2 fatcat:3dyxwifaknc6rbbja23nuwh7u4

Unsupervised Semantic Segmentation by Contrasting Object Mask Proposals [article]

Wouter Van Gansbeke, Simon Vandenhende, Stamatios Georgoulis, Luc Van Gool
2021 arXiv   pre-print
In this paper, we make a first attempt to tackle the problem on datasets that have been traditionally utilized for the supervised case.  ...  Experimental evaluation shows that our method comes with key advantages over existing works. First, the learned pixel embeddings can be directly clustered in semantic groups using K-Means on PASCAL.  ...  Semi-Supervised Learning This section describes the semi-supervised learning setup. In each case, we report the average result for three randomly sampled splits. ImageNet Pre-Trained Baseline.  ... 
arXiv:2102.06191v3 fatcat:6kbodv6wbjblzorfimrs4hz5v4

Discovering Hypernymy in Text-Rich Heterogeneous Information Network by Exploiting Context Granularity [article]

Yu Shi, Jiaming Shen, Yuchen Li, Naijing Zhang, Xinwei He, Zhengzhi Lou, Qi Zhu, Matthew Walker, Myunghwan Kim, Jiawei Han
2019 arXiv   pre-print
HyperMine learns this model using weak supervision acquired based on high-precision textual patterns.  ...  Existing methods of hypernymy discovery either leverage textual patterns to extract explicitly mentioned hypernym-hyponym pairs, or learn a distributional representation for each term of interest based  ...  Mining t1 = Literature Mining t1 = Graph Mining t1 = Supervised Learning t1 = Unsupervised Learning t1 = Semi-supervised Learning t1 = Reinforcement Learning t2 = Data mining t2 = Learning algorithm t2  ... 
arXiv:1909.01584v1 fatcat:c6f4uokevreufhnp6256zssymm

Sequence clustering and labeling for unsupervised query intent discovery

Jackie Chi Kit Cheung, Xiao Li
2012 Proceedings of the fifth ACM international conference on Web search and data mining - WSDM '12  
We present an unsupervised method for clustering queries with similar intent and for producing a pattern consisting of a sequence of semantic concepts and/or lexical items for each intent.  ...  Furthermore, we leverage the discovered intent patterns to automatically annotate a large number of queries beyond those used in clustering.  ...  ACKNOWLEDGMENTS The authors would like to thank Alex Acero and Ye-Yi Wang for useful discussions.  ... 
doi:10.1145/2124295.2124342 dblp:conf/wsdm/CheungL12 fatcat:abor24tep5f5laokcnfqon7yhy

Intent Discovery Through Unsupervised Semantic Text Clustering

A Padmasundari, Srinivas Bangalore
2018 Interspeech 2018  
Our experiments on public datasets demonstrate the effectiveness of our approach generating homogeneous clusters with 89% cluster accuracy, leading to better semantic intent alignments.  ...  While classification methods that rely on labeled data are often used for SLU, creating large supervised data sets is extremely tedious and time consuming.  ...  Designing unsupervised or semi-supervised solution to cluster and discover intents for dialogs may be another direction to look at.  ... 
doi:10.21437/interspeech.2018-2436 dblp:conf/interspeech/PadmasundariB18 fatcat:xpojukqy7zcmxhbl4ujauo3gty
« Previous Showing results 1 — 15 out of 12,841 results