Multi-Response Awareness for Retrieval-Based Conversations: Respond with Diversity via Dynamic Representation Learning

2021 ◽  
Vol 39 (4) ◽  
pp. 1-29
Author(s):  
Rui Yan ◽  
Weiheng Liao ◽  
Dongyan Zhao ◽  
Ji-Rong Wen

Conversational systems now attract great attention due to their promising potential and commercial values. To build a conversational system with moderate intelligence is challenging and requires big (conversational) data, as well as interdisciplinary techniques. Thanks to the prosperity of the Web, the massive data available greatly facilitate data-driven methods such as deep learning for human-computer conversational systems. In general, retrieval-based conversational systems apply various matching schema between query utterances and responses, but the classic retrieval paradigm suffers from prominent weakness for conversations: the system finds similar responses given a particular query. For real human-to-human conversations, on the contrary, responses can be greatly different yet all are possibly appropriate. The observation reveals the diversity phenomenon in conversations. In this article, we ascribe the lack of conversational diversity to the reason that the query utterances are statically modeled regardless of candidate responses through traditional methods. To this end, we propose a dynamic representation learning strategy that models the query utterances and different response candidates in an interactive way. To be more specific, we propose a Respond-with-Diversity model augmented by the memory module interacting with both the query utterances and multiple candidate responses. Hence, we obtain dynamic representations for the input queries conditioned on different response candidates. We frame the model as an end-to-end learnable neural network. In the experiments, we demonstrate the effectiveness of the proposed model by achieving a good appropriateness score and much better diversity in retrieval-based conversations between humans and computers.

2020 ◽  
Vol 34 (04) ◽  
pp. 4132-4139
Author(s):  
Huiting Hong ◽  
Hantao Guo ◽  
Yucheng Lin ◽  
Xiaoqing Yang ◽  
Zang Li ◽  
...  

In this paper, we focus on graph representation learning of heterogeneous information network (HIN), in which various types of vertices are connected by various types of relations. Most of the existing methods conducted on HIN revise homogeneous graph embedding models via meta-paths to learn low-dimensional vector space of HIN. In this paper, we propose a novel Heterogeneous Graph Structural Attention Neural Network (HetSANN) to directly encode structural information of HIN without meta-path and achieve more informative representations. With this method, domain experts will not be needed to design meta-path schemes and the heterogeneous information can be processed automatically by our proposed model. Specifically, we implicitly represent heterogeneous information using the following two methods: 1) we model the transformation between heterogeneous vertices through a projection in low-dimensional entity spaces; 2) afterwards, we apply the graph neural network to aggregate multi-relational information of projected neighborhood by means of attention mechanism. We also present three extensions of HetSANN, i.e., voices-sharing product attention for the pairwise relationships in HIN, cycle-consistency loss to retain the transformation between heterogeneous entity spaces, and multi-task learning with full use of information. The experiments conducted on three public datasets demonstrate that our proposed models achieve significant and consistent improvements compared to state-of-the-art solutions.


Author(s):  
Yung-An Hsieh ◽  
Yichang (James) Tsai

Raveling is one of the most common asphalt pavement distresses. The survey of its condition is required for transportation agencies to ensure roadway safety and appropriately apply preservation and rehabilitation treatments. However, the traditional raveling condition survey, including the determination of the raveling severity, is typically manually conducted by in-field visual inspection methods that are time consuming, labor intensive, and error prone. Although automated raveling detection and severity classification models have been developed, these existing models have shortcomings. Therefore, there is an urgent need to develop a more accurate and reliable model to automatically detect and classify raveling. This study proposes the first convolutional neural network (CNN)-based model for automated raveling detection and classification. Compared with general CNNs, the proposed model combines the data-driven features learned from training data and macrotexture features of 3D pavement surface data to achieve better performance. The proposed model was evaluated and compared with existing machine learning models using real-world 3D pavement surface data collected from the state of Georgia, U.S. By combining data-driven features with macrotexture features, the proposed model achieved the highest accuracy of 90.8% on raveling classification. The proposed model also achieved classification precision and recall higher than 85% for all raveling severity levels, which is more accurate and robust than existing models. It is concluded that, with multi-type features extraction and proper model design, the proposed model can provide more accurate and reliable predictions for raveling detection and classification.


Author(s):  
Jinhyuk Lee ◽  
Hyunjae Kim ◽  
Miyoung Ko ◽  
Donghee Choi ◽  
Jaehoon Choi ◽  
...  

Personal names tend to have many variations differing from country to country. Though there exists a large amount of personal names on the Web, nationality prediction solely based on names has not been fully studied due to its difficulties in extracting subtle character level features. We propose a recurrent neural network based model which predicts nationalities of each name using automatic feature extraction. Evaluation of Olympic record data shows that our model achieves greater accuracy than previous feature based approaches in nationality prediction tasks. We also evaluate our proposed model and baseline models on name ethnicity classification task, again achieving better or comparable performances. We further investigate the effectiveness of character embeddings used in our proposed model.


Author(s):  
Ying Song ◽  
Shuangjia Zheng ◽  
Zhangming Niu ◽  
Zhang-hua Fu ◽  
Yutong Lu ◽  
...  

Constructing proper representations of molecules lies at the core of numerous tasks such as molecular property prediction and drug design. Graph neural networks, especially message passing neural network (MPNN) and its variants, have recently made remarkable achievements in molecular graph modeling. Albeit powerful, the one-sided focuses on atom (node) or bond (edge) information of existing MPNN methods lead to the insufficient representations of the attributed molecular graphs. Herein, we propose a Communicative Message Passing Neural Network (CMPNN) to improve the molecular embedding by strengthening the message interactions between nodes and edges through a communicative kernel. In addition, the message generation process is enriched by introducing a new message booster module. Extensive experiments demonstrated that the proposed model obtained superior performances against state-of-the-art baselines on six chemical property datasets. Further visualization also showed better representation capacity of our model.


Sensors ◽  
2020 ◽  
Vol 20 (12) ◽  
pp. 3535 ◽  
Author(s):  
Qian Yan ◽  
Baohua Yang ◽  
Wenyan Wang ◽  
Bing Wang ◽  
Peng Chen ◽  
...  

Scab, frogeye spot, and cedar rust are three common types of apple leaf diseases, and the rapid diagnosis and accurate identification of them play an important role in the development of apple production. In this work, an improved model based on VGG16 is proposed to identify apple leaf diseases, in which the global average poling layer is used to replace the fully connected layer to reduce the parameters and a batch normalization layer is added to improve the convergence speed. A transfer learning strategy is used to avoid a long training time. The experimental results show that the overall accuracy of apple leaf classification based on the proposed model can reach 99.01%. Compared with the classical VGG16, the model parameters are reduced by 89%, the recognition accuracy is improved by 6.3%, and the training time is reduced to 0.56% of that of the original model. Therefore, the deep convolutional neural network model proposed in this work provides a better solution for the identification of apple leaf diseases with higher accuracy and a faster convergence speed.


Author(s):  
Raed Waheed Kadhim ◽  
Methaq Talib Gaata

<span>Cross-site scripting (XSS) is today one of the biggest threatthat could targeting the Web application. Based on study published by the open web applications security project (OWASP), XSS vulnerability has been present among the TOP 10 Web application vulnerabilities.Still,an important security-related issue remains how to effectively protect web applications from XSS attacks.In first part of this paper, a method for detecting XSS attack was proposed by combining </span><span lang="EN-GB">convolutional</span><span> neural network (CNN) with long short term memories<strong> (</strong>LSTM), Initially, pre-processing was applied to XSS Data Set by decoding, generalization and tokanization, and then word2vec was applied to convert words into word vectors in XSS payloads. And then we use the combination CNN with LSTM to train and test word vectors to produce a model that can be used in a web application. Based on the obtaned results, it is observed that the proposed model achevied an excellent result with accuracy of 99.4%.</span>


2020 ◽  
Author(s):  
Mikołaj Morzy ◽  
Bartłomiej Balcerzak ◽  
Adam Wierzbicki ◽  
Adam Wierzbicki

BACKGROUND With the rapidly accelerating spread of dissemination of false medical information on the Web, the task of establishing the credibility of online sources of medical information becomes a pressing necessity. The sheer number of websites offering questionable medical information presented as reliable and actionable suggestions with possibly harmful effects poses an additional requirement for potential solutions, as they have to scale to the size of the problem. Machine learning is one such solution which, when properly deployed, can be an effective tool in fighting medical disinformation on the Web. OBJECTIVE We present a comprehensive framework for designing and curating of machine learning training datasets for online medical information credibility assessment. We show how the annotation process should be constructed and what pitfalls should be avoided. Our main objective is to provide researchers from medical and computer science communities with guidelines on how to construct datasets for machine learning models for various areas of medical information wars. METHODS The key component of our approach is the active annotation process. We begin by outlining the annotation protocol for the curation of high-quality training dataset, which then can be augmented and rapidly extended by employing the human-in-the-loop paradigm to machine learning training. To circumvent the cold start problem of insufficient gold standard annotations, we propose a pre-processing pipeline consisting of representation learning, clustering, and re-ranking of sentences for the acceleration of the training process and the optimization of human resources involved in the annotation. RESULTS We collect over 10 000 annotations of sentences related to selected subjects (psychiatry, cholesterol, autism, antibiotics, vaccines, steroids, birth methods, food allergy testing) for less than $7 000 employing 9 highly qualified annotators (certified medical professionals) and we release this dataset to the general public. We develop an active annotation framework for more efficient annotation of non-credible medical statements. The results of the qualitative analysis support our claims of the efficacy of the presented method. CONCLUSIONS A set of very diverse incentives is driving the widespread dissemination of medical disinformation on the Web. An effective strategy of countering this spread is to use machine learning for automatically establishing the credibility of online medical information. This, however, requires a thoughtful design of the training pipeline. In this paper we present a comprehensive framework of active annotation. In addition, we publish a large curated dataset of medical statements labelled as credible, non-credible, or neutral.


2021 ◽  
Vol 11 (4) ◽  
pp. 1829
Author(s):  
Davide Grande ◽  
Catherine A. Harris ◽  
Giles Thomas ◽  
Enrico Anderlini

Recurrent Neural Networks (RNNs) are increasingly being used for model identification, forecasting and control. When identifying physical models with unknown mathematical knowledge of the system, Nonlinear AutoRegressive models with eXogenous inputs (NARX) or Nonlinear AutoRegressive Moving-Average models with eXogenous inputs (NARMAX) methods are typically used. In the context of data-driven control, machine learning algorithms are proven to have comparable performances to advanced control techniques, but lack the properties of the traditional stability theory. This paper illustrates a method to prove a posteriori the stability of a generic neural network, showing its application to the state-of-the-art RNN architecture. The presented method relies on identifying the poles associated with the network designed starting from the input/output data. Providing a framework to guarantee the stability of any neural network architecture combined with the generalisability properties and applicability to different fields can significantly broaden their use in dynamic systems modelling and control.


2021 ◽  
Vol 25 (3) ◽  
pp. 711-738
Author(s):  
Phu Pham ◽  
Phuc Do

Link prediction on heterogeneous information network (HIN) is considered as a challenge problem due to the complexity and diversity in types of nodes and links. Currently, there are remained challenges of meta-path-based link prediction in HIN. Previous works of link prediction in HIN via network embedding approach are mainly focused on exploiting features of node rather than existing relations in forms of meta-paths between nodes. In fact, predicting the existence of new links between non-linked nodes is absolutely inconvincible. Moreover, recent HIN-based embedding models also lack of thorough evaluations on the topic similarity between text-based nodes along given meta-paths. To tackle these challenges, in this paper, we proposed a novel approach of topic-driven multiple meta-path-based HIN representation learning framework, namely W-MMP2Vec. Our model leverages the quality of node representations by combining multiple meta-paths as well as calculating the topic similarity weight for each meta-path during the processes of network embedding learning in content-based HINs. To validate our approach, we apply W-TMP2Vec model in solving several link prediction tasks in both content-based and non-content-based HINs (DBLP, IMDB and BlogCatalog). The experimental outputs demonstrate the effectiveness of proposed model which outperforms recent state-of-the-art HIN representation learning models.


Sign in / Sign up

Export Citation Format

Share Document