scholarly journals Learning 2D Temporal Adjacent Networks for Moment Localization with Natural Language

2020 ◽  
Vol 34 (07) ◽  
pp. 12870-12877 ◽  
Author(s):  
Songyang Zhang ◽  
Houwen Peng ◽  
Jianlong Fu ◽  
Jiebo Luo
Keyword(s):  
Natural Language ◽  
State Of The Art ◽  
The Other ◽  
Temporal Relation ◽  
One Dimension ◽  
Single Shot ◽  
Two Dimensional ◽  
Challenging Problem ◽  
Starting Time ◽  
Temporal Moments

We address the problem of retrieving a specific moment from an untrimmed video by a query sentence. This is a challenging problem because a target moment may take place in relations to other temporal moments in the untrimmed video. Existing methods cannot tackle this challenge well since they consider temporal moments individually and neglect the temporal dependencies. In this paper, we model the temporal relations between video moments by a two-dimensional map, where one dimension indicates the starting time of a moment and the other indicates the end time. This 2D temporal map can cover diverse video moments with different lengths, while representing their adjacent relations. Based on the 2D map, we propose a Temporal Adjacent Network (2D-TAN), a single-shot framework for moment localization. It is capable of encoding the adjacent temporal relation, while learning discriminative features for matching video moments with referring expressions. We evaluate the proposed 2D-TAN on three challenging benchmarks, i.e., Charades-STA, ActivityNet Captions, and TACoS, where our 2D-TAN outperforms the state-of-the-art.

scholarly journals Planar Prior Assisted PatchMatch Multi-View Stereo

2020 ◽  
Vol 34 (07) ◽  
pp. 12516-12523
Author(s):  
Qingshan Xu ◽  
Wenbing Tao
Keyword(s):  
Graphical Model ◽  
State Of The Art ◽  
Depth Estimation ◽  
3D Models ◽  
The Other ◽  
Depth Information ◽  
Probabilistic Graphical Model ◽  
Challenging Problem ◽  
Matching Cost ◽  
Planar Regions

The completeness of 3D models is still a challenging problem in multi-view stereo (MVS) due to the unreliable photometric consistency in low-textured areas. Since low-textured areas usually exhibit strong planarity, planar models are advantageous to the depth estimation of low-textured areas. On the other hand, PatchMatch multi-view stereo is very efficient for its sampling and propagation scheme. By taking advantage of planar models and PatchMatch multi-view stereo, we propose a planar prior assisted PatchMatch multi-view stereo framework in this paper. In detail, we utilize a probabilistic graphical model to embed planar models into PatchMatch multi-view stereo and contribute a novel multi-view aggregated matching cost. This novel cost takes both photometric consistency and planar compatibility into consideration, making it suited for the depth estimation of both non-planar and planar regions. Experimental results demonstrate that our method can efficiently recover the depth information of extremely low-textured areas, thus obtaining high complete 3D models and achieving state-of-the-art performance.

scholarly journals Bilateral Multi-Perspective Matching for Natural Language Sentences

2017 ◽  
Author(s):  
Zhiguo Wang ◽  
Wael Hamza ◽  
Radu Florian
Keyword(s):  
Natural Language ◽  
State Of The Art ◽  
The State ◽  
Experimental Results ◽  
The Other ◽  
Multiple Perspectives ◽  
Time Step ◽  
Benchmark Datasets ◽  
Sentence Matching ◽  
Fully Connected

Natural language sentence matching is a fundamental technology for a variety of tasks. Previous approaches either match sentences from a single direction or only apply single granular (word-by-word or sentence-by-sentence) matching. In this work, we propose a bilateral multi-perspective matching (BiMPM) model. Given two sentences P and Q, our model first encodes them with a BiLSTM encoder. Next, we match the two encoded sentences in two directions P against Q and P against Q. In each matching direction, each time step of one sentence is matched against all time-steps of the other sentence from multiple perspectives. Then, another BiLSTM layer is utilized to aggregate the matching results into a fix-length matching vector. Finally, based on the matching vector, a decision is made through a fully connected layer. We evaluate our model on three tasks: paraphrase identification, natural language inference and answer sentence selection. Experimental results on standard benchmark datasets show that our model achieves the state-of-the-art performance on all tasks.

scholarly journals Natural Language Processing in Serious Games: A state of the art.

2015 ◽  
Vol 2 (3) ◽  
Author(s):  
Davide Picca ◽  
Dominique Jaccard ◽  
Gérald Eberlé
Keyword(s):  
Natural Language Processing ◽  
Natural Language ◽  
Language Processing ◽  
Serious Games ◽  
State Of The Art ◽  
Serious Game ◽  
The Other ◽  
Other Hand ◽  
The One ◽  
High Level

In the last decades, Natural Language Processing (NLP) has obtained a high level of success. Interactions between NLP and Serious Games have started and some of them already include NLP techniques. The objectives of this paper are twofold: on the one hand, providing a simple framework to enable analysis of potential uses of NLP in Serious Games and, on the other hand, applying the NLP framework to existing Serious Games and giving an overview of the use of NLP in pedagogical Serious Games. In this paper we present 11 serious games exploiting NLP techniques. We present them systematically, according to the following structure:  first, we highlight possible uses of NLP techniques in Serious Games, second, we describe the type of NLP implemented in the each specific Serious Game and, third, we provide a link to possible purposes of use for the different actors interacting in the Serious Game.

Multidimensional Scaling of Terms Used to Describe Personality

1970 ◽  
Vol 26 (1) ◽  
pp. 3-8
Author(s):  
Herbert Gerjuoy ◽  
Bernard S. Aaronson
Keyword(s):  
Multidimensional Scaling ◽  
Personality Traits ◽  
The Other ◽  
One Dimension ◽  
Two Dimensional ◽  
Intensity Scale ◽  
Personality Dimensions ◽  
The Mean ◽  
Highly Correlated

Buss and Gerjuoy classified 293 personality-descriptive terms into 18 personality “dimensions.” They reported “abnormality” and “intensity” scale values for each term. The present study multidimensionally scaled the highest-intensity, lowest-intensity, and lowest-abnormality terms from 6 Buss-Gerjuoy dimensions. Expert judges estimated the correlations among the personality traits referred to by these terms. The mean estimates had a two-dimensional configuration. One dimension was Abnormality; the other, Behavior Elicitation-Emission. Communications among psychologists should take into account psychologists' assumptions about trait intercorrelations. Psychologists may perceive personality descriptions as redundant if the traits specified are believed highly correlated. Individuals with trait combinations not fitting the consensual model should be explicitly so characterized.

scholarly journals A REJECTION MECHANISM IN 2D BOUNDED CONFIDENCE PROVIDES MORE CONFORMITY

2008 ◽  
Vol 11 (04) ◽  
pp. 529-549 ◽  
Author(s):  
S. HUET ◽  
G. DEFFUANT ◽  
W. JAGER
Keyword(s):  
Attitude Change ◽  
Large Range ◽  
Two Dimensions ◽  
The Other ◽  
One Dimension ◽  
Two Dimensional ◽  
Number Of Clusters ◽  
Bounded Confidence ◽  
Rejection Mechanism ◽  
Model Apply

This paper explores the dynamics of attitude change in two dimensions resulting from social interaction. We add a rejection mechanism into the 2D bounded confidence (BC) model proposed by Deffuant et al. (2001). Individuals are characterized by two-dimensional continuous attitudes, each associated with an uncertainty u, supposed constant in this first study. Individuals interact through random pairs. If their attitudes are closer than u on both dimensions, or further than u on both dimensions, or closer than u on one dimension and not further than u + δ u on the other dimension, then the rules of the BC model apply. But if their attitudes are closer than u on one dimension and further than u + δ u on the other dimension, then the individuals are in a dissonant state. They tend to solve this problem by shifting away their close attitudes. The model shows metastable clusters, which maintain themselves through opposite influences of competitor clusters. Our analysis and first experiments support the hypothesis that, for a large range of uncertainty values, the number of clusters grows linearly with the inverse of the uncertainty, whereas this growth is quadratic in the BC model.

scholarly journals Conceptual change of disaster management models: A thematic analysis

2018 ◽  
Vol 10 (1) ◽  
Author(s):  
Mehdi Nojavan ◽  
Esmail Salehi ◽  
Babak Omidvar
Keyword(s):  
Disaster Management ◽  
Thematic Analysis ◽  
The Other ◽  
One Dimension ◽  
Comprehensive Model ◽  
Two Dimensional ◽  
Final Phase ◽  
Management Models ◽  
Management Actions ◽  
The Ideal

Different models have been proposed for disaster management by researchers and agencies. Despite their efficiency in some locations, disasters are still a fundamental challenge in the way of sustainable development. The purpose of this research is developing a comprehensive conceptual model for disaster management using thematic analysis. In this regard, first, disaster management models are collected. In the next stage, the themes of each model are extracted and categorised in three phases. In the first phase that is descriptive coding, available elements in each model are extracted as code and the basic themes are recognised. Then, in the phase of interpretive coding, basic themes are classified in three categories that are called organising themes (i.e. hazard assessment, risk management and management actions). In the final phase, strategic management is selected as the global or overarching theme to integrate all the other themes. Based on thematic analysis, it can be concluded that disaster management has three main elements that are the three organising themes. Therefore, comprehensive model of disaster management should include these three elements and their sub-basic themes that is called the ideal or criterion type. Results showed that some scientists have looked at disaster management one dimensionally (one theme). Even in two-dimensional models, one dimension has advantage over the other one. While the proposed typology in this study showed that the comprehensive model should include all the three mentioned elements.

Level phase independence for GI/M/1-type Markov chains

2000 ◽  
Vol 37 (4) ◽  
pp. 984-998 ◽  
Author(s):  
Guy Latouche ◽  
P. G. Taylor
Keyword(s):  
Phase Transitions ◽  
Markov Chains ◽  
State Space ◽  
Transition Probabilities ◽  
The Other ◽  
One Dimension ◽  
Death Process ◽  
Two Dimensional ◽  
Birth And Death Process ◽  
Special Case

GI/M/1-type Markov chains make up a class of two-dimensional Markov chains. One dimension is usually called the level, and the other is often called the phase. Transitions from states in level k are restricted to states in levels less than or equal to k+1. For given transition probabilities in the interior of the state space, we show that it is always possible to define the boundary transition probabilities in such a way that the level and phase are independent under the stationary distribution. We motivate our analysis by first considering the quasi-birth-and-death process special case in which transitions from any state are restricted to states in the same, or adjacent, levels.

scholarly journals The Semantics and Collocations Relation in Food Reviews

2021 ◽  
Vol 34 (1) ◽  
Author(s):  
Fazel Keshtkar ◽  
Ledong Shi ◽  
Syed Ahmad Chan Bukhari
Keyword(s):  
Natural Language Processing ◽  
Natural Language ◽  
Computational Linguistics ◽  
Language Processing ◽  
Topic Modeling ◽  
State Of The Art ◽  
Semantic Relation ◽  
The Other ◽  
Good Place ◽  
The Common

Finding our favorite dishes have became a hard task since restaurants are providing more choices and va- rieties. On the other hand, comments and reviews of restaurants are a good place to look for the answer. The purpose of this study is to use computational linguistics and natural language processing to categorise and find semantic relation in various dishes based on reviewers’ comments and menus description. Our goal is to imple- ment a state-of-the-art computational linguistics meth- ods such as, word embedding model, word2vec, topic modeling, PCA, classification algorithm. For visualiza- tions, t-Distributed Stochastic Neighbor Embedding (t- SNE) was used to explore the relation within dishes and their reviews. We also aim to extract the common pat- terns between different dishes among restaurants and reviews comment, and in reverse, explore the dishes with a semantics relations. A dataset of articles related to restaurant and located dishes within articles used to find comment patterns. Then we applied t-SNE visual- izations to identify the root of each feature of the dishes. As a result, to find a dish our model is able to assist users by several words of description and their inter- est. Our dataset contains 1,000 articles from food re- views agency on a variety of dishes from different cul- tures: American, i.e. ’steak’, hamburger; Chinese, i.e. ’stir fry’, ’dumplings’; Japanese, i.e., ’sushi’.

Emotion and Tension

2018 ◽  
Author(s):  
David Temperley
Keyword(s):  
Moderate Level ◽  
The Other ◽  
One Dimension ◽  
Two Dimensional ◽  
Dimensional Representation ◽  
Unexpected Events ◽  
Aesthetic Enjoyment ◽  
Third Dimension ◽  
Musical Emotion

Following much other research on musical emotion, this chapter assumes a two-dimensional representation, with one dimension representing valence (positive/negative) and the other representing energy/activity. It is argued that the valence dimension in rock is conveyed primarily by the location of a song’s scale (relative to the tonic) on the “line of fifths”; this captures the well-known major/minor contrast but also allows finer distinctions. The energy dimension is conveyed by a variety of musical parameters including loudness, register, tempo, rhythmic density, and timbral brightness. The chapter also posits a third dimension, complexity, which is taken to be represented experientially by tension; increased tension is caused by unexpected events and by an increase in event density. It has been hypothesized that a moderate level of complexity is optimal for aesthetic enjoyment; this may in part account for the appeal of certain rhythmic patterns, a phenomenon known as “groove.”

Physics Subject Headings (PhySH)

2020 ◽  
Vol 47 (3) ◽  
pp. 257-266
Author(s):  
Arthur Smith
Keyword(s):  
Ad Hoc ◽  
Classification Scheme ◽  
Physical Review ◽  
Controlled Vocabulary ◽  
The Other ◽  
One Dimension ◽  
Two Dimensional ◽  
Dimensional Classification ◽  
Subject Headings ◽  
American Physical

PhySH (Physics Subject Headings) was developed by the American Physical Society and first used in 2016 as a faceted hierarchical controlled vocabulary for physics, with some basic terms from related fields. It was developed mainly for the purpose of associating subjects with papers submitted to and published in the Physical Review family of journals. The scheme is organized at the top level with a two-dimensional classification, with one dimension (labeled “disciplines”) representing professional divisions within physics, and the other dimension (labeled “facets”) providing a conceptual partitioning of terms. PhySH was preceded in use by PACS (“Physics and Astronomy Classification Scheme”), which was in turn preceded by more ad hoc approaches, and this history and related vocabularies or categorizations will also be briefly discussed.

Sign in / Sign up

Export Citation Format

Share Document