Topic-relevant Response Generation using Optimal Transport for an Open-domain Dialog System

The workshop program of the Association for the Advancement of Artificial Intelligence’s 33rd Conference on Artificial Intelligence (AAAI-19) was held in Honolulu, Hawaii, on Sunday and Monday, January 27–28, 2019. There were fifteen workshops in the program: Affective Content Analysis: Modeling Affect-in-Action, Agile Robotics for Industrial Automation Competition, Artificial Intelligence for Cyber Security, Artificial Intelligence Safety, Dialog System Technology Challenge, Engineering Dependable and Secure Machine Learning Systems, Games and Simulations for Artificial Intelligence, Health Intelligence, Knowledge Extraction from Games, Network Interpretability for Deep Learning, Plan, Activity, and Intent Recognition, Reasoning and Learning for Human-Machine Dialogues, Reasoning for Complex Question Answering, Recommender Systems Meet Natural Language Processing, Reinforcement Learning in Games, and Reproducible AI. This report contains brief summaries of the all the workshops that were held.

Download Full-text

Open Domain New Word Detection Using Condition Random Field Method

Journal of Software ◽

10.3724/sp.j.1001.2013.04254 ◽

2013 ◽

Vol 24 (5) ◽

pp. 1051-1060 ◽

Cited By ~ 7

Author(s):

Fei CHEN ◽

Yi-Qun LIU ◽

Chao WEI ◽

Yun-Liang ZHANG ◽

Min ZHANG ◽

...

Keyword(s):

Random Field ◽

Field Method ◽

Open Domain ◽

Word Detection

Download Full-text

End-to-End Task-Oriented Dialog System Through Template Slot Value Generation

10.21437/interspeech.2020-2011 ◽

2020 ◽

Author(s):

Teakgyu Hong ◽

Oh-Woog Kwon ◽

Young-Kil Kim

Keyword(s):

Dialog System ◽

End To End ◽

Task Oriented

Download Full-text

Optimal transport and unitary orbits in C⁎-algebras

Journal of Functional Analysis ◽

10.1016/j.jfa.2021.109068 ◽

2021 ◽

Vol 281 (5) ◽

pp. 109068

Author(s):

Bhishan Jacelon ◽

Karen R. Strung ◽

Alessandro Vignati

Keyword(s):

Optimal Transport ◽

C Algebras

Download Full-text

Modeling Epidemic Spread among a Commuting Population Using Transport Schemes

Mathematics ◽

10.3390/math9161861 ◽

2021 ◽

Vol 9 (16) ◽

pp. 1861

Author(s):

Daniela Calvetti ◽

Alexander P. Hoover ◽

Johnie Rose ◽

Erkki Somersalo

Keyword(s):

Network Model ◽

Transport Theory ◽

Optimal Transport ◽

State Level ◽

Compartment Model ◽

Census Bureau ◽

Mitigation Strategies ◽

Mitigation Measures ◽

Metapopulation Model ◽

Infection Dynamics

Understanding the dynamics of the spread of COVID-19 between connected communities is fundamental in planning appropriate mitigation measures. To that end, we propose and analyze a novel metapopulation network model, particularly suitable for modeling commuter traffic patterns, that takes into account the connectivity between a heterogeneous set of communities, each with its own infection dynamics. In the novel metapopulation model that we propose here, transport schemes developed in optimal transport theory provide an efficient and easily implementable way of describing the temporary population redistribution due to traffic, such as the daily commuter traffic between work and residence. Locally, infection dynamics in individual communities are described in terms of a susceptible-exposed-infected-recovered (SEIR) compartment model, modified to account for the specific features of COVID-19, most notably its spread by asymptomatic and presymptomatic infected individuals. The mathematical foundation of our metapopulation network model is akin to a transport scheme between two population distributions, namely the residential distribution and the workplace distribution, whose interface can be inferred from commuter mobility data made available by the US Census Bureau. We use the proposed metapopulation model to test the dynamics of the spread of COVID-19 on two networks, a smaller one comprising 7 counties in the Greater Cleveland area in Ohio, and a larger one consisting of 74 counties in the Pittsburgh–Cleveland–Detroit corridor following the Lake Erie’s American coastline. The model simulations indicate that densely populated regions effectively act as amplifiers of the infection for the surrounding, less densely populated areas, in agreement with the pattern of infections observed in the course of the COVID-19 pandemic. Computed examples show that the model can be used also to test different mitigation strategies, including one based on state-level travel restrictions, another on county level triggered social distancing, as well as a combination of the two.

Download Full-text