End-to-end neural models for intelligent dialogue systems suffer from the problem of generating uninformative responses. Various methods were proposed to generate more informative responses by leveraging external knowledge. However, few previous work has focused on selecting appropriate knowledge in the learning process. The inappropriate selection of knowledge could prohibit the model from learning to make full use of the knowledge. Motivated by this, we propose an end-to-end neural model which employs a novel knowledge selection mechanism where both prior and posterior distributions over knowledge are used to facilitate knowledge selection. Specifically, a posterior distribution over knowledge is inferred from both utterances and responses, and it ensures the appropriate selection of knowledge during the training process. Meanwhile, a prior distribution, which is inferred from utterances only, is used to approximate the posterior distribution so that appropriate knowledge can be selected even without responses during the inference process. Compared with the previous work, our model can better incorporate appropriate knowledge in response generation. Experiments on both automatic and human evaluation verify the superiority of our model over previous baselines.
Abstract:We use reactive molecular dynamics (RMD) simulations to study the interface between cyclotrimethylene trinitramine (RDX) and Aluminum (Al) with different oxide layers to elucidate the effect of nano-sized Al on thermal decomposition of RDX. A published ReaxFF force field for C/H/N/O elements was retrained to incorporate Al interactions, and then used in RMD simulations to characterize compound energetic materials. We find that the predicted adsorption energies for RDX on the Al (111) for RDX(210)/Al 2 O 3 (0001) provide a more accurate description. We conclude that the origin of these differences in dynamic behavior is due to the variations in the oxide layer morphologies.
In human conversation an input post is open to multiple potential responses, which is typically regarded as a one-to-many problem. Promising approaches mainly incorporate multiple latent mechanisms to build the one-to-many relationship. However, without accurate selection of the latent mechanism corresponding to the target response during training, these methods suffer from a rough optimization of latent mechanisms. In this paper, we propose a multi-mapping mechanism to better capture the one-to-many relationship, where multiple mapping modules are employed as latent mechanisms to model the semantic mappings from an input post to its diverse responses. For accurate optimization of latent mechanisms, a posterior mapping selection module is designed to select the corresponding mapping module according to the target response for further optimization. We also introduce an auxiliary matching loss to facilitate the optimization of posterior mapping selection. Empirical results demonstrate the superiority of our model in generating multiple diverse and informative responses over the state-of-the-art methods.1 Multi-modal means the property with multiple modes.
<sec>Recently, helicon plasma sources have aroused the great interest particularly in plasma-material interaction under fusion conditions. In this paper, the helicon wave antenna in helicon physics prototype experiment (HPPX) is optimized. To reveal the effect of the radial density configuration on wave field and energy flow, Maxwell's equations for a radially nonuniform plasma with standard cold-plasma dielectric tensor are solved. Helicon wave coupling and power deposition are studied under different types of antennas, different antenna lengths and driving frequencies by using HELIC. Through the numerical simulation, the optimal antenna structure and size are obtained, that is, half helix antenna, which works at 13.56 MHz and has a length of 0.4 m, can generate nonaxisymmetric radio frequency energy coupling to excite higher electron density.</sec><sec>The influences of different static magnetic fields and axis plasma densities on power deposition are also analyzed. It is found that the absorbed power of the plasma to the helicon wave has different peak power points in a multiple static magnetic field and axial plasma densities, and the overall coupling trend increases with static magnetic field increasing, but decreases with axis plasma density increasing. According to the simulation results, the ionization mechanism of helicon plasma is discussed. In order to further study the coupling of helicon wave with plasma in HPPX, the induced electromagnetic field and current density distribution are given when the plasma discharges. Under parabolic density distribution, the field intensity of the induced electric field at the edge is large, while neither the induced magnetic field nor current density changes much along the radial direction, the energy is distributed evenly in the whole plasma. Under the Gaussian density distribution, the induced electric field intensity is higher at the edge, while the induced magnetic field and current density in the center are much higher than at the edge. </sec><sec>In this paper studied are the structure and size of helicon wave antenna, the influences of static magnetic field and axial plasma density on plasma power deposition and the distribution of induced electromagnetic field and current density during plasma discharge under different density distributions. This work will provide important theoretical basis for helicon wave antena design and relevant physical experiments on HPPX.</sec>
In order to improve the detonation performance of hexahydro-1,3,5-trinitro-1,3,5-triazine (RDX) explosive, addictives with high heat values were used, and magnesium hydride (MgH2) is one of the candidates. However, it is important to see whether MgH2is a safe addictive. In this paper, the thermal and kinetic properties of RDX and mixture of RDX/MgH2were investigated by differential scanning calorimeter (DSC) and accelerating rate calorimeter (ARC), respectively. The apparent activation energy (E) and frequency factor (A) of thermal explosion were calculated based on the data of DSC experiments using the Kissinger and Ozawa approaches. The results show that the addition of MgH2decreases bothEandAof RDX, which means that the mixture of RDX/MgH2has a lower thermal stability than RDX, and the calculation results obtained from the ARC experiments data support this too. Besides, the most probable mechanism functions about the decomposition of RDX and RDX/MgH2were given in this paper which confirmed the change of the decomposition mechanism.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.