Spurious Paths Research Articles

Path-based multi-hop reasoning over knowledge graph for answering questions via adversarial reinforcement learning

Multi-hop knowledge graph question answering targets at pinpointing the answer entities by inferring across multiple triples in knowledge graphs. To enhance model interpretability, path-based methods are proposed. Specifically, with the advances of deep reinforcement learning (DRL), this paper explores to extend the line of RL-based approaches. However, existing solutions suffer from the issue of spurious paths. A major reason lies in that the agent takes an opportunistic way to explicitly pursue the predictive accuracy of answer entities instead of considering correct reasoning paths. To overcome this challenge, our idea is inspired by adversarial learning and we expect that a discriminator could effectively distinguish whether the reasoning chain is correct or not. To this end, we propose an interpretable reasoning method based on adversarial reinforcement learning for multi-hop KGQA, namely Adversarial Reinforcement Reasoning Network (AR2N). AR2N consists of two crucial components: an answer generator (i.e., policy network of RL) and a path discriminator. By alternately updating two components in an adversarial manner, the generator is able to infer answer entities by following the correct reasoning chain, the discriminator is capable of evaluating the plausibility of reasoning paths. Extensive experiments conducted over three benchmark datasets well demonstrate the effectiveness of our method.11For reproducibility, we publicly release the code at https://github.com/iDylanCui/AR2N.

A Physical Basis for Kinetic Compensation.

Kinetic compensation is a strong, positive correlation between the Arrhenius activation energy E and the frequency factor A for a reaction between the same reactants under similar experimental conditions or similar reactants under the same conditions, even though these parameters are supposed to be independent. The kinetic compensation effect (KCE) is demonstrated by a linear relationship between ln[A] and E/R in the eponymous Constable plot and has been the subject of more than 50 000 publications over the past 100 years, with no consensus opinion about the cause of this effect. In this paper, it is suggested that the linear relationship between ln[A] and E is the result of a real or spurious path dependence of the reaction history between the initial state of the pure reactant(s) and the final state of the pure product(s) having standard enthalpy and entropy differences, ΔH° and ΔS°, respectively. The single-step rate law approximation of a reversible reaction leads to T0 = H°/ΔS° as the dynamic thermal (thermodynamic) equilibrium temperature and 1/T0 = (ln[A̅/k0])/(E̅/R) as the slope of a Constable/KCE plot or the crossover temperature of Arrhenius lines in an isokinetic relationship (IKR), where A̅ and E̅ are mean values for the ensemble of compensating {Ei, Ai} pairs and k0 is a constant that accounts for the path dependence of the reaction history and reconciles the KCE with the IKR. This proposed physical basis for the KCE and IKR is supported by qualitative agreement between ΔH° and ΔS° calculated from the statistics of compensating {Ei, Ai} pairs in the literature, and the difference in the standard enthalpies and entropies of formation of the products and reactants for thermal decomposition of organic peroxides, calcium carbonate, and poly(methyl methacrylate).

An Intelligent Question-Answering Model over Educational Knowledge Graph for Sustainable Urban Living

With the development of education informatization and the accumulation of massive educational resources and teaching data in urban environments, educational knowledge graphs that provide good conditions for developing data-driven intelligent education have been proposed. Based on such educational knowledge graphs, the question-answering method can provide students with immediate coaching and significantly increase their learning interest and productivity. However, there is little research on knowledge graph question-answering focused on the educational field. Students tend to consult complex questions that require reasoning; however, the existing QA system cannot satisfy their complex information needs. To help improve sustainable learning efficiency, we propose a novel intelligent question-answering model applied in smart cities, which can reason over the educational knowledge graph to locate the answers to given questions. Our approach uses a highly expressive bilinear graph neural network technology to perform forward reasoning, utilizing the contextual information between graph nodes to improve reasoning ability. On this basis, we propose two-teacher knowledge distillation. We construct two distinct teacher networks by combining forward and backward reasoning, then incorporate the intermediate supervision signals from the two networks to guide the reasoning process, thereby mitigating the phenomenon of spurious path reasoning. Extensive experiments on the MOOC Q&A dataset prove the effectiveness of our approach.

The Spurious Path Problem in Abstraction

Abstraction is a powerful technique in search and planning. A fundamental problem of abstraction is that it can create spurious paths, i.e., abstract paths that do not correspond to valid concrete paths. In this paper, we define spurious paths as a generalization of spurious states. We show that spurious paths can be categorized into two types: state-independent spurious paths and state-specific spurious paths. We present a practical method that eliminates state-independent spurious paths, as well as state-specific spurious paths when integrated with mutex detection methods. We provide syntactical conditions under which our method can remove state-independent spurious paths completely. We demonstrate that eliminating spurious paths can improve a heuristic substantially, even in abstract spaces that are free of spurious states.

Impacts of Point Cloud Modeling on the Accuracy of Ray-Based Multipath Propagation Simulations

This article reports the comparison of multipath channels of outdoor small cells at 28 GHz simulated by point-cloud-based ray tracer and those from measurements. The comparison shows that the ray tracer is capable of reproducing measured channels with reasonable accuracy, given proper permittivity values set for major building walls. The accuracy, however, depends on the types of point clouds we input to the ray tracer: 1) raw point cloud from laser scanning measurements of the cellular site; 2) flattened point cloud where structural details of the environments are all lost like what we find in publicly available digital maps; and 3) processed point cloud where missing sections are complemented, flat and rough sections of surfaces are identified and wedges are detected. For condensed parameters of channels such as path loss, delay, and angular spreads, ray tracing with the processed and raw point clouds shows equally good accuracy. In contrast, the ray tracing with the processed point cloud shows clear advantage of accuracy over that with the raw point cloud when simulating physical propagation paths. The raw point cloud produces many spurious paths that do not exist in measurements, whereas the processed point cloud reproduces multipath richness in form of many weak diffracted paths that are found in measurements and cannot be reproduced by diffuse scattering. Ray tracing with the flattened point cloud is in least agreement with measurements. Still, first-order specular reflections from large smooth walls are reproduced well by ray tracer with any of the tested point clouds. We thereby demonstrate the superiority of using the processed point cloud as inputs to ray tracer to obtain multipath channels with the highest fidelity to measurements.

You Shall Not Pass: Avoiding Spurious Paths in Shortest-Path Based Centralities in Multidimensional Complex Networks

The aggregation process may create spurious paths on the aggregated view of multidimensional (high order) complex networks. Consequently, these spurious paths may then cause shortest-path based centrality metrics to produce incorrect results, thus undermining the network centrality analysis. In this context, we propose a method built upon MultiAspect Graphs (MAGs) able to avoid taking into account spurious paths when computing centralities based on shortest paths in multidimensional complex networks. We show that, by using our proposed method, pitfalls usually associated with spurious paths resulting from aggregation in multidimensional networks can be avoided at the time of the aggregation process. As a result, shortest-path based centralities are assured to be computed correctly for multidimensional networks, without taking into account spurious paths that could otherwise lead to incorrect results. We also present case studies that show the impact of spurious paths in the computing of shortest paths and consequently of shortest-path based centralities, such as betweenness and closeness, thus illustrating the importance of this contribution.

Tailoring Multiloop Atom Interferometers with Adjustable Momentum Transfer.

Multiloop matter-wave interferometers are essential in quantum sensing to measure the derivatives of physical quantities in time or space. Because multiloop interferometers require multiple reflections, imperfections of the matter-wave mirrors create spurious paths that scramble the signal of interest. Here, we demonstrate a method of adjustable momentum transfer that prevents the recombination of the spurious paths in a double-loop atom interferometer aimed at measuring rotation rates. We experimentally study the recombination condition of the spurious matter waves, which is quantitatively supported by a model accounting for the coherence properties of the atomic source. We finally demonstrate the effectiveness of the method in building a cold-atom gyroscope with a single-shot acceleration sensitivity suppressed by a factor of at least 50. Our study will impact the design of multiloop atom interferometers that measure a single inertial quantity.

Semiclassical analysis of the quantum instanton approximation.

We explore the relation between the quantum and semiclassical instanton approximations for the reaction rate constant. From the quantum instanton expression, we analyze the contributions to the rate constant in terms of minimum-action paths and find that two such paths dominate the expression. For symmetric barriers, these two paths join together to describe the semiclassical instanton periodic orbit. However, for asymmetric barriers, one of the two paths takes an unphysically low energy and dominates the expression, leading to order-of-magnitude errors in the rate predictions. Nevertheless, semiclassical instanton theory remains accurate. We conclude that semiclassical instanton theory can be obtained directly from the semiclassical limit of the quantum instanton only for symmetric systems. We suggest a modification of the quantum instanton approach which avoids sampling the spurious path and thus has a stronger connection to semiclassical instanton theory, giving numerically accurate predictions even for very asymmetric systems in the low temperature limit.

Making CEGAR More Efficient in Software Model Checking

Counter-example guided abstraction refinement (CEGAR) is widely used in software model checking. With an abstract model, the state space is largely reduced, however, a counterexample found in such a model that does not satisfy the desired property may not exist in the concrete model. Therefore, how to check whether a reported counterexample is spurious is a key problem in the abstraction-refinement loop. Next, in the case that a spurious counterexample is found, the abstract model needs to be further refined where an NP-hard state separation problem is often involved. Thus, how to refine the abstract model efficiently has attracted a great attention in the past years. In this paper, by re-analyzing spurious counterexamples, a new formal definition of spurious paths is given. Based on it, efficient algorithms for detecting spurious counterexamples are presented. By the new algorithms, when dealing with infinite counterexamples, the finite prefix to be analyzed will be polynomially shorter than the one dealt with by the existing algorithms. Moreover, in practical terms, the new algorithms can naturally be parallelized that enables multi-core processors contributes more in spurious counterexample checking. In addition, a novel refining approach by adding extra Boolean variables to the abstract model is presented. With this approach, not only the NP-hard state separation problem can be avoided, but also a smaller refined abstract model can be obtained. Experimental results show that the new algorithms perform well in practice.

AUTOMATIC SEGMENTATION AND STRUCTURAL CHARACTERIZATION OF LOW DENSITY FIBREBOARDS

In this paper, a new skeleton-based algorithm for the segmentation of individual fibres in 3D tomographic images is described. The proposed method is designed to deal with low density materials featuring fibres with varied sizes, shapes and tortuosities, like composite fibreboards used for buildings insulation. To this end the paths of the skeleton are first classified according to their connectivity, the connectivity of their adjacent nodes, their orientation, their average radius and the variation of the distance transform along each path. This allows for the identification of spurious paths and paths linking two fibres. Reconstruction of the path of the fibres is done thanks to an optimal pairing algorithm which joins paths that show the most similar orientation and radius at each node/link. The segmented skeleton is finally dilated by means of a growing algorithm ordonned by the average radius of the fibres in order to reconstruct each identified fibres. As an application, the algorithm is used to segment a 3D tomographic image of hemp/polymer fibreboard for buildings insulation. Information such as the number of contacts, tortuosity, length, average radius, orientation of fibres are finally measured on both the segmented skeleton and reconstructed image, which allow for a thorough characterization of the fibre network.

An enhanced flow analysis technique for detecting unreachability faults in concurrent systems

We present a flow analysis technique for detecting unreachable states and actions in concurrent systems. It is an enhancement of the approach by Cheung and Kramer. Each process of a concurrent system is modeled as a finite state machine, whose states represent process execution states and whose transitions are labeled by actions. We construct dependency sets incrementally and eliminate spurious paths by checking the execution sequences of actions. We prove mathematically that our algorithm can detect more unreachability faults than the well-known Reif/Smolka and Cheung/Kramer algorithms. The algorithm is easy to manage and its complexity is still polynomial to the system size. Case studies on two commonly used communication protocols show that the technique is effective.

Self-aligning arrays for communication in multipath conditions

Highly directional aerial systems have been used to exclude signals arriving via spurious paths in communication systems, in order to reduce intersymbol distortion or frequencyselective fading. It is then necessary to use a tracking facility if one of the terminals is moving with respect to the other. In this letter, the problems associated with self-tracking aerials or acoustic transducers in multipath-propagation conditions are considered, and some proposals are made for the realisation of such arrays.

"High-gain traveling-wave amplifiers"

Modern techniques make it possible to build traveling-wave amplifiers with stable gains as high as 60 or 70 db. Such tubes can have relatively flat frequency response, and high gain over frequency ranges well in excess of two to one. In design it is necessary to pay strict attention to adequate isolation of the input and output so that spurious feedback paths do not cause oscillation or excessive variation of gain with frequency. In many practical applications, such tubes are required to operate with source and load impedances which produce large reflections of energy. Special design techniques result in tubes which are stable under such conditions. Another feature which can be achieved in modern tubes is a high ratio of backward loss to forward gain, so that the amplifier can very effectively isolate an oscillator. The recent design trend is toward higher currents and lower voltages, which is of direct benefit to the equipment designer, since it results in shorter tubes as well as simpler power supply designs, With the use of highly convergent electron beams, cathode current density can be kept quite low, and ion bombardment effects minimized, so that excellent life is achieved. Traveling-wave amplifiers may be readily phase modulated since small variations in the beam voltage yield large phase excursions. This characteristic is used for high level mixing. By proper gun design, it is possible to obtain good mixer action with modulation frequencies as high as a few hundred megacycles. The problems involved in obtaining the above characteristics and typical experimental results are illustrated by data on the STS-69, a half-watt twt for the 2 to 4 kmc frequency range.