Search for: All records

Explaining the results of Machine learning algorithms is crucial given the rapid growth and potential applicability of these methods in critical domains including healthcare, defense, autonomous driving, etc. In this paper, we address this problem in the context of Markov Logic Networks (MLNs) which are highly expressive statistical relational models that combine first-order logic with probabilistic graphical models. MLNs in general are known to be interpretable models, i.e., MLNs can be understood more easily by humans as compared to models learned by approaches such as deep learning. However, at the same time, it is not straightforward to obtain human-understandable explanations specific to an observed inference result (e.g. marginal probability estimate). This is because, the MLN provides a lifted interpretation, one that generalizes to all possible worlds/instantiations, which are not query/evidence specific. In this paper, we extract grounded-explanations, i.e., explanations defined w.r.t specific inference queries and observed evidence. We extract these explanations from importance weights defined over the MLN formulas that encode the contribution of formulas towards the final inference results. We validate our approach in real world problems related to analyzing reviews from Yelp, and show through user-studies that our explanations are richer than state-of-the-art non-relational explainers such as LIME . 

In order to understand the characteristics of long‐lasting “C‐type” structure in the Sodium (Na) lidargram, six cases from different observational locations have been analyzed. The Na lidargram, collected from low‐, middle‐, and high‐latitude sites, show long lifetime of the C‐type structures which is believed to be the manifestation of Kelvin‐Helmholtz (KH) billows in the Mesosphere and Lower Thermosphere (MLT) region. In order to explore the characteristics of the long‐lasting C‐type structures, the altitude profile of square of Brunt‐Väisälä frequency in the MLT region has been derived using the temperature profile collected from the Na lidar instruments and the SABER instrument onboard TIMED satellite. It is found to be positive in the C‐type structure region for all the six cases which indicates that the regions are convectively stable. Simultaneous wind measurements, which allowed us to calculate the Richardson numbers and Reynolds numbers for three cases, suggest that the regions where the C‐type structure appeared were dynamically stable and nonturbulent. This paper brings out a hypothesis wherein the low temperature can increase the magnitude of the Prandtl number and convectively stable atmospheric region can cause the magnitude of Reynolds number to decrease. As a consequence, the remnant of previously generated KH billows in nearly “frozen‐in” condition can be advected through this conducive region to a different location by the background wind where they can sustain for a long time without much deformation. These long‐lived KH billows in the MLT region will eventually manifest the long‐lasting C‐type structures in the Na lidargram.