Connections between integration along hypersufaces, Radon transforms, and neural networks are exploited to highlight an integral geometric mathematical interpretation of neural networks. By analyzing the properties of neural networks as operators on probability distributions for observed data, we show that the distribution of outputs for any node in a neural network can be interpreted as a nonlinear projection along hypersurfaces defined by level surfaces over the input data space. We utilize these descriptions to provide new interpretation for phenomena such as nonlinearity, pooling, activation functions, and adversarial examples in neural network-based learning problems.

Machine learning is often used in virtual screening to find compounds that are pharmacologically active on a target protein. The weave module is a type of graph convolutional deep neural network that uses not only features focusing on atoms alone (atom features) but also features focusing on atom pairs (pair features); thus, it can consider information of nonadjacent atoms. However, the correlation between the distance on the graph and the three-dimensional coordinate distance is uncertain. In this paper, we propose three improvements for modifying the weave module. First, the distances between ring atoms on the graph were modified to bring the distances on the graph closer to the coordinate distance. Second, different weight matrices were used depending on the distance on the graph in the convolution layers of the pair features. Finally, a weighted sum, by distance, was used when converting pair features to atom features. The experimental results show that the performance of the proposed method is slightly better than that of the weave module, and the improvement in the distance representation might be useful for compound activity prediction.

The increasing rate of information pollution on the Web requires novel solutions to tackle that. Question Answering (QA) interfaces are simplified and user-friendly interfaces to access information on the Web. However, similar to other AI applications, they are black boxes which do not manifest the details of the learning or reasoning steps for augmenting an answer. The Explainable Question Answering (XQA) system can alleviate the pain of information pollution where it provides transparency to the underlying computational model and exposes an interface enabling the end-user to access and validate provenance, validity, context, circulation, interpretation, and feedbacks of information. This position paper sheds light on the core concepts, expectations, and challenges in favor of the following questions (i) What is an XQA system?, (ii) Why do we need XQA?, (iii) When do we need XQA? (iv) How to represent the explanations? (iv) How to evaluate XQA systems?

Understanding how "black-box" models arrive at their predictions has sparked significant interest from both within and outside the AI community. Our work focuses on doing this by generating local explanations about individual predictions for tree-based ensembles, specifically Gradient Boosting Decision Trees (GBDTs). Given a correctly predicted instance in the training set, we wish to generate a counterfactual explanation for this instance, that is, the minimal perturbation of this instance such that the prediction flips to the opposite class. Most existing methods for counterfactual explanations are (1) model-agnostic, so they do not take into account the structure of the original model, and/or (2) involve building a surrogate model on top of the original model, which is not guaranteed to represent the original model accurately. There exists a method specifically for random forests; we wish to extend this method for GBDTs. This involves accounting for (1) the sequential dependency between trees and (2) training on the negative gradients instead of the original labels.

Procedural generation of initial states of state-space search problems have applications in human and machine learning as well as in the evaluation of planning systems. In this paper we deal with the task of generating hard and solvable initial states of Sokoban puzzles. We propose hardness metrics based on pattern database heuristics and the use of novelty to improve the exploration of search methods in the task of generating initial states. We then present a system called Beta that uses our hardness metrics and novelty to generate initial states. Experiments show that Beta is able to generate initial states that are harder to solve by a specialized solver than those designed by human experts.

Recent years have witnessed a fast-growing interest in computing explanations for Machine Learning (ML) models predictions. For non-interpretable ML models, the most commonly used approaches for computing explanations are heuristic in nature. In contrast, recent work proposed rigorous approaches for computing explanations, which hold for a given ML model and prediction over the entire instance space. This paper extends earlier work to the case of boosted trees and assesses the quality of explanations obtained with state-of-the-art heuristic approaches. On most of the datasets considered, and for the vast majority of instances, the explanations obtained with heuristic approaches are shown to be inadequate when the entire instance space is (implicitly) considered.

The goal of this paper is to set a constraint programming framework to solve lot-sizing problems. More specifically, we consider a single-item lot-sizing problem with time-varying lower and upper bounds for production and inventory. The cost structure includes time-varying holding costs, unitary production costs and setup costs. We establish a new lower bound for this problem by using a subtle time decomposition. We formulate this NP-hard problem as a global constraint and show that bound consistency can be achieved in pseudo-polynomial time and when not including the costs, in polynomial time. We develop filtering rules based on existing dynamic programming algorithms, exploiting the above mentioned time decomposition for difficult instances. In a numerical study, we compare several formulations of the problem: mixed integer linear programming, constraint programming and dynamic programming. We show that our global constraint is able to find solutions, unlike the decomposed constraint programming model and that constraint programming can be competitive, in particular when adding combinatorial side constraints.

Moral responsibility is a major concern in automated decision-making, with applications ranging from self-driving cars to kidney exchanges. From the viewpoint of automated systems, the urgent questions are: (a) How can models of moral scenarios and blameworthiness be extracted and learnt automatically from data? (b) How can judgements be computed tractably, given the split-second decision points faced by the system? By building on deep tractable probabilistic learning, we propose a learning regime for inducing models of such scenarios automatically from data and reasoning tractably from them. We report on experiments that compare our system with human judgement in three illustrative domains: lung cancer staging, teamwork management, and trolley problems.

Abstraction is a powerful idea widely used in science, to model, reason and explain the behavior of systems in a more tractable search space, by omitting irrelevant details. While notions of abstraction have matured for deterministic systems, the case for abstracting probabilistic models is not yet fully understood. In this paper, we provide a semantical framework for analyzing such abstractions from first principles. We develop the framework in a general way, allowing for expressive languages, including logic-based ones that admit relational and hierarchical constructs with stochastic primitives. We motivate a definition of consistency between a high-level model and its low-level counterpart, but also treat the case when the high-level model is missing critical information present in the low-level model. We prove properties of abstractions, both at the level of the parameter as well as the structure of the models. We conclude with some observations about how abstractions can be derived automatically.

We will first briefly consider how DL has contributed to the research on visual object recognition. In the main part we will assess whether DL could also help us to clarify the computations underlying higher cognitive functions such as Theory of Mind. In addition, we will compare the objectives and learning signals of brains and machines, leading us to conclude that simply scaling up the current DL algorithms will most likely not lead to human level Theory of Mind.