We demonstrate the need and potential of systematically integrated vision and semantics solutions for visual sensemaking in the backdrop of autonomous driving. A general neurosymbolic method for online visual sensemaking using answer set programming (ASP) is systematically formalised and fully implemented. The method integrates state of the art in visual computing, and is developed as a modular framework that is generally usable within hybrid architectures for realtime perception and control. We evaluate and demonstrate with community established benchmarks KITTIMOD, MOT-2017, and MOT-2020. As use-case, we focus on the significance of human-centred visual sensemaking -- e.g., involving semantic representation and explainability, question-answering, commonsense interpolation -- in safety-critical autonomous driving situations. The developed neurosymbolic framework is domain-independent, with the case of autonomous driving designed to serve as an exemplar for online visual sensemaking in diverse cognitive interaction settings in the backdrop of select human-centred AI technology design considerations. Keywords: Cognitive Vision, Deep Semantics, Declarative Spatial Reasoning, Knowledge Representation and Reasoning, Commonsense Reasoning, Visual Abduction, Answer Set Programming, Autonomous Driving, Human-Centred Computing and Design, Standardisation in Driving Technology, Spatial Cognition and AI.

In this paper, we investigate CNF encodings, for which unit propagation is strong enough to derive a contradiction if the encoding is not consistent with a partial assignment of the variables (unit refutation complete or URC encoding) or additionally to derive all implied literals if the encoding is consistent with the partial assignment (propagation complete or PC encoding). We prove an exponential separation between the sizes of PC and URC encodings without auxiliary variables and strengthen the known results on their relationship to the PC and URC encodings that can use auxiliary variables. Besides of this, we prove that the sizes of any two irredundant PC formulas representing the same function differ at most by a factor polynomial in the number of the variables and present an example of a function demonstrating that a similar statement is not true for URC formulas. One of the separations above implies that a q-Horn formula may require an exponential number of additional clauses to become a URC formula. On the other hand, for every q-Horn formula, we present a polynomial size URC encoding of the same function using auxiliary variables. This encoding is not q-Horn in general.

Transformers have been shown to emulate logical deduction over natural language theories (logical rules expressed in natural language), reliably assigning true/false labels to candidate implications. However, their ability to generate implications of a theory has not yet been demonstrated, and methods for reconstructing proofs of answers are imperfect. In this work we show that a generative model, called ProofWriter, can reliably generate both implications of a theory and the natural language proof(s) that support them. In particular, iterating a 1-step implication generator results in proofs that are highly reliable, and represent actual model decisions (rather than post-hoc rationalizations). On the RuleTaker dataset, the accuracy of ProofWriter's proofs exceed previous methods by 9% absolute, and in a way that generalizes to proof depths unseen in training and on out-of-domain problems. We also show that generative techniques can perform a type of abduction with high precision: Given a theory and an unprovable conclusion, identify a missing fact that allows the conclusion to be proved, along with a proof. These results Figure 1: Given facts, rules, and a question all expressed significantly improve the viability of neural in natural language, ProofWriter answers the methods for systematically reasoning over question and generates a proof of the answer.

Heifetz, Meier and Schipper (HMS) present a lattice model of awareness. The HMS model is syntax-free, which precludes the simple option to rely on formal language to induce lattices, and represents uncertainty and unawareness with one entangled construct, making it difficult to assess the properties of either. Here, we present a model based on a lattice of Kripke models, induced by atom subset inclusion, in which uncertainty and unawareness are separate. We show the models to be equivalent by defining transformations between them which preserve formula satisfaction, and obtain completeness through our and HMS' results.

We introduce a new programming paradigm called oracle-guided decision programming in which a program specifies a Markov Decision Process (MDP) and the language provides a universal policy. We prototype a new programming language, Dodona, that manifests this paradigm using a primitive 'choose' representing nondeterministic choice. The Dodona interpreter returns either a value or a choicepoint that includes a lossless encoding of all information necessary in principle to make an optimal decision. Meta-interpreters query Dodona's (neural) oracle on these choicepoints to get policy and value estimates, which they can use to perform heuristic search on the underlying MDP. We demonstrate Dodona's potential for zero-shot heuristic guidance by meta-learning over hundreds of synthetic tasks that simulate basic operations over lists, trees, Church datastructures, polynomials, first-order terms and higher-order terms.

Given a system that does not work as expected, Sequential Diagnosis (SD) aims at suggesting a series of system measurements to isolate the true explanation for the system's misbehavior from a potentially exponential set of possible explanations. To reason about the best next measurement, SD methods usually require a sample of possible fault explanations at each step of the iterative diagnostic process. The computation of this sample can be accomplished by various diagnostic search algorithms. Among those, Reiter's HS-Tree is one of the most popular due its desirable properties and general applicability. Usually, HS-Tree is used in a stateless fashion throughout the SD process to (re)compute a sample of possible fault explanations in each iteration, each time given the latest (updated) system knowledge including all so-far collected measurements. At this, the built search tree is discarded between two iterations, although often large parts of the tree have to be rebuilt in the next iteration, involving redundant operations and calls to costly reasoning services. As a remedy to this, we propose DynamicHS, a variant of HS-Tree that maintains state throughout the diagnostic session and additionally embraces special strategies to minimize the number of expensive reasoner invocations. In this vein, DynamicHS provides an answer to a longstanding question posed by Raymond Reiter in his seminal paper from 1987. Extensive evaluations on real-world diagnosis problems prove the reasonability of the DynamicHS and testify its clear superiority to HS-Tree wrt. computation time. More specifically, DynamicHS outperformed HS-Tree in 96% of the executed sequential diagnosis sessions and, per run, the latter required up to 800% the time of the former. Remarkably, DynamicHS achieves these performance improvements while preserving all desirable properties as well as the general applicability of HS-Tree.

Logical formalisms provide a natural and concise means for specifying and reasoning about preferences. In this paper, we propose lexicographic logic, an extension of classical propositional logic that can express a variety of preferences, most notably lexicographic ones. The proposed logic supports a simple new connective whose semantics can be defined in terms of finite lists of truth values. We demonstrate that, despite the well-known theoretical limitations that pose barriers to the quantitative representation of lexicographic preferences, there exists a subset of the rational numbers over which the proposed new connective can be naturally defined. Lexicographic logic can be used to define in a simple way some well-known preferential operators, like "$A$ and if possible $B$", and "$A$ or failing that $B$". Moreover, many other hierarchical preferential operators can be defined using a systematic approach. We argue that the new logic is an effective formalism for ranking query results according to the satisfaction level of user preferences.

Logical forgetting is removing some variables from a logical formula while preserving its information regarding the others [LLM03]. Seen from a different angle, it is restricting a formula to keep only its information about some of its variables [Del17]. Each of the two interpretations has its own applications. Removing information allow reducing memory requirements [EKI19], simplifying reasoning [DW15, EF07, WSS05] and clarifying the relationship about variables [Del17]. Restricting a formula on some variables allows formalizing the limited knowledge of agents [FHMV95, RHPT14], ensuring privacy [GKLW17] and removing inconsistency [LM10].

A recent work has shown that transformers are able to "reason" with facts and rules in a limited setting where the rules are natural language expressions of conjunctions of conditions implying a conclusion. Since this suggests that transformers may be used for reasoning with knowledge given in natural language, we do a rigorous evaluation of this with respect to a common form of knowledge and its corresponding reasoning -- the reasoning about effects of actions. Reasoning about action and change has been a top focus in the knowledge representation subfield of AI from the early days of AI and more recently it has been a highlight aspect in common sense question answering. We consider four action domains (Blocks World, Logistics, Dock-Worker-Robots and a Generic Domain) in natural language and create QA datasets that involve reasoning about the effects of actions in these domains. We investigate the ability of transformers to (a) learn to reason in these domains and (b) transfer that learning from the generic domains to the other domains.

A sliding puzzle is a combination puzzle where a player slide pieces along certain routes on a board to reach a certain end-configuration. In this paper, we propose a novel measurement of complexity of massive sliding puzzles with paramodulation which is an inference method of automated reasoning. It turned out that by counting the number of clauses yielded with paramodulation, we can evaluate the difficulty of each puzzle. In experiment, we have generated 100 * 8 puzzles which passed the solvability checking by countering inversions. By doing this, we can distinguish the complexity of 8 puzzles with the number of generated with paramodulation. For example, board [2,3,6,1,7,8,5,4, hole] is the easiest with score 3008 and board [6,5,8,7,4,3,2,1, hole] is the most difficult with score 48653. Besides, we have succeeded to obverse several layers of complexity (the number of clauses generated) in 100 puzzles. We can conclude that proposal method can provide a new perspective of paramodulation complexity concerning sliding block puzzles.