Traditional clustering identifies groups of objects that share certain qualities. Tangles do the converse: they identify groups of qualities that often occur together. They can thereby discover, relate, and structure types: of behaviour, political views, texts, or viruses. If desired, tangles can also be used for direct clustering of objects. They offer a precise, quantitative paradigm suited particularly to fuzzy clusters, since they do not require any `hard' assignments of objects to the clusters they collectively form. This is a draft of the introductory chapter of a book I am preparing on the application of tangles in the empirical sciences. The purpose of posting this draft early is to give authors of tangle application papers a generic reference for the basic guiding principles underlying tangle applications outside mathematics, so that in their own papers they can concentrate on the ideas specific to their particular application rather than having to repeat the generic story each time. The text starts with three separate generic introductions to tangles in the natural sciences, in the social sciences, and in data science including machine learning. It then gives a short informal description of the abstract notion of tangles that encompasses all these potential applications.

Approaches to Grounded Language Learning typically focus on a single task-based final performance measure that may not depend on desirable properties of the learned hidden representations, such as their ability to predict salient attributes or to generalise to unseen situations. To remedy this, we present GROLLA, an evaluation framework for Grounded Language Learning with Attributes with three sub-tasks: 1) Goal-oriented evaluation; 2) Object attribute prediction evaluation; and 3) Zero-shot evaluation. We also propose a new dataset CompGuessWhat?! as an instance of this framework for evaluating the quality of learned neural representations, in particular concerning attribute grounding. To this end, we extend the original GuessWhat?! dataset by including a semantic layer on top of the perceptual one. Specifically, we enrich the VisualGenome scene graphs associated with the GuessWhat?! images with abstract and situated attributes. By using diagnostic classifiers, we show that current models learn representations that are not expressive enough to encode object attributes (average F1 of 44.27). In addition, they do not learn strategies nor representations that are robust enough to perform well when novel scenes or objects are involved in gameplay (zero-shot best accuracy 50.06%).

Designing reward functions is a challenging problem in AI and robotics. Humans usually have a difficult time directly specifying all the desirable behaviors that a robot needs to optimize. One common approach is to learn reward functions from collected expert demonstrations. However, learning reward functions from demonstrations introduces many challenges: some methods require highly structured models, e.g. reward functions that are linear in some predefined set of features, while others adopt less structured reward functions that on the other hand require tremendous amount of data. In addition, humans tend to have a difficult time providing demonstrations on robots with high degrees of freedom, or even quantifying reward values for given demonstrations. To address these challenges, we present a preference-based learning approach, where as an alternative, the human feedback is only in the form of comparisons between trajectories. Furthermore, we do not assume highly constrained structures on the reward function. Instead, we model the reward function using a Gaussian Process (GP) and propose a mathematical formulation to actively find a GP using only human preferences. Our approach enables us to tackle both inflexibility and data-inefficiency problems within a preference-based learning framework. Our results in simulations and a user study suggest that our approach can efficiently learn expressive reward functions for robotics tasks.

Recent work in cognitive science has uncovered a diversity of explanatory values, or dimensions along which we judge explanations as better or worse. We propose a Bayesian account of how these values fit together to guide explanation. The resulting taxonomy provides a set of predictors for which explanations people prefer and shows how core values from psychology, statistics, and the philosophy of science emerge from a common mathematical framework. In addition to operationalizing the explanatory virtues associated with, for example, scientific argument-making, this framework also enables us to reinterpret the explanatory vices that drive conspiracy theories, delusions, and extremist ideologies. Intuitively, philosophically, and as seen in laboratory experiments, explanations are judged as better or worse on the basis of many different criteria. These explanatory values appear in early childhood [1, 2, 3, 4, 5] and their influence extends to some of the most sophisticated social knowledge formation processes we know [6]. We lack, however, an understanding of the origin of these values or an account of how they fit together to guide belief formation. The multiplicity of values also appears to conflict with Bayesian models of cognition, which speak solely in terms of degrees of beliefs and suggest we judge explanations as better or worse on the basis of a single quantity, the posterior likelihood (see Glossary). In this opinion, we show how to resolve these conflicts by arguing that previously-identified explanatory values capture different components of a full Bayesian calculation and, when considered together and weighed appropriately, implement Bayesian cognition. This framework shows how key explanatory values identified by laboratory experiments and philosophers of science--co-explanation, descriptiveness, precision, unification, power, and simplicity--emerge naturally from the mathematical structure of probabilistic inference, thereby reconciling them with Bayesian models of cognition [7, 8]. Second, it shows how these values combine to produce preferences for one explanation over another.

Roads have well defined geometries, topologies, and traffic rules. While this has been widely exploited in motion planning methods to produce maneuvers that obey the law, little work has been devoted to utilize these priors in perception and motion forecasting methods. In this paper we propose to incorporate these structured priors as a loss function. In contrast to imposing hard constraints, this approach allows the model to handle non-compliant maneuvers when those happen in the real world. Safe motion planning is the end goal, and thus a probabilistic characterization of the possible future developments of the scene is key to choose the plan with the lowest expected cost. Towards this goal, we design a framework that leverages REINFORCE to incorporate non-differentiable priors over sample trajectories from a probabilistic model, thus optimizing the whole distribution. We demonstrate the effectiveness of our approach on real-world self-driving datasets containing complex road topologies and multi-agent interactions. Our motion forecasts not only exhibit better precision and map understanding, but most importantly result in safer motion plans taken by our self-driving vehicle. We emphasize that despite the importance of this evaluation, it has been often overlooked by previous perception and motion forecasting works.

Reinforcement learning algorithms have had tremendous successes in online learning settings. However, these successes have relied on low-stakes interactions between the algorithmic agent and its environment. In many settings where RL could be of use, such as health care and autonomous driving, the mistakes made by most online RL algorithms during early training come with unacceptable costs. These settings require developing reinforcement learning algorithms that can operate in the so-called batch setting, where the algorithms must learn from set of data that is fixed, finite, and generated from some (possibly unknown) policy. Evaluating policies different from the one that collected the data is called off-policy evaluation, and naturally poses counter-factual questions. In this project we show how off-policy evaluation and the estimation of treatment effects in causal inference are two approaches to the same problem, and compare recent progress in these two areas.

Industry~4.0 (I4.0) standards and standardization frameworks have been proposed with the goal of \emph{empowering interoperability} in smart factories. These standards enable the description and interaction of the main components, systems, and processes inside of a smart factory. Due to the growing number of frameworks and standards, there is an increasing need for approaches that automatically analyze the landscape of I4.0 standards. Standardization frameworks classify standards according to their functions into layers and dimensions. However, similar standards can be classified differently across the frameworks, producing, thus, interoperability conflicts among them. Semantic-based approaches that rely on ontologies and knowledge graphs, have been proposed to represent standards, known relations among them, as well as their classification according to existing frameworks. Albeit informative, the structured modeling of the I4.0 landscape only provides the foundations for detecting interoperability issues. Thus, graph-based analytical methods able to exploit knowledge encoded by these approaches, are required to uncover alignments among standards. We study the relatedness among standards and frameworks based on community analysis to discover knowledge that helps to cope with interoperability conflicts between standards. We use knowledge graph embeddings to automatically create these communities exploiting the meaning of the existing relationships. In particular, we focus on the identification of similar standards, i.e., communities of standards, and analyze their properties to detect unknown relations. We empirically evaluate our approach on a knowledge graph of I4.0 standards using the Trans$^*$ family of embedding models for knowledge graph entities. Our results are promising and suggest that relations among standards can be detected accurately.

The morality of algorithms and their potential for bias and discrimination are important concerns. A popular approach to machine learning and artificial intelligence is via the numerical optimization of objective functions, and adapting such an approach to handle ethics could seem natural: with a hammer in hand, everything looks like a nail. The hammer of much artificial intelligence is the optimization of objective values, so some might like to treat morality solely through such objective functions. However, relying solely on the optimization of scalar objective values is fraught with unavoidable flaws when dealing with real people.

We propose to explain the behavior of black-box prediction methods (e.g., deep neural networks trained on image pixel data) using causal graphical models. Specifically, we explore learning the structure of a causal graph where the nodes represent prediction outcomes along with a set of macro-level "interpretable" features, while allowing for arbitrary unmeasured confounding among these variables. The resulting graph may indicate which of the interpretable features, if any, are possible causes of the prediction outcome and which may be merely associated with prediction outcomes due to confounding. The approach is motivated by a counterfactual theory of causal explanation wherein good explanations point to factors which are "difference-makers" in an interventionist sense. The resulting analysis may be useful in algorithm auditing and evaluation, by identifying features which make a causal difference to the algorithm's output.

Concrete domains, especially those that allow to compare features with numeric values, have long been recognized as a very desirable extension of description logics (DLs), and significant efforts have been invested into adding them to usual DLs while keeping the complexity of reasoning in check. For expressive DLs and in the presence of general TBoxes, for standard reasoning tasks like consistency, the most general decidability results are for the so-called $\omega$-admissible domains, which are required to be dense. Supporting non-dense domains for features that range over integers or natural numbers remained largely open, despite often being singled out as a highly desirable extension. The decidability of some extensions of $\mathcal{ALC}$ with non-dense domains has been shown, but existing results rely on powerful machinery that does not allow to infer any elementary bounds on the complexity of the problem. In this paper, we study an extension of $\mathcal{ALC}$ with a rich integer domain that allows for comparisons (between features, and between features and constants coded in unary), and prove that consistency can be solved using automata-theoretic techniques in single exponential time, and thus has no higher worst-case complexity than standard $\mathcal{ALC}$. Our upper bounds apply to some extensions of DLs with concrete domains known from the literature, support general TBoxes, and allow for comparing values along paths of ordinary (not necessarily functional) roles.