We saw in Chapter 2 that Born’s statistical interpretation of the wave function was one of the building blocks of quantum theory. According to Born’s interpretation, the wave function of a particle at a given moment of time defined a probability density $|\psi (x){|}^2$ with respect to the position $x$. This result is generalized to state vectors of an Hilbert space and to general observables through the following procedure.

Previous chapters considered detection of backdoors before/during training and post-training. Here, our objective is to detect use of a backdoor trigger operationally, that is, at test time. Such detection may prevent potentially catastrophic decisions, as well as potentially catching culprits in the act of exploiting a learned backdoor mapping. We also refer to such detection as “in-flight.” A likelihood based backdoor trigger detector is developed and compared against other detectors.

Backdoor attacks have been considered in non-image data domains, including speech and audio, text, as well as for regression applications (Chapter 12). In this chapter, we consider classification of point cloud data, for example, LiDAR data used by autonomous vehicles. Point cloud data differs significantly from images, with the former representing a given scene/object by a collection of points in 3D (or a higher-dimensional) space. Accordingly, point cloud DNN classifiers (such as PointNet) deviate significantly from the DNN architectures commonly used for image classification. So, backdoor (as well as test-time evasion) attacks also need to be customized to the nature of the (point cloud) data. Such attacks typically involve either adding points, deleting points, or modifying (transforming) the points representing a given scene/object. While test-time evasion attacks against point cloud classifiers were previously proposed, in this chapter we develop backdoor attacks against point cloud classifiers (based on insertion of points designed to defeat the classifier, as well as to defeat anomaly detectors that identify point outliers and remove them). We also devise a post-training detector designed to defeat this attack, as well as other point cloud backdoor attacks.

This chapter begins (Module 1.1) by comparing prescriptive and descriptive approaches to grammar, and evaluating different sources of linguistic data; it goes on to consider the nature of grammatical categories and features. Module 1.2 then turns to look at the merge and adjunction operations which generate syntactic structures, and at the underlying principles of X-bar Syntax. Next Module 1.3 examines the syntax of null constituents, and the role played by the relation c-command in a range of syntactic phenomena (e.g. case-marking). Module 1.4 goes on to explore three different types of movement operation, namely A bar Movement, A movement, and Head Movement. Subsequently Module 1.5 examines the role of constraints in blocking illicit operations, and of filters in blocking illicit structures. The chapter concludes with a Summary (Module 1.6), Bibliography (Module 1.7), and Workbook (Module 1.8), with some Workbook exercise examples designed for self-study, and others for assignments/seminar discussion.

In this chapter we describe reverse-engineering attacks (REAs) on classifiers and defenses against them. REAs involve querying (probing) a classifier to discover its decision rules. One primary application of REAs is to enable TTEs. Another is to reveal a private (e.g., proprietary) classifier’s decision-making. For example, an adversary may seek to discover the workings of a military automated target-recognition system. Early work demonstrates that, with a modest number of (random) queries, which do not rely on any knowledge of the nominal data distribution, one can learn a surrogate classifier on a given domain that closely mimics an unknown classifier. However, a critical weakness of this attack is that random querying makes the attack easily detectable – randomly selected query patterns will typically look nothing like legitimate examples. They are likely to be extreme outliers of all the classes. Each such query is thus individually highly suspicious, let alone thousands or millions of such queries (required for accurate reverse-engineering). However, more recent REAs, which are akin to active learning strategies, are stealthier. Here, we use the ADA method (developed in Chapter 4 for TTE detection) to detect REAs. This method is demonstrated to provide significant detection power against stealthy REAs.

In Chapter 1, we presented the fundamental principles of classical physics, and then we motivated and presented the fundamental principles of quantum physics. The two sets of principles are summarized and compared in Table 10.1.

This chapter explains the importance of social support for relationship maintenance and individual functioning. It first reviews common stressors (e.g., life events, low socioeconomic status, minority stress and stigma), their accompanying personal and relational costs, and the consequences of social support in adverse or stressful contexts. In particular, this section highlights the different consequences of perceiving support availability and actual support receipt during stress. Next, this chapter reviews the role of supportive relationships to facilitate personal goal pursuit and desired self-change in non-adverse contexts. Finally, this chapter considers social support from the perspective of the support provider and describes how caregiving can be both rewarding and costly.