For both humans and machines, understanding the visual world requires relating new percepts with past experience. We argue that a good visual representation for an image should encode what makes it similar to other images, enabling the recall of associated experiences. Current machine implementations of visual representations can capture some aspects of similarity, but fall far short of human ability overall. Even if one explicitly labels objects in millions of images to tell the computer what should be considered similar—a very expensive procedure—the labels still do not capture everything that might be relevant. This thesis shows that one can often train a representation which captures similarity beyond what is labeled in a given dataset. That means we can begin with a dataset that has uninteresting labels, or no labels at all, and still build a useful representation. To do this, we propose to using pretext tasks: tasks that are not useful in and of themselves, but serve as an excuse to learn a more general-purpose representation. The labels for a pretext task can be inexpensive or even free. Furthermore, since this approach assumes training labels differ from the desired outputs, it can handle output spaces where the correct answer is ambiguous, and therefore impossible to annotate by hand. The thesis explores two broad classes of supervision. The first isweak image-level supervision, which is exploited to train mid-level discriminative patch classifiers. For example, given a dataset of street-level imagery labeled only with GPS coordinates, patch classifiers are trained to differentiate one specific geographical region (e.g. the city of Paris) from others. The resulting classifiers each automatically collect and associate a set of patches which all depict the same distinctive architectural element. In this way, we can learn to detect elements like balconies, signs, and lamps without annotations. The second type of supervision requires no information about images other than the pixels themselves. Instead, the algorithm is trained to predict the context around image patches. The context serves as a sort of weak label: to predict well, the algorithm must associate similar-looking patches which also have similar contexts. After training, the feature representation learned using this within-image context indeed captures visual similarity across images, which ultimately makes it useful for real tasks like object detection and geometry estimation.

The idea that music acts in part as a vehicle for meaning is a truism in both popular reception and music scholarship. The language used to speak and to write about music is replete with words that describe it metaphorically. Melodies descend; rhythms speed up; timbre is smooth. Certainly, we use these terms for communicative facility, yet by applying this language to music, we create metaphors that, according to Ludwig Wittgenstein, act as frames that direct interpretation. In the paper, I put forth a theory that views metaphor as the process of semantic transfer or substitution in which a non-musical concept stands in for a musical feature, effectively enabling us to hear music as more than simply sound. The use of certain metaphors receives inspiration from previously heard music, programs, a perceived similarity with non-musical phenomena, or a combination of these. The methodology that I propose coordinates these metaphors—places them within a single frame—and enables them to interact with one another and to create a more palpable musical experience for the listener. I use Chopin's E minor and A major preludes from Op. 28 as the primary models for expounding this hermeneutic. / text

The diploma thesis deals with the interpretation of Deml's prose anthology Moji přátelé/My Friends from an intertextual viewpoint. The work is focused mainly on the reconstruction of My Friends' genesis and intertextual structures and on the interpretation of identified references, i.e. allusions and quotations. The thesis also explores Deml's vocabulary of flowers by reference to Biedermeier's work, it surveys repertoire of the main motifs and offers analysis of author's poetics. The work is also focused on Deml's specific dialogue.

My first exposure to Brecht and his theories was as a high school drama student. One of our year twelve assessment tasks was to write and perform our own Brechtian drama using three or more alienation techniques. I wrote a piece about Religion and Fundamentalism, an issue that I felt strongly about at the time. By carefully following my teacher’s instructions and adhering to the assessment criteria I received a VHA. I concluded from this experience that political theatre could be made by following a simple recipe and combining key ingredients. As my knowledge of theatre and my own creative practice developed I came to understand the great complexity of Brechtian theory and the extreme difficulty of creating effective political theatre, that is, theatre that changes the world. Brecht’s theories have been so thoroughly absorbed into contemporary theatre practice that we no longer identify the techniques of Epic Theatre as necessarily political, nor do we acknowledge its radical origins. I have not yet seen a professional production of a Brechtian play but I’ve absorbed on countless occasions the brilliant reinterpretations of Brecht’s theories within the work of contemporary dramatists. My approach to creating political drama is eclectic and irreverent and I’m prepared to beg borrow and steal from the cannon of political theatre and popular media to create a drama that works, a drama that is both entertaining and provocative. Hoods is an adaptation for young audiences of my original play Kingswood Kids (2001). The process of re-purposing Kingwood Kids to Hoods has been a long and complex one. The process has triggered an analysis of my own creative practice and theory, and demanded an in-depth engagement with the theories and practice of key political theatre makers, most notably Brecht and Boal and more contemporary theatre makers such as Churchill, Kane, and Zeal Theatre. The focus of my exegesis is an inquiry into how the dramatist can create a theatre of currency that challenges the dominant culture and provokes critical thinking and political engagement in young audiences. It will particularly examine Brecht’s theory of alienation and argue its continued relevance, exploring how Brechtian techniques can be applied and re-interpreted through an in-depth analysis of my two works for young people, Hoods and Children of the Black Skirt. For the purposes of this short exegesis I have narrowed the inquiry by focusing on four key areas: Transformation, Structure, Pretext, Metatext.

Hyperspectral imaging is an expanding topic within the field of computer vision, that uses images of high spectral granularity. Contrastive learning is a discrim- inative approach to self-supervised learning, a form of unsupervised learning where the network is trained using self-created pseudo-labels. This work com- bines these two research areas and investigates how a pretrained network based on contrastive learning can be used for hyperspectral images. The hyperspectral images used in this work are generated from simulated RGB images and spec- tra from a spectral library. The network is trained with a pretext task based on data augmentations, and is evaluated through transfer learning and fine-tuning for a downstream task. The goal is to determine the impact of the pretext task on the downstream task and to determine the required amount of labelled data. The results show that the downstream task (a classifier) based on the pretrained network barely performs better than a classifier without a pretrained network. In the end, more research needs to be done to confirm or reject the benefit of a pretrained network based on contrastive learning for hyperspectral images. Also, the pretrained network should be tested on real-world hyperspectral data and trained with a pretext task designed for hyperspectral images.

Indiana University-Purdue University Indianapolis (IUPUI) / Wharton’s heroines are ordinary women who fight to secure material comfort and create selves that satisfy their emotional and sexual needs. These women often find that the two goals are mutually exclusive, since society strictly dictates appropriate behavior. This code of behavior stems from their relation to men: as objects to be won, as wives, and as mothers. In many instances, women are not even aware of their prescriptive roles and confuse their search for self with a search for security. Material comfort does not nurture Wharton’s heroines’ inner selves and they feel a metaphysical dissatisfaction, often seeking to find contentment through divorce or affairs. What they find in either case is that the cure to their ennui is not material, but mental. Wharton’s women seek a transcendent self—a self that is not dependent upon popular notions of respectability; a spiritual state that is independent from any attachment to social imperatives.