DOCSLIB.ORG

The Computational Creativity Complex

Total Page：16

File Type：pdf, Size：1020Kb

Download full-text PDF Read full-text
The Computational Creativity Complex The Computational Creativity Complex Dan Ventura Abstract We briefly examine the subject of computational creativity through the lenses of three different systems for producing creative artifacts in three different domains: music, visual art and cookery. From these we attempt to abstract some- thing of a “general-purpose” computationally creative agent and examine how this agent might behave in the context of an archetypical “algorithm” for creativity. Fi- nally, we consider this agent’s behavior from the point of view of the theory of (Turing) computability and suggest that computational creativity as a field provides an interesting opportunity for looking at computational complexity in new ways. 1 Inception The question of computational creativity can be approached from several different angles. Here we assume that creativity is possible in computational systems and examine the idea of a general, abstract mechanism for computational creativity that exists somehow independently from any particular domain. This is a difficulty ques- tion, and it is clear that much of creativity is not a domain independent construct. However, we posit that there does exist some core abstraction or creativity “algo- rithm” that can be applied to any domain (with suitable domain-specific augmen- tation, of course). Given this hypothesis, we attempt an inductive approach to the problem by first examining three computationally creative systems in three differ- ent domains and then attempt to generalize from these specific examples an abstract model of a creative system. As a complementary view of the problem, we also discuss an abstract “al- gorithm” for creativity and relate this “algorithm” to the abstract system, asking whether this “algorithm” could, in fact, become an algorithm in the formal sense, Dan Ventura Computer Science Department, Brigham Young University, e-mail: [email protected] 1 2 Dan Ventura and attempt to reason about the answer to that question. In other words, we try to ascertain whether computational creativity is actually computable. 2 Three Blind Mice Our three specific examples, from which we will try to generalize an abstract sys- tem are DARCI, a computational artist that creates visualizations for communicating concepts [31, 32, 19, 33, 20]; CARL, a computational composer that discovers mu- sical motifs in non-musical sources and composes music around them [22, 21]; and PIERRE a computational chef that creates original slow cooker recipes[29]. Each of these systems has been presented in more detail elsewhere, and we only give enough detail here to support the generalization that is our goal. 2.1 DARCI DARCI is a system for generating original images that convey intention and is in- spired by other artistic image generating systems such as AARON [28] and The Painting Fool [7]. Central to the design philosophy of DARCI is the notion that the communication of meaning in art is a necessary part of eliciting an aesthetic expe- rience in the viewer [11], and it is unique in that it creates images that explicitly express a given concept using visual metaphor. This is currently done at two levels: using iconic nouns as surrogates for the target concept and using image filters to convey associated adjectival content. DARCI is composed of two major subsystems, an image understanding com- ponent, and an image generation component. The image understanding component learns how to associate images with concepts in the forms of nouns and adjectives. The image generation component composes an original source image as a collage of iconic noun concepts and then uses a genetic algorithm, governed by the analysis component, to render this source image to visually convey an adjective. Figure 1 outlines this process of creating artifacts. 2.1.1 Image Understanding DARCI’s understanding of images is derived from two sources: a mapping from low-level image features to descriptive adjectives and semantic associations between linguistic concepts. Visuo-Linguistic Association In order for DARCI to make associations between images and associated adjectives, the system learns a mapping from low-level com- puter vision features [18, 13, 25, 43, 23, 42] to words using images that are hand- The Computational Creativity Complex 3 Fig. 1 A diagram outlining the two major components of DARCI. Image analysis learns how to annotate new images with adjectives using a series of appreciation networks trained with labeled images (outlined in blue). Image generation uses a semantic memory model to identify nouns and adjectives associated with a given concept. The nouns are composed into a source image that is rendered to reflect the adjectives, using a genetic algorithm that is governed by a set of evaluation metrics. The final product (outlined in red) is an image that communicates the given concept. labeled with adjective tags. The use of WordNet’s [17] database of adjective synsets allows images to be described by their affect, most of their aesthetic qualities, many of their possible associations, and even, to some extent, by their subject. To collect training data we have created a public website for training DARCI (http://darci.cs.byu.edu), where users are presented with a random image and asked to provide adjectives that describe the image. When users input a word with multiple senses, they are presented with a list of the available senses, along with the WordNet gloss, and asked to select the most appropriate one. Additionally, for each image presented to the user, DARCI lists seven adjectives that it associates with the image. The user is then allowed to flag those labels that are not accurate. Learning image to synset associations is a multi-label classification problem [39], meaning each image can be associated with more than one synset. To handle this, we use a collection of artificial neural networks (ANNs) that we call appreciation networks, each of which outputs a single real value, between 0 and 1, indicating the degree to which a given image can be described by the networks’ corresponding synset (adjective). An appreciation network is created for each synset that has a sufficient number of training data, and as data is incrementally accumulated, new neural networks are dynamically added to the collection to accommodate any new synsets. There are currently close to 300 appreciation networks in the system. Semantic Memory Model The system also contains a simple cognitive model, built as a semantic network forming a graph of associations between words [37, 4 Dan Ventura 14]. These word associations are acquired in one of two ways: from people and by automatic inference from a corpus, with the idea being to use the human word associations to capture general knowledge and then to fill in the gaps using the corpus associations. For the human word associations, we use two pre-existing databases of free as- sociation norms (FANs): the Edinburgh Associative Thesaurus [24] and the Univer- sity of Florida’s Word Association Norms [30]. These word associations were ac- quired by asking hundreds of human volunteers to provide the first word that comes to mind when given a cue word. This technique is able to capture many different types of word associations including word co-ordination (pepper, salt), collocation (trash, can), super-ordination (insect, butterfly), synonymy (starving, hungry), and antonymy (good, bad). The association strength between two words is simply a count of the number of volunteers that said the second word given the first word. FANs are considered to be one of the best methods for understanding how people, in general, associate words in their own minds [30]. For the corpus-based associations, we build a (term × term) co-occurrence matrix from a large corpus, in a manner similar to that employed in the Hyperspace Analog to Language (HAL) model [27]. For our corpus, we use the entire (English) text of Wikipedia, as it is large, easily accessible, and covers a wide range of human knowledge [15]. Once the co-occurrence matrix is built, we use the co-occurrence values themselves as association strengths between words. This approach works, since we only care about the strongest associations between words, and it allows us to reduce the number of irrelevant associations by ignoring any word pairs with a co-occurrence count less than some threshold. Our final semantic network is a composition of the human- and corpus-based as- sociations, which essentially merges the two separate graphs into a single network before querying it for associations. This method assumes that the human data con- tains more valuable word associations than the corpus data because such human data is typically used as the gold standard in the literature. However, the corpus data does contain some valuable associations not present in the human data. To combine the graphs, we add the top n associations for each word from the corpus data to the human data but weight the corpus-based association strengths lower than the human-based associations. This is beneficial for two reasons. First, if there are any associations that overlap, adding them again will strengthen the association in the combined network. Second, corpus-based associations not present in the human data will be added to the combined network and provide a greater variety of word asso- ciations. We keep the association strength low because we want the corpus data to reinforce, but not dominate, the human data. 2.1.2 Image Generation DARCI generates images in two stages: the creation of a source image composed of a collage of concept icons and the rendering of this source image using various pa- rameterized image filters. The collage generation is driven by the semantic network, The Computational Creativity Complex 5 while the filtered rendering is achieved using an evolutionary mechanism whose fit- ness function is defined in terms of the outputs of the visuo-linguistic association networks.
Load more

Recommended publications
  • Combinational Creativity and Computational Creativity
    Combinational Creativity and Computational Creativity By Ji HAN Dyson School of Design Engineering Imperial College London A thesis submitted for the degree of Doctor of Philosophy April 2018 Originality Declaration The coding of the Combinator and the Retriever was accomplished in collaboration with Feng SHI, a PhD colleague in the Creativity Design and Innovation Lab at the Dyson School of Design Engineering, Imperial College London, and co-author with me of relevant publications. Except where otherwise stated, this thesis is the result of my own research. This research was conducted in the Dyson School of Design Engineering at Imperial College London, between October 2014 and December 2017. I certify that this thesis has not been submitted in whole or in parts as consideration for any other degree or qualification at this or any other institute of learning. Ji HAN April 2018 II Copyright Declaration The copyright of this thesis rests with the author and is made available under a Creative Commons Attribution Non-Commericial No Derivitives licence. Researchers are free to copy, distribute or transmit the thesis on the condition that they attribute it, that they do not use it for commercial purposes and that they do not alter, transform or build upon it. For any reuse or redistribution, researchers must make clear to others the license terms of this work. Ji HAN April 2018 III List of Publications The following journal and conference papers were published during this PhD research. Journal Publications 1. Han J., Shi F., Chen L., Childs P. R. N., 2018. The Combinator – A computer-based tool for creative idea generation based on a simulation approach.
    [Show full text]
  • Computational Creativity
    Computational Creativity Hannu Toivonen University of Helsinki [email protected] www.cs.helsinki.fi/hannu.toivonen ECSS, Gothenburg, 9 Oct 2018 8.10.2018 1 The following video clips, pictures and audio files have been removed from this file to save space: – Video: Poemcatcher tests ”Brain Poetry” machine at Frankfurt Book Fair: https://www.youtube.com/watch?v=cNnbTQL8j B4 – Images produced by Deep Dream, see e.g. https://en.wikipedia.org/wiki/DeepDream – Audio clip: music produced by a programme by Turing and his colleagues www.helsinki.fi/yliopisto 9.10.2018 2 Remote Associates Test (RAT) – What word relates to all of these three words? – coin – silver coin – quick – quick silver – spoon – silver spoon – Measures the ability to discover relationships between remotely associated concepts – A (controversial) psychometrical test of creativity – Correlates with IQ and originality in brain storming www.helsinki.fi/yliopisto 8.10.2018 3 Modeling RATs computationally – A single RAT question is a quadruple – A probabilistic approach: find that maximizes – Maximize (cf. naïve Bayes) www.helsinki.fi/yliopisto 8.10.2018 4 Modeling RATs computationally – Learn word frequencies from a large corpus – Use Google 1 and 2-grams to estimate probabilities and – (Google n-grams: a large, publically available collection of word sequences and their probabilities) – A lot more could be done, but we want to keep things as simple as possible – Creative behavior without an explicit semantic resource (such as WordNet) www.helsinki.fi/yliopisto 8.10.2018
    [Show full text]
  • Collaboratively Designing Metrics to Evaluate Creative Machines
    ISEA 2020 MEASURING COMPUTATIONAL CREATIVITY: COLLABORATIVELY DESIGNING METRICS TO EVALUATE CREATIVE MACHINES EUNSU KANG JEAN OH ROBERT TWOMEY WELCOME! :D WE UNDERSTAND WE CANNOT MEASURE EVERY ASPECT OF CREATIVITY BUT WE WANT TO FIGURE OUT THE MEASURABLE ASPECTS OF CREATIVITY WE ARE HERE FOR • Collectively brainstorming evaluation metrics and producing a set of metrics. • Creating a map of metrics to help us have a better view on the problem • Building a community that contributes to this difficult task OUR SCHEDULE TODAY IS • Introduction (20 min) • Small group discussion: Elements of creative AI (30 min) • Sharing the first discussion results (15 min) • Invited speaker panel 1 (35 min) • Small group discussion 2: Evaluating selected projects (30 min) • Sharing the second discussion results (15 min) • Invited speaker panel 2 (35 min) • Small group discussion 3: Revising metrics, evaluating the second project (30 min) • Presentation of results and Q&A (30 min) • 5 min breaks as needed PANEL DISCUSSIONS BY INVITED SPEAKERS Panel 1: Graham Wakefield, Haru Ji, Fabrizio Poltronieri, Allison Parrish, Aaron Hertzmann • What are your methods and metrics for evaluating the creative AI system that you and your colleagues have developed? • How does the study of creative AI systems inform human creative practice? • How do we attribute (who is responsible for) the creativity in collaborative creative systems? • Do you think there is a difference between creative AI and human creativity? What would be the difference? If not, why not. PANEL DISCUSSIONS
    [Show full text]
  • Explainable Computational Creativity
    Explainable Computational Creativity Maria Teresa Llano Mark d’Inverno Matthew Yee-King SensiLab, Monash University Goldsmiths, University of London Goldsmiths, University of London Melbourne, Australia London, United Kingdom London, United Kingdom Jon McCormack Alon Ilsar Alison Pease Simon Colton SensiLab, Monash University SensiLab, Monash University Dundee University SensiLab, Monash University Melbourne, Australia Melbourne, Australia Dundee, United Kingdom Queen Mary, University of London Abstract for intermediate explanations as the process progresses, let Human collaboration with systems within the Computational alone place for exchanges of information that can exploit Creativity (CC) field is often restricted to shallow interac- human-machine co-creation. tions, where the creative processes, of systems and humans alike, are carried out in isolation, without any (or little) inter- In this paper we propose Explainable Computational Cre- vention from the user, and without any discussion about how ativity (XCC) as a subfield of XAI. The focus of this subfield the unfolding decisions are taking place. Fruitful co-creation is the study of bidirectional explainable models in the con- requires a sustained ongoing interaction that can include dis- text of computational creativity – where the term explainable cussions of ideas, comparisons to previous/other works, in- is used with a broader sense to cover not only one shot-style cremental improvements and revisions, etc. For these interac- explanations, but also for co-creative interventions
    [Show full text]
  • Download Thesis
    Department of Computer Science Series of Publications A Report A-2021-4 Computational Understanding, Generation and Evaluation of Creative Expressions Khalid Alnajjar Doctoral dissertation, to be presented for public discussion with the permission of the Faculty of Science of the University of Helsinki, in Auditorium B123, Exactum, Pietari Kalmin katu 5, on the 22nd of March, 2021 at 12:00 o’clock. The defence is also open for the audience through remote access. University of Helsinki Finland Supervisor Hannu Toivonen, University of Helsinki, Finland Pre-examiners Josep Blat, Universitat Pompeu Fabra, Spain Tapio Salakoski, University of Turku, Finland Opponent Pablo Gervás, Universidad Complutense de Madrid, Spain Custos Hannu Toivonen, University of Helsinki, Finland Faculty Representative Teemu Roos, University of Helsinki, Finland Contact information Department of Computer Science P.O. Box 68 (Pietari Kalmin katu 5) FI-00014 University of Helsinki Finland Email address: [email protected].fi URL: https://cs.helsinki.fi/ Telephone: +358 2941 911 Copyright c 2021 Khalid Alnajjar ISSN 1238-8645 ISBN 978-951-51-7145-0 (paperback) ISBN 978-951-51-7146-7 (PDF) Helsinki 2021 Unigrafia Computational Understanding, Generation and Evaluation of Creative Expressions Khalid Alnajjar Department of Computer Science P.O. Box 68, FI-00014 University of Helsinki, Finland khalid.alnajjar@helsinki.fi https://www.khalidalnajjar.com/ https://www.rootroo.com/ PhD Thesis, Series of Publications A, Report A-2021-4 Helsinki, March 2021, 56 pages ISSN 1238-8645 ISBN 978-951-51-7145-0 (paperback) ISBN 978-951-51-7146-7 (PDF) Abstract Computational creativity has received a good amount of research interest in generating creative artefacts programmatically.
    [Show full text]
  • Teaching Computational Creativity Margareta Ackerman1, Ashok Goel2, Colin G
    Teaching Computational Creativity Margareta Ackerman1, Ashok Goel2, Colin G. Johnson3, Anna Jordanous3, Carlos Le´on4, Rafael P´erezy P´erez5, Hannu Toivonen6, and Dan Ventura7. 1 Computer Engineering Department, Santa Clara University. [email protected]. 2 School of Interactive Computing, Georgia Tech. [email protected] 3 School of Computing, University of Kent. fC.G.Johnson, [email protected] 4 Department of Software Engineering and Artificial Intelligence, Complutense University of Madrid. [email protected] 5 Depto. de Tecnolog´ıasde la Informaci´on,Universidad Aut´onomaMetropolitana. [email protected] 6 Department of Computer Science, University of Helsinki. [email protected] 7 Computer Science Department, Brigham Young University. [email protected] Abstract • Master theoretical and conceptual tools for analysing and discussing creative systems The increasing popularity of computational creativity (CC) in recent years gives rise to the need for educa- • Develop and understand one's own creative processes tional resources. This paper presents several modules that together act as a guide for developing new CC • Become familiar with the latest advances in CC, and courses as well as improving existing curricula. As well be able to critically discuss current state-of-the-art as introducing core CC concepts, we address pedagog- • Become able to create CC systems and/or techniques ical approaches to this interdisciplinary subject. An accessible overview of the field lets this paper double • Gain the ability to describe, employ and debate as an introductory tutorial to computational creativity. methods for evaluation of computational creativity • Be able to identify appropriate contexts for using CC Introduction In what follows, we offer examples of extant CC sys- As computational creativity (CC) grows in popular- tems that can be used as archetypes for introducing ity, it is increasingly being taught in some form as a key questions in the field; discuss different modeling university-level course.
    [Show full text]
  • Computational Creativity: Coming of Age
    Editorial Computational Creativity: Coming of Age Simon Colton, Ramon López de Mántaras, and Oliviero Stock n This editorial provides an introduction to current AI research on computationally created A artifacts as well as to the papers and topics cov- t its heart, computational creativity is the study of build- ered by this special issue on computational cre- ing software that exhibits behavior that would be deemed cre- ativity. ative in humans. Such creative software can be used for autonomous creative tasks, such as inventing mathematical the- ories, writing poems, painting pictures, and composing music. However, computational creativity studies also enable us to understand human creativity and to produce programs for cre- ative people to use, where the software acts as a creative collab- orator rather than a mere tool. Historically, it’s been difficult for society to come to terms with machines that purport to be intel- ligent and even more difficult to admit that they might be cre- ative. For instance, in 1934, some professors at the University of Manchester in the United Kingdom built meccano models that were able to solve some mathematical equations. Groundbreak- ing for its time, this project was written up in a piece in Mecca- no Magazine. The article was titled “Are Thinking Machines Pos- sible” and was very upbeat, but surprisingly ends by stating that “Truly creative thinking of course will always remain beyond the power of any machine.” Surely, though, this attitude has changed in light of the amaz- ing advances in hardware and software technology that fol- lowed those meccano models? Sadly, no.
    [Show full text]
  • A Critical Review of Computational Creativity in Built Environment Design
    buildings Review A Critical Review of Computational Creativity in Built Environment Design Ning Gu * and Peiman Amini Behbahani UniSA Creative, University of South Australia, Adelaide 5001, Australia; [email protected] * Correspondence: [email protected] Abstract: Computational creativity in built environment (BE) design has been a subject of research interest in the discipline. This paper presents a critical review of various ways computational creativ- ity has been and can be defined and approached in BE design. The paper examines a comprehensive body of contemporary literature on the topics of creativity, computational creativity, and their assess- ment to identify levels of computational creativity. The paper then proceeds to a further review of the implications of these levels specifically in BE design. The paper identifies four areas in BE design where computational creativity is relevant. In two areas—synthesis (generation) and analysis—there is considerable literature on lower levels of computational creativity. However, in two other areas— interfacing and communication—even the definition of computational creativity is not as defined and clear for the discipline, and most works only consider the role of computers as a supporting tool or medium. These open up future research opportunities for the discipline. Keywords: computational creativity; human–computer (co-)creativity; creativity support tools; built environment design 1. Introduction As the use of computation and artificial intelligence has increased in built environment Citation: Gu, N.; Amini Behbahani, P. (BE) design, their roles in supporting creativity, a quintessential quality of design, have A Critical Review of Computational been examined more frequently [1–4]. Each of these topics—computation, creativity, and Creativity in Built Environment Design.
    [Show full text]
  • Artistic Creativity in Artificial Intelligence
    Artistic Creativity in Artificial Intelligence a dissertation presented by Deniz E. Kurt to The Department of Creative Industries in partial fulfillment of the requirements for the degree of Master of Arts in the subject of Arts & Culture Radboud University Nijmegen, Netherlands April 2018 ©2018 – Deniz E. Kurt all rights reserved. Thesis advisor: Dr. Vincent Meelberg Deniz E. Kurt Artistic Creativity in Artificial Intelligence Abstract Computational art is a creative field that indicates to a futuristic idea of artificial intelligence. De- spite the common belief that a machine is unable to create art, current developments and examples in computational art present a new form of art. Reaching to a broad variety of artistic dimensions, artificial intelligence programs are generating poetry, music, visual art, architecture and design. This study introduces artificial intelligence as an artistic phenomenon and analyzes the artifacts that are produced by computer algorithms. Thus, it provides a theoretical framework and a philosophical dis- cussion on the creative abilities of artificial intelligence. Furthermore, it concerns with the entitlement of artiness, by presenting a reflection on the dynamics between the artwork, the art-maker and the art audience. The analysis of computational artworks and the computer programs that generate those artworks show that the function of artificial intelligence is far beyond being merely a tool to create art, it is rather an actor that have an artistic and creative agency. With this study, I suggest an alternative approach in the conceptual definition of art, which can help to explore the possibilities of a new art genre. iii Contents 0 Introduction 1 1 Human Intelligence vs.
    [Show full text]
  • Associative Algorithms for Computational Creativity
    LAV R. VARSHNEY JUN WANG KUSH R. VARSHNEY Special Issue on Creativity and Food Associative Algorithms for Computational Creativity ABSTRACT Computational creativity, the generation of new, unimagined ideas or artifacts by a machine that are deemed creative by people, can be applied in the culinary domain to create novel and flavorful dishes. In fact, we have done so successfully using a combina- torial algorithm for recipe generation combined with statistical models for recipe ranking and selection. However, the algorithm of creation in our prior work may be difficult for people to interpret, understand, and ultimately adopt because the process differs from the process of human creativity theorized in the psychology and neuroscience literatures. In this paper, to address this issue, we discuss how human creativity, including in the food arena, may be built on associations and how an algorithm also built on associations can be more relatable to people so they can interact with the tool more easily. We pro- pose a computational creativity approach that extends the data mining technique of asso- ciation rule mining to generate new food recipes. We illustrate this associative algorithm on real-world culinary data. Keywords: association mining, computational creativity, food. Computational creativity is the use of computers to generate and select ideas or arti- facts that would be regarded as creative if produced by people alone. This area of study has a long history with workers motivated by disparate goals. Some researchers are after scientific understanding of the mind, whereas others have the technological goal of pro- ducing useful tools; some practitioners may try to illuminate the psychological processes underlying creativity, whereas others try to produce esthetically interesting results (Boden, 2015).
    [Show full text]
  • Developing Creative AI to Generate Sculptural Objects
    Developing Creative AI to Generate Sculptural Objects Songwei Ge∗, Austin Dill?, Eunsu Kang?, Chun-Liang Li?, Lingyao Zhang Manzil Zaheer, Barnabas Poczos Carnegie Mellon University Pittsburgh, PA, United States fchunlial, eunsuk, songweig, lingyaoz, abdill, manzil, [email protected] Abstract We explore the intersection of human and machine creativ- ity by generating sculptural objects through machine learning. This research raises questions about both the technical details of automatic art generation and the interaction between AI and people, as both artists and the audience of art. We introduce two algorithms for generating 3D point clouds and then dis- cuss their actualization as sculpture and incorporation into a holistic art installation. Specifically, the Amalgamated Deep- Dream (ADD) algorithm solves the sparsity problem caused by the naive DeepDream-inspired approach and generates cre- ative and printable point clouds. The Partitioned DeepDream (PDD) algorithm further allows us to explore more diverse 3D object creation by combining point cloud clustering algorithms Figure 1: Sculpture generated by creative AI, PDD. and ADD. sculptures. Nima et al. [27] introduced Vox2Netadapted from Keywords pix2pix GAN which can extract the abstract shape of the input Partitioned DeepDream, Amalgamated DeepDream, 3D, sculpture. DeepCloud [5] exploited an autoencoder to gener- Point Cloud, Sculpture, Art, Interactive Installation, Creative ate point clouds, and constructed a web interface with analog AI, Machine Learning input devices for users to interact with the generation process. Similar to the work of Lehman et al [22], the results generated Introduction by DeepCloud are limited to previously seen objects. In this paper, we introduce two ML algorithms to create Will Artificial Intelligence (AI) replace human artists or will original sculptural objects.
    [Show full text]
  • Bridging Generative Deep Learning and Computational Creativity
    Bridging Generative Deep Learning and Computational Creativity Sebastian Berns 1 and Simon Colton 1, 2 1 Game AI Group, EECS, Queen Mary University of London, United Kingdom 2 SensiLab, Faculty of IT, Monash University, Australia Abstract by way of the creative AI community, and we propose av- enues for improvements and cross-community engagement. We aim to help bridge the research fields of generative deep learning and computational creativity by way of the creative In the section below, we make a case for generative mod- AI community, and to advocate the common objective of els as a successful and powerful technology which is inher- more creatively autonomous generative learning systems. We ently limited in its creative abilities by its uni-dimensional argue here that generative deep learning models are inher- objective of perfection. The following section discusses ently limited in their creative abilities because of a focus on how, in spite of its limitations, GANs have been used and learning for perfection. To highlight this, we present a se- abused as artwork production engines. We then explore how ries of techniques which actively diverge from standard us- computational creativity research can contribute to further age of deep learning, with the specific intention of producing evolve such models into more autonomous creative systems, novel and interesting artefacts. We sketch out some avenues looking specifically at novelty measures as a first step to- for improvement of the training and application of genera- wards this goal. We conclude by returning to the notion of tive models and discuss how previous work on the evaluation of novelty in a computational creativity setting could be har- bridging the two fields and describing future work.
    [Show full text]