Portugal

&quot;There are two classes of generative art approaches: neural, where a deep model is trained to generate samples from a data distribution, and “symbolic” or algorithmic, where an artist designs the primary parameters and an autonomous system generates samples within these constraints. In this work, we propose a new hybrid
genre: neuro-symbolic generative art. As a preliminary study, we train a generative deep neural network on samples from the symbolic approach. We demonstrate through human studies that subjects find the final artifacts and the creation process using our neuro-symbolic approach to be more creative than the symbolic approach 61% and 82% of the time respectively.&quot;

ICCC&#39;20

Neuro-Symbolic Generative Art: A Preliminary Study

symbolic art

generative art

gans

deep neural networks

stages 

COLING, the International Conference on Computational Linguistics, is one of the premier conferences for natural language processing and computational linguistics. Often grouped within the field of artificial intelligence, but actually pre-dating the development of artificial intelligence, advances in computational linguistics and natural language processing are now some of the major drivers behind the use of artificial intelligence for commercial and social applications – for example, on-line search, machine translation and with voice-assisted conversational devices.

First established in 1965, the biennial COLING conference is held in diverse parts of the globe and attracts participants from both top-ranked research centers and emerging countries. Today, the most important developments in our field are taking place not only in universities and academic research institutes, but also in industrial research departments and in technological startups. COLING conferences provide opportunities for all these communities to showcase their exciting developments.

COLING 2020

COLING, the International Conference on Computational Linguistics, is one of the premier conferences for natural language processing and computational linguistics. Often grouped within the field of artificial intelligence, but actually pre-dating the development of artificial intelligence, advances in computational linguistics and natural language processing are now some of the major drivers behind the use of artificial intelligence for commercial and social applications – for example, on-line search, machine translation and with voice-assisted conversational devices.

"There are two classes of generative art approaches: neural, where a deep model is trained to generate samples from a data distribution, and “symbolic” or algorithmic, where an artist designs the primary parameters and an autonomous system generates samples within these constraints. In this work, we propose a new hybrid
genre: neuro-symbolic generative art. As a preliminary study, we train a generative deep neural network on samples from the symbolic approach. We demonstrate through human studies that subjects find the final artifacts and the creation process using our neuro-symbolic approach to be more creative than the symbolic approach 61% and 82% of the time respectively."

poster

Computational creativity is the art, science, philosophy and engineering of computational systems which, by taking on particular responsibilities, exhibit behaviours that unbiased observers would deem to be creative. As a field of research, this area is thriving, with progress in formalising what it means for software to be creative, along with many exciting and valuable applications of creative software in the sciences, the arts, literature, gaming and elsewhere.

The ICCC conference series organized by the Association for Computational Creativity since 2010 is the only scientific conference that focuses on computational creativity alone and also covers all aspects of it.


ICCC'20

The ICCC conference series organized by the Association for Computational Creativity since 2010 is the only scientific conference that focuses on computational creativity alone and also covers all aspects of it.
Computational creativity is the art, science, philosophy and engineering of computational systems which, by taking on particular responsibilities, exhibit behaviors that unbiased observers would deem to be creative. As a field of research, this area is thriving, with progress in formalizing what it means for software to be creative, along with many exciting and valuable applications of creative software in the sciences, the arts, literature, gaming and elsewhere.



We present how we built the dataset and models, summarize evaluation results and show a few examples of generated music at the end.

Automated Music Generation for Visual Art through Emotion

We present a proposal for meta-level evaluation in creative systems. In this paper, the Mexica creative system is the subject of our analysis. To achieve meta-level evaluation, we have constructed a program that interacts with Mexica's inputs and its outputs. The meta-level system evaluates the results and makes decisions based on them. For decision-making it includes heuristics that are defined in terms of Mexica's results; these are rules particular to the object- level system. We also include other general metrics that apply in any other creative system. For example, Ritchie's criteria for evaluating creative systems are included in this proposal.

Meta-level Evaluation and Transformational Creativity; An analysis of MEXICA

"Music variations consists of modifying an original music piece so that it remains harmonious and provides a sense of novelty while still remain relatable to the original song. The melody should progress in a harmonious way and under a reasonable chord hierarchy, due to this constraints, music 
variations can be considered as a rule based composition system. In this research, we develop a genetic algorithm to compose a variation on a given music piece. To achieve this we design a novel evaluation function and evolutionary operators which favor the modification process. These are designed to include components of melody similarity and tempo diversity in addition to harmony. The experimental results show that the proposed  system can generate variations that preserve musicality. A human evaluation study is also included to validate the proposed evaluation function. Additionally a link to listen to our generated compositions is provided."

Melody Similarity and Tempo Diversity as Evolutionary Factors for Music Variations by Genetic Algorithms

The automated evaluation of creative products promises both good-and-scalable creativity assessments and new forms of visual analysis of whole corpora. Where creative works are not ‘born digital’, such automated evaluation requires fast and frugal ways of transforming them into data representations that can be meaningfully assessed with common creativity metrics like novelty. In this paper, we report the results of training a Spatiotemporal DeepInfomax Variational Autoencoder (STDIM-VAE) on a digital photo pool of 162 LEGO ducks to generate a phenotypical landscape of clusters of similar ducks and dissimilarity scores for individual ducks. Visual inspection suggests that our system produces plausible results from image pixels alone. We conclude that under certain conditions, STDIM-VAEs may provide fast and frugal ways of automatically assessing corpora of creative works.

Automatic Similarity Detection in LEGO Ducks

Natural language generation provides designers with methods for automatically generating text, e.g. for creating summaries, chatbots and game content. In practise, text generators are often either learned and hard to interpret, or created by hand using techniques such as grammars and templates. In this paper, we introduce a novel grammar induction algorithm for learning interpretable grammars for generative purposes, called GITTA. We also introduce the novel notion of template trees to discover latent templates in corpora to derive these generative grammars. By using existing human-created grammars, we found that the algorithm can reasonably approximate these grammars using only a few examples. These results indicate that GITTA could be used to automatically learn interpretable and easily modifiable grammars, and thus provide a stepping stone for human-machine co-creation of generative models.

Discovering Textual Structures: Generative Grammar Induction using Template Trees

We introduce CANIS, our computationally creative system for generating novel constellations. We explain the motivations behind its creation, how it works, and demonstrate several of its artifacts. 

Creative Constellation Generation

"Humor is incorporated into our daily interactions, but conceiving jokes ideas can be inherently difficult to produce spontaneously. The PAUL BOT system, “puns are usually lame but occasionally terrific”, includes features of the JAPE system by incorporating aspects of the SAD generator, templates, and sentence forms to automatically generate a pun (Ritchie 2003). A two-word database is composed of adjective-noun phrases that contain a homophone. The punchline to the pun is replacing a word in the two- word phrase with a homophone. A synonym is derived from the homophone, and a hypernym is chosen from the non- homophone word in the original phrase. The synonym and the hypernym are incorporated into a predetermined sentence structure to construct the question. Surveys were conducted to evaluate our artifacts produced by PAUL BOT in order to identify future improvements of the system. The creativity in the system is attributed to the novelty of the unique artifacts, the level of surprise, the typicality of the artifacts as classic puns, and the intentionality by providing the connection between the setup and punchline. This paper proposes the PAUL BOT system that incorporates the JAPE system to create puns in order to produce a humorous, creative system utilized for entertainment.
Source: github.com/knw7x9/PunGenerator"

Computational Humor: Automated Pun Generation

"We present ERwEM (Events Represented with Emotive Music), a system that generates cross-domain artefacts between text and music by drawing from emotional sentiments. Our system uses keywords extracted from a
news article to filter live-stream tweets containing these keywords and analyze the tweet collection’s emotion data. It then uses the emotion sentiment (a scale from positive to negative) to influence the mood of the music it writes using an adapted bi-axial long short term memory (BALSTM) model. From this, our system determines
which mode the generated song will be transposed to. We use the FACE model to characterize the creative aspects of our system as a collection of generative acts. We provide examples of the artefacts our system generates as well as the response we received from external sources who reviewed the generated songs."

ERwEM: Events Represented with Emotive Music Using Topic-Filtered Tweets

"We propose three methods for the adaptation of well-known sayings~(e.g., proverbs, movie titles) for a context, given by a textual input~(e.g. news headline).
These methods -- word substitution, analogy and vector difference -- present different ways of exploiting word embeddings for word replacement towards a new text that should be semantically related to the input, in the sense that it could be used as a sub-title or comment, though more creative.
They were further combined with two selection methods, based on word overlap and on sentence embeddings, and used in the production of text in context with a small set of Portuguese headlines. To better understand how well our purpose was suit, results were manually assessed. All methods produced text with both syntax and relatedness to the input above average, contrasting the underachieving funniness scores."

Exploring Word Embeddings for Text Adaptation to a given Context

In this paper, we present a system that generates letterpress-inspired typographic posters from a user-given text. This system employs lexicon-based approaches not only to recognise emotions and sentiments but also the colours related to the content. Moreover, the system generates and evolves outputs employing an Evolutionary Computation approach. The generated outputs are evaluated according to their legibility, aesthetics and semantics, throughout a multi-criteria hardwired fitness function scheme. Also, this system enables its users to guide the generation process through the dynamic definition of several setting parameters.

Neuro-Symbolic Generative Art: A Preliminary Study

Downloads

Next from ICCC'20

Automated Music Generation for Visual Art through Emotion

Similar lecture

Adversarial Learning on the Latent Space for Diverse Dialog Generation