Activities per year
Project Details
Description
Lrn2Cre8 aims to understand the relationship between learning and creativity by means of practical engineering, theoretical study, and cognitive comparison. We begin from the position that creativity is a function of memory, that generates new structures based on memorised ones, by processes which are essentially statistical. Thus, the project is situated in the tradition of frequentist statistical models of mind, and it builds on statistical understandings of perception of sequence (these are not naive 1st-order statistics) to consider its domain at the level of sequence processing in terms of percepts. Representations of these percepts are also learned, in parallel with the structural information in the data itself, and the guiding principle is one of information efficiency: the representations conspire to produce the most efficient possible representation of the data in memory. In Lre2Cre8, we wish to avoid the problem of extrinsic domain reasoning (e.g., physics in the real world) so we use music as our domain; extrinsic reasoning for music is small in comparison. We propose to build systems that take musical data as input, both in continuous and discrete forms, and learn the necessary representations and structure to memorise it efficiently. We hypothesise that this is a cognitive model of human musical behaviour, and we will test our hypthesis though empirical studies and experiments that compare the behaviour of our computational models with human behaviours. We will study the relationship between our well-understood and cognitively validated learning mechanisms and creative behaviour, in musical composition and performance. We aim to devise new methods for evaluation of creative behaviour in machines and humans, and to apply them, comparatively, to creative processes and outputs of the project. We aim ultimately to produce music which will be of genuine interest to society, and we will launch a record label to promote it as part of our evaluation methods.
Acronym | Lrn2Cre8 |
---|---|
Status | Finished |
Effective start/end date | 01/10/2013 → 30/09/2016 |
Collaborative partners
- Queen Mary and Westfield College, University of London (Project partner) (lead)
- Re-Compose GMBH (Project partner)
- Universidad del Pais Vasco EHU UPV (Project partner)
- Sony Group Corporation (Project partner)
- Oesterreichische Studiengesellschaft für Kybernetik (Project partner)
Funding
- EU Seventh Framework Programme (FP7): DKK18,600,000.00
Keywords
- Computational creativity
- Machine learning
- Music composition and performance
Fingerprint
Explore the research topics touched on by this project. These labels are generated based on the underlying awards/grants. Together they form a unique fingerprint.
Activities
- 1 Talks and presentations in private or public companies
-
Geometric, compression-based pattern discovery in music
David Meredith (Lecturer)
18 Sept 2016 → 22 Sept 2016Activity: Talks and presentations › Talks and presentations in private or public companies
File
Research output
-
Music analysis and data compression
Meredith, D., 26 Sept 2019, Oxford Handbook of Sound and Imagination. Grimshaw-Aagaard, M., Walther-Hansen, M. & Knakkegaard, M. (eds.). Oxford University Press, Vol. 2.Research output: Contribution to book/anthology/report/conference proceeding › Book chapter › Research › peer-review
-
Convolution-based classification of audio and symbolic representations of music
Velarde, G., Cancino Chacón, C., Meredith, D., Weyde, T. & Grachten, M., 27 May 2018, In: Journal of New Music Research. 47, 3, p. 191-205 15 p.Research output: Contribution to journal › Journal article › Research › peer-review
Open AccessFile10 Citations (Scopus)121 Downloads (Pure) -
Generating new musical works in the style of Milton Babbitt
Bemman, B. & Meredith, D., 1 Apr 2018, In: Computer Music Journal. 42, 1, p. 60-79 20 p.Research output: Contribution to journal › Journal article › Research › peer-review
1 Citation (Scopus)