-
Collaborative Control for Geometry-Conditioned PBR Image Generation
Authors:
Shimon Vainer,
Mark Boss,
Mathias Parger,
Konstantin Kutsy,
Dante De Nigris,
Ciara Rowles,
Nicolas Perony,
Simon Donné
Abstract:
Current 3D content generation approaches build on diffusion models that output RGB images. Modern graphics pipelines, however, require physically-based rendering (PBR) material properties. We propose to model the PBR image distribution directly, avoiding photometric inaccuracies in RGB generation and the inherent ambiguity in extracting PBR from RGB. Existing paradigms for cross-modal fine-tuning…
▽ More
Current 3D content generation approaches build on diffusion models that output RGB images. Modern graphics pipelines, however, require physically-based rendering (PBR) material properties. We propose to model the PBR image distribution directly, avoiding photometric inaccuracies in RGB generation and the inherent ambiguity in extracting PBR from RGB. Existing paradigms for cross-modal fine-tuning are not suited for PBR generation due to both a lack of data and the high dimensionality of the output modalities: we overcome both challenges by retaining a frozen RGB model and tightly linking a newly trained PBR model using a novel cross-network communication paradigm. As the base RGB model is fully frozen, the proposed method does not risk catastrophic forgetting during fine-tuning and remains compatible with techniques such as IPAdapter pretrained for the base RGB model. We validate our design choices, robustness to data sparsity, and compare against existing paradigms with an extensive experimental section.
△ Less
Submitted 20 February, 2024; v1 submitted 8 February, 2024;
originally announced February 2024.
-
Chimpanzee voice prints? Insights from transfer learning experiments from human voices
Authors:
Mael Leroux,
Orestes Gutierrez Al-Khudhairy,
Nicolas Perony,
Simon W. Townsend
Abstract:
Individual vocal differences are ubiquitous in the animal kingdom. In humans, these differences pervade the entire vocal repertoire and constitute a "voice print". Apes, our closest-living relatives, possess individual signatures within specific call types, but the potential for a unique voice print has been little investigated. This is partially attributed to the limitations associated with extra…
▽ More
Individual vocal differences are ubiquitous in the animal kingdom. In humans, these differences pervade the entire vocal repertoire and constitute a "voice print". Apes, our closest-living relatives, possess individual signatures within specific call types, but the potential for a unique voice print has been little investigated. This is partially attributed to the limitations associated with extracting meaningful features from small data sets. Advances in machine learning have highlighted an alternative to traditional acoustic features, namely pre-trained learnt extractors. Here, we present an approach building on these developments: leveraging a feature extractor based on a deep neural network trained on over 10,000 human voice prints to provide an informative space over which we identify chimpanzee voice prints. We compare our results with those obtained by using traditional acoustic features and discuss the benefits of our methodology and the significance of our findings for the identification of "voice prints" in non-human animals.
△ Less
Submitted 15 December, 2021;
originally announced December 2021.
-
The digital traces of bubbles: feedback cycles between socio-economic signals in the Bitcoin economy
Authors:
David Garcia,
Claudio Juan Tessone,
Pavlin Mavrodiev,
Nicolas Perony
Abstract:
What is the role of social interactions in the creation of price bubbles? Answering this question requires obtaining collective behavioural traces generated by the activity of a large number of actors. Digital currencies offer a unique possibility to measure socio-economic signals from such digital traces. Here, we focus on Bitcoin, the most popular cryptocurrency. Bitcoin has experienced periods…
▽ More
What is the role of social interactions in the creation of price bubbles? Answering this question requires obtaining collective behavioural traces generated by the activity of a large number of actors. Digital currencies offer a unique possibility to measure socio-economic signals from such digital traces. Here, we focus on Bitcoin, the most popular cryptocurrency. Bitcoin has experienced periods of rapid increase in exchange rates (price) followed by sharp decline; we hypothesise that these fluctuations are largely driven by the interplay between different social phenomena. We thus quantify four socio-economic signals about Bitcoin from large data sets: price on on-line exchanges, volume of word-of-mouth communication in on-line social media, volume of information search, and user base growth. By using vector autoregression, we identify two positive feedback loops that lead to price bubbles in the absence of exogenous stimuli: one driven by word of mouth, and the other by new Bitcoin adopters. We also observe that spikes in information search, presumably linked to external events, precede drastic price declines. Understanding the interplay between the socio-economic signals we measured can lead to applications beyond cryptocurrencies to other phenomena which leave digital footprints, such as on-line social network usage.
△ Less
Submitted 7 August, 2014;
originally announced August 2014.
-
Dynamical coupling during collective animal motion
Authors:
Thomas O. Richardson,
Nicolas Perony,
Claudio J. Tessone,
Christophe A. H. Bousquet,
Marta B. Manser,
Frank Schweitzer
Abstract:
The measurement of information flows within moving animal groups has recently been a topic of considerable interest, and it has become clear that the individual(s) that drive collective movement may change over time, and that such individuals may not necessarily always lead from the front. However, methods to quantify the influence of specific individuals on the behaviour of other group members an…
▽ More
The measurement of information flows within moving animal groups has recently been a topic of considerable interest, and it has become clear that the individual(s) that drive collective movement may change over time, and that such individuals may not necessarily always lead from the front. However, methods to quantify the influence of specific individuals on the behaviour of other group members and the direction of information flow in moving group, are lacking on the level of empirical studies and theoretical models. Using high spatio-temporal resolution GPS trajectories of foraging meerkats, Suricata suricatta, we provide an information-theoretic framework to identify dynamical coupling between animals independent of their relative spatial positions. Based on this identification, we then compare designations of individuals as either drivers or responders against designations provided by the relative spatial position. We find that not only does coupling occur both from the frontal to the trailing individuals and vice versa, but also that the coupling direction is a non-linear function of the relative position. This provides evidence for (i) intermittent fluctuation of the coupling strength and (ii) alternation in the coupling direction within foraging meerkat pairs. The framework we introduce allows for a detailed description of the dynamical patterns of mutual influence between all pairs of individuals within moving animal groups. We argue that applying an information-theoretic perspective to the study of coordinated phenomena in animal groups will eventually help to understand cause and effect in collective behaviour.
△ Less
Submitted 6 November, 2013;
originally announced November 2013.
-
A stochastic model of social interaction in wild house mice
Authors:
Nicolas Perony,
Barbara König,
Frank Schweitzer
Abstract:
We investigate to what extent the interaction dynamics of a population of wild house mouse (Mus musculus domesticus) in their environment can be explained by a simple stochastic model. We use a Markov chain model to describe the transitions of mice in a discrete space of nestboxes, and implement a multi-agent simulation of the model. We find that some important features of our behavioural dataset…
▽ More
We investigate to what extent the interaction dynamics of a population of wild house mouse (Mus musculus domesticus) in their environment can be explained by a simple stochastic model. We use a Markov chain model to describe the transitions of mice in a discrete space of nestboxes, and implement a multi-agent simulation of the model. We find that some important features of our behavioural dataset can be reproduced using this simplified stochastic representation, and discuss the improvements that could be made to our model in order to increase the accuracy of its predictions. Our findings have implications for the understanding of the complexity underlying social behaviour in the animal kingdom and the cognitive requirements of such behaviour.
△ Less
Submitted 4 December, 2012;
originally announced December 2012.
-
Hierarchical Consensus Formation Reduces the Influence of Opinion Bias
Authors:
Nicolas Perony,
René Pfitzner,
Ingo Scholtes,
Claudio J. Tessone,
Frank Schweitzer
Abstract:
We study the role of hierarchical structures in a simple model of collective consensus formation based on the bounded confidence model with continuous individual opinions. For the particular variation of this model considered in this paper, we assume that a bias towards an extreme opinion is introduced whenever two individuals interact and form a common decision. As a simple proxy for hierarchical…
▽ More
We study the role of hierarchical structures in a simple model of collective consensus formation based on the bounded confidence model with continuous individual opinions. For the particular variation of this model considered in this paper, we assume that a bias towards an extreme opinion is introduced whenever two individuals interact and form a common decision. As a simple proxy for hierarchical social structures, we introduce a two-step decision making process in which in the second step groups of like-minded individuals are replaced by representatives once they have reached local consensus, and the representatives in turn form a collective decision in a downstream process. We find that the introduction of such a hierarchical decision making structure can improve consensus formation, in the sense that the eventual collective opinion is closer to the true average of individual opinions than without it. In particular, we numerically study how the size of groups of like-minded individuals being represented by delegate individuals affects the impact of the bias on the final population-wide consensus. These results are of interest for the design of organisational policies and the optimisation of hierarchical structures in the context of group decision making.
△ Less
Submitted 21 July, 2012; v1 submitted 30 March, 2012;
originally announced April 2012.
