Curiosity-driven phonetic learning

Moulin-Frier, Clément; Oudeyer, Pierre-Yves

doi:10.1109/devlrn.2012.6400583

Cited by 25 publications

(27 citation statements)

References 21 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Such systems will naturally and spontaneously move from one stage of development to the next, from one level of complexity to another, as error reduction becomes less available (either all that's left is uninteresting noise or the complexity is yet too high to be managed given the skill level of the organism). A neat outcome of such systems is the self-organization of developmental and learning trajectories that naturally move agents from acquiring simple to more complex skills over time (Oudeyer and Kaplan 2006;Oudeyer et al 2007;Kaplan and Oudeyer 2011;Moulin-Frier and Oudeyer 2012).…”

Section: Novelty Seeking and Learning Progressmentioning

confidence: 99%

The feeling of grip: novelty, error dynamics, and the predictive brain

2017

View full text Add to dashboard Cite

According to the free energy principle biological agents resist a tendency to disorder in their interactions with a dynamically changing environment by keeping themselves in sensory and physiological states that are expected given their embodiment and the niche they inhabit (Friston in Nat Rev Neurosci 11(2):127-138, 2010. doi:10.1038/nrn2787). Why would a biological agent that aims at minimising uncertainty in its encounters with the world ever be motivated to seek out novelty? Novelty for such an agent would arrive in the form of sensory and physiological states that are unexpected. Such an agent ought therefore to avoid novel and surprising interactions with the world one might think. Yet humans and many other animals find play and other forms of novelty-seeking and exploration hugely rewarding. How can this be understood in frameworks for studying the mind that emphasise prediction error minMany thanks to Andy Clark for excellent feedback and suggestions on earlier drafts of this paper, and to Jelle Bruineberg for helping us to understand the recent literature in cognitive neuroscience on epistemic value. We are extremely grateful for detailed reviews we received from two anonymous reviewers that helped us to significantly improve and sharpen our arguments. Erik Rietveld would like to acknowledge the research funding awarded by the Netherlands Organisation for Scientific Research (NWO) 123Synthese imisation? This problem has been taken up in recent research concerned with epistemic action-actions an agent engages in to reduce uncertainty. However that work leaves two questions unanswered, which it is the aim of our paper to address. First, no account has been given yet of why it should feel good to the agent to engage the world playfully and with curiosity. Second an appeal is made to precision-estimation to explain epistemic action, yet it remains unclear how precision-weighting works in action more generally, or active inference. We argue that an answer to both questions may lie in the bodily states of an agent that track the rate at which free energy is being reduced. The recent literature on the predictive brain has connected the valence of emotional experiences to the rate of change in the reduction of prediction error (Joffily and Coricelli in PLoS Comput Biol 9(6):e1003094, 2013. doi:10.1371/journal.pcbi.1003094; Van de Cruys, in Metzinger and Wiese (eds) Philosophy and predictive processing, vol 24, MIND Group, Frankfurt am Main, 2017. doi:10.15502/9783958573253). In this literature valenced emotional experiences are hypothesised to be identical with changes in the rate at which prediction error is reduced. Experiences are negatively valenced when overall prediction error increases and are positively valenced when the sum of prediction errors decrease. We offer an ecological-enactive interpretation of the concept of valence and its connection to rate of change of prediction error. We show how rate of change should be understood in terms of embodied states of affordance-related action read...

show abstract

Section: Novelty Seeking and Learning Progressmentioning

confidence: 99%

The feeling of grip: novelty, error dynamics, and the predictive brain

2017

View full text Add to dashboard Cite

show abstract

“…Instead of exploring in the space of motor parameters, babbling can also be organized in the space of outcomes [7], [8]. In [9] it was shown that such goal babbling can bootstrap vowel sounds quicker than motor babbling. Goal babbling has subsequently been successfully applied for learning F0 contours [10] or for modeling the emergence of speech-like sounds in general [11], [12].…”

Section: Introductionmentioning

confidence: 99%

Hyperarticulation aids learning of new vowels in a developmental speech acquisition model

Philippsen

Reinhart

Wrede

et al. 2017

2017 International Joint Conference on Neural Networks (IJCNN)

View full text Add to dashboard Cite

Abstract-Many studies emphasize the importance of infantdirected speech: stronger articulated, higher-quality speech helps infants to better distinguish different speech sounds. This effect has been widely investigated in terms of the infant's perceptual capabilities, but few studies examined whether infantdirected speech has an effect on articulatory learning.In earlier studies, we developed a model that learns articulatory control for a 3D vocal tract model via goal babbling. Exploration is organized in the space of outcomes. This so called goal space is generated from a set of ambient speech sounds. Similarly to how speech from the environment shapes infant's speech perception, the data from which the goal space is learned shapes the later learning process: it determines which sounds the model is able to discriminate, and thus, which sounds it can eventually learn to produce.We investigate how speech sound quality in early learning affects the model's capability to learn new vowel sounds. The model is trained either on hyperarticulated (tense) or on hypoarticulated (lax) vowels. Then we retrain the model with vowels from the other set.Results show that new vowels can be acquired although they were not included in early learning. There is, however, an effect of learning order, showing that models first trained on the stronger articulated tense vowels easier accommodate to new vowel sounds later on.

show abstract

“…Studies by Moulin-Frier et al [13], [14], [15], [16] have applied the idea of goal babbling to the speech domain. By using formant frequencies as goals, they could demonstrate the emergence of articulated speech sounds [16] and the bootstrapping of vowel sounds [13].…”

Section: Introductionmentioning

confidence: 99%

“…By using formant frequencies as goals, they could demonstrate the emergence of articulated speech sounds [16] and the bootstrapping of vowel sounds [13]. In a recent work [17], Liu and Xu used goal babbling to control F0 for Chinese language.…”

Section: Introductionmentioning

confidence: 99%

Goal babbling of acoustic-articulatory models with adaptive exploration noise

Philippsen

Reinhart

Wrede

2016

2016 Joint IEEE International Conference on Development and Learning and Epigenetic Robotics (ICDL-EpiRob)

View full text Add to dashboard Cite

Abstract-We use goal babbling to bootstrap a parametric model of speech production for a complex 3D vocal tract model. The system learns to control the articulators for producing five different vowel sounds. Ambient speech influences learning on two levels: it organizes the learning process because it is used to generate a space of goals in which exploration takes place. A distribution learned from ambient speech provides the system with targets during exploration.Previous research with this vocal tract model showed that visual information have to be included for acquiring the vowel [u] via reward-based optimization. We model the learning process instead with goal-directed exploration where all targets are learned in parallel. As some vowels require more exploratory noise in the articulators than others, we propose a mechanism to adapt the noise amplitude depending on the system's competence in different regions of the goal space. We demonstrate that this self-aware learning leads to more stable results. The implemented system succeeds in acquiring vocalization skills for rounded as well as unrounded vowels using only a single modality.

show abstract

Curiosity-driven phonetic learning

Cited by 25 publications

References 21 publications

The feeling of grip: novelty, error dynamics, and the predictive brain

The feeling of grip: novelty, error dynamics, and the predictive brain

Hyperarticulation aids learning of new vowels in a developmental speech acquisition model

Goal babbling of acoustic-articulatory models with adaptive exploration noise

Contact Info

Product

Resources

About