Efficient movement representation by embedding Dynamic Movement Primitives in deep autoencoders

Chen, Nutan; Bayer, Justin; Urban, Sebastian; Smagt, Patrick van der

doi:10.1109/humanoids.2015.7363570

Cited by 41 publications

(36 citation statements)

References 9 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Artificial neural networks have, with the enormous scientific and economical success of deep learning in particular and of artificial intelligence in general, strongly facilitated the view that also biological neuronal networks can approximate general transformations which adjust the output of the network to arbitrarily changing conditions (Chen et al, 2015). In mathematical terms, they are said to perform universal classification and function approximation.…”

Section: The Spinal Cord As Transformer Of World Viewsmentioning

confidence: 99%

“…Functional insights on the use of synergies can be obtained from robotics control theory. In recent work, an artificial neural network, which formed a similar network structure as a sensorimotor synergy, was trained to encode meaningful motor primitives within the intermediate synergy layer (Chen et al, 2015). The underlying type of artificial neural networks is called autoencoder and is typically used in in the field of deep learning to reduce the dimensionality of data (Hinton and Salakhutdinov, 2006).…”

Section: Lightening the Burden Of Freedommentioning

confidence: 99%

See 1 more Smart Citation

Scaling Our World View: How Monoamines Can Put Context Into Brain Circuitry

Stratmann

Albu‐Schäffer

Jörntell

2018

Front. Cell. Neurosci.

View full text Add to dashboard Cite

Monoamines are presumed to be diffuse metabotropic neuromodulators of the topographically and temporally precise ionotropic circuitry which dominates CNS functions. Their malfunction is strongly implicated in motor and cognitive disorders, but their function in behavioral and cognitive processing is scarcely understood. In this paper, the principles of such a monoaminergic function are conceptualized for locomotor control. We find that the serotonergic system in the ventral spinal cord scales ionotropic signals and shows topographic order that agrees with differential gain modulation of ionotropic subcircuits. Whereas the subcircuits can collectively signal predictive models of the world based on life-long learning, their differential scaling continuously adjusts these models to changing mechanical contexts based on sensory input on a fast time scale of a few 100 ms. The control theory of biomimetic robots demonstrates that this precision scaling is an effective and resource-efficient solution to adapt the activation of individual muscle groups during locomotion to changing conditions such as ground compliance and carried load. Although it is not unconceivable that spinal ionotropic circuitry could achieve scaling by itself, neurophysiological findings emphasize that this is a unique functionality of metabotropic effects since recent recordings in sensorimotor circuitry conflict with mechanisms proposed for ionotropic scaling in other CNS areas. We substantiate that precision scaling of ionotropic subcircuits is a main functional principle for many monoaminergic projections throughout the CNS, implying that the monoaminergic circuitry forms a network within the network composed of the ionotropic circuitry. Thereby, we provide an early-level interpretation of the mechanisms of psychopharmacological drugs that interfere with the monoaminergic systems.

show abstract

Section: The Spinal Cord As Transformer Of World Viewsmentioning

confidence: 99%

Section: Lightening the Burden Of Freedommentioning

confidence: 99%

Scaling Our World View: How Monoamines Can Put Context Into Brain Circuitry

Stratmann

Albu‐Schäffer

Jörntell

2018

Front. Cell. Neurosci.

View full text Add to dashboard Cite

show abstract

“…Since the discovery of mirror neurons (Gallese et al, 1996;Rizzolatti et al, 1996), that react similarly for goal-oriented self-motions and for similar motions of others and thus allow estimations of the intentions of others, scientists have considered that the mirror neuron system may be impaired or possibly dysfunctional in autism (Williams et al, 2001). In other words, it was thought that in typical cases, mirror neurons can provide supervisory information for training and for copying behavioral templates, as in the case of successful models of the learning of dynamic movement primitives (Schaal, 2006;Ijspeert et al, 2013;Chen et al, 2015).…”

Section: Mirror Neuron Theorymentioning

confidence: 99%

Colors of Autism Spectrum: A Single Paradigm Explains the Heterogeneity in Autism Spectrum Disorder

Lörincz¹

2018

Preprint

View full text Add to dashboard Cite

We propose a theory of ASD as a condition of comorbid cognitive impairments that corrupt the learning, encoding, and manipulation of episodic and semantic memories. We consider (i) episodic and semantic memory functions of the entorhinal-hippocampal complex, (ii) constraints on the transfer and encoding of these memory components into neocortical areas, and (iii) the demands of cognitively manipulating memories in distributed computations being necessary for goaloriented interactions. In ASD, learning and cognitive challenges manifest in diverse ways but especially in high-complexity-model predictive control tasks with latent variables. ASD impairments in social interactions represent a prototypical example. Social interactions are at the high end of complexity and require processes (i) -(iii) to work in a concerted fashion due to the need for the learning and estimation of many, sometimes latent, parameters, including emotions, intention, physical and mental capabilities as well as the predictive modeling of these parameters for decision making and timed-action series. We put forth the idea that autism is a result of an arbitrary combination of otherwise not prominent corruptions in processes (i)-(iii). Together, these corruptions may severely impair intelligence and slow down learning, especially in highcomplexity learning tasks. Over time, slow learning may spare the spontaneous learning-by-doing method -namely, repetitive behavioral patterns, whereas behavioral failures related to complex tasks can restrict interest in such task, thus inducing a fear of novelty; conversely, the fear of novelty restricts interest and can slow learning down. We embed our thoughts into a predictive autoencoding, goal-oriented model of a deterministic world. We compare this model to others, such as the noisy brain model, the Bayesian prior theory, the mirror neuron theory and the weak central coherence theory. We argue that the predictive autoencoder model of the deterministic world harmonizes with these other models and embraces them in a straightforward way. AUTHOR SUMMARYThe core of our arguments is based on the autoencoder model of the entorhinal-hippocampal complex and its role in forming declarative memories. We argue that social behavior and language usage are highly complex cognitive tasks involving the encoding, learning, and manipulation of declarative memories with latent variables. All these processes can be corrupted in many ways. Corruptions may manifest as epilepsy, schizophrenia, attention deficit, and depression, among other disorders. However, if these symptoms are weak but co-occur, then the most strongly affected behavioral feature is the ability to solve complex cognitive tasks. This condition can be brought about by diverse combinations of impairments that sum up in corrupted social behavior the most.

show abstract

“…Herein, we use the behavioral approach to construct a natural and flexible theoretical framework for analyzing the input/output relationship by means of machine learning. First, we give some examples of previous research that used system behaviors and trajectories [34][35][36][37][38][39][40][41][42]. In one case, focusing on only human motion trajectories, an unlearned motion pattern was generated by learning two types of movement [37,40,41].…”

Section: Introductionmentioning

confidence: 99%

“…First, we give some examples of previous research that used system behaviors and trajectories [34][35][36][37][38][39][40][41][42]. In one case, focusing on only human motion trajectories, an unlearned motion pattern was generated by learning two types of movement [37,40,41]. In another case, the two basic stepping patterns in neonates were retained through development, augmented by two new patterns first revealed in toddlers [34].…”

Section: Introductionmentioning

confidence: 99%

Generalization of movements in quadruped robot locomotion by learning specialized motion data

et al. 2020

View full text Add to dashboard Cite

Machines that are sensitive to environmental fluctuations, such as autonomous and pet robots, are currently in demand, rendering the ability to control huge and complex systems crucial. However, controlling such a system in its entirety using only one control device is difficult; for this purpose, a system must be both diverse and flexible. Herein, we derive and analyze the feature values of robot sensor and actuator data, thereby investigating the role that each feature value plays in robot locomotion. We conduct experiments using a developed quadruped robot from which we acquire multi-point motion information as the movement data; we extract the features of these movement data using an autoencoder. Next, we decompose the movement data into three features and extract various gait patterns. Despite learning only the "walking" movement, the movement patterns of trotting and bounding are also extracted herein, which suggests that movement data obtained via hardware contain various gait patterns. Although the present robot cannot locomote with these movements, this research suggests the possibility of generating unlearned movements.

show abstract

Efficient movement representation by embedding Dynamic Movement Primitives in deep autoencoders

Cited by 41 publications

References 9 publications

Scaling Our World View: How Monoamines Can Put Context Into Brain Circuitry

Scaling Our World View: How Monoamines Can Put Context Into Brain Circuitry

Colors of Autism Spectrum: A Single Paradigm Explains the Heterogeneity in Autism Spectrum Disorder

Generalization of movements in quadruped robot locomotion by learning specialized motion data

Contact Info

Product

Resources

About

Efficient movement representation by embedding Dynamic Movement Primitives in deep autoencoders

Cited by 41 publications

References 9 publications

Scaling Our World View: How Monoamines Can Put Context Into Brain Circuitry

Scaling Our World View: How Monoamines Can Put Context Into Brain Circuitry

Colors of Autism Spectrum: A Single Paradigm Explains the Heterogeneity&nbsp;in Autism Spectrum Disorder

Generalization of movements in quadruped robot locomotion by learning specialized motion data

Contact Info

Product

Resources

About

Colors of Autism Spectrum: A Single Paradigm Explains the Heterogeneity in Autism Spectrum Disorder