Unsupervised learning of haptic material properties

Authors: Metzger, A. and Toscani, M.

Journal: eLife

Volume: 11

eISSN: 2050-084X

DOI: 10.7554/ELIFE.64876

Abstract:

When touching the surface of an object, its spatial structure translates into a vibration on the skin. The perceptual system evolved to translate this pattern into a representation that allows to distinguish between different materials. Here, we show that perceptual haptic representation of materials emerges from efficient encoding of vibratory patterns elicited by the interaction with materials. We trained a deep neural network with unsupervised learning (Autoencoder) to recon-struct vibratory patterns elicited by human haptic exploration of different materials. The learned compressed representation (i.e., latent space) allows for classification of material categories (i.e., plastic, stone, wood, fabric, leather/wool, paper, and metal). More importantly, classification performance is higher with perceptual category labels as compared to ground truth ones, and distances between categories in the latent space resemble perceptual distances, suggesting a similar coding. Crucially, the classification performance and the similarity between the perceptual and the latent space decrease with decreasing compression level. We could further show that the temporal tuning of the emergent latent dimensions is similar to properties of human tactile receptors.

https://eprints.bournemouth.ac.uk/36687/

Source: Scopus

Unsupervised learning of haptic material properties.

Authors: Metzger, A. and Toscani, M.

Journal: Elife

Volume: 11

eISSN: 2050-084X

DOI: 10.7554/eLife.64876

Abstract:

When touching the surface of an object, its spatial structure translates into a vibration on the skin. The perceptual system evolved to translate this pattern into a representation that allows to distinguish between different materials. Here, we show that perceptual haptic representation of materials emerges from efficient encoding of vibratory patterns elicited by the interaction with materials. We trained a deep neural network with unsupervised learning (Autoencoder) to reconstruct vibratory patterns elicited by human haptic exploration of different materials. The learned compressed representation (i.e., latent space) allows for classification of material categories (i.e., plastic, stone, wood, fabric, leather/wool, paper, and metal). More importantly, classification performance is higher with perceptual category labels as compared to ground truth ones, and distances between categories in the latent space resemble perceptual distances, suggesting a similar coding. Crucially, the classification performance and the similarity between the perceptual and the latent space decrease with decreasing compression level. We could further show that the temporal tuning of the emergent latent dimensions is similar to properties of human tactile receptors.

https://eprints.bournemouth.ac.uk/36687/

Source: PubMed

Unsupervised learning of haptic material properties

Authors: Metzger, A. and Toscani, M.

Journal: ELIFE

Volume: 11

ISSN: 2050-084X

DOI: 10.7554/eLife.64876

https://eprints.bournemouth.ac.uk/36687/

Source: Web of Science (Lite)

Unsupervised learning of haptic material properties

Authors: Metzger, A. and Toscani, M.

Journal: eLife

Volume: 11

eISSN: 2050-084X

DOI: 10.7554/ELIFE.64876

Abstract:

When touching the surface of an object, its spatial structure translates into a vibration on the skin. The perceptual system evolved to translate this pattern into a representation that allows to distinguish between different materials. Here, we show that perceptual haptic representation of materials emerges from efficient encoding of vibratory patterns elicited by the interaction with materials. We trained a deep neural network with unsupervised learning (Autoencoder) to recon-struct vibratory patterns elicited by human haptic exploration of different materials. The learned compressed representation (i.e., latent space) allows for classification of material categories (i.e., plastic, stone, wood, fabric, leather/wool, paper, and metal). More importantly, classification performance is higher with perceptual category labels as compared to ground truth ones, and distances between categories in the latent space resemble perceptual distances, suggesting a similar coding. Crucially, the classification performance and the similarity between the perceptual and the latent space decrease with decreasing compression level. We could further show that the temporal tuning of the emergent latent dimensions is similar to properties of human tactile receptors.

https://eprints.bournemouth.ac.uk/36687/

Source: Manual

Preferred by: Anna Metzger

Unsupervised learning of haptic material properties.

Authors: Metzger, A. and Toscani, M.

Journal: eLife

Volume: 11

Pages: e64876

eISSN: 2050-084X

ISSN: 2050-084X

DOI: 10.7554/elife.64876

Abstract:

When touching the surface of an object, its spatial structure translates into a vibration on the skin. The perceptual system evolved to translate this pattern into a representation that allows to distinguish between different materials. Here, we show that perceptual haptic representation of materials emerges from efficient encoding of vibratory patterns elicited by the interaction with materials. We trained a deep neural network with unsupervised learning (Autoencoder) to reconstruct vibratory patterns elicited by human haptic exploration of different materials. The learned compressed representation (i.e., latent space) allows for classification of material categories (i.e., plastic, stone, wood, fabric, leather/wool, paper, and metal). More importantly, classification performance is higher with perceptual category labels as compared to ground truth ones, and distances between categories in the latent space resemble perceptual distances, suggesting a similar coding. Crucially, the classification performance and the similarity between the perceptual and the latent space decrease with decreasing compression level. We could further show that the temporal tuning of the emergent latent dimensions is similar to properties of human tactile receptors.

https://eprints.bournemouth.ac.uk/36687/

Source: Europe PubMed Central

Unsupervised learning of haptic material properties.

Authors: Metzger, A. and Toscani, M.

Journal: eLife

Volume: 11

ISSN: 2050-084X

Abstract:

When touching the surface of an object, its spatial structure translates into a vibration on the skin. The perceptual system evolved to translate this pattern into a representation that allows to distinguish between different materials. Here, we show that perceptual haptic representation of materials emerges from efficient encoding of vibratory patterns elicited by the interaction with materials. We trained a deep neural network with unsupervised learning (Autoencoder) to reconstruct vibratory patterns elicited by human haptic exploration of different materials. The learned compressed representation (i.e., latent space) allows for classification of material categories (i.e., plastic, stone, wood, fabric, leather/wool, paper, and metal). More importantly, classification performance is higher with perceptual category labels as compared to ground truth ones, and distances between categories in the latent space resemble perceptual distances, suggesting a similar coding. Crucially, the classification performance and the similarity between the perceptual and the latent space decrease with decreasing compression level. We could further show that the temporal tuning of the emergent latent dimensions is similar to properties of human tactile receptors.

https://eprints.bournemouth.ac.uk/36687/

Source: BURO EPrints