Unsupervised Learning and Adaptive Classification of Neuromorphic Tactile Encoding of Textures

Iskarous, Mark M.; Nguyen, Harrison; Osborn, Luke; Betthauser, Joseph L.; Thakor, Nitish V.

doi:10.1109/biocas.2018.8584702

Cited by 17 publications

(5 citation statements)

References 13 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Unsupervised learning methods have also been applied to texture classification tasks [10]. This work achieved a high accuracy of mean = 86.46, albeit on a small number of textures (n = 3).…”

Section: A Texture Classificationmentioning

confidence: 99%

See 1 more Smart Citation

Temporal and Spatio-temporal domains for Neuromorphic Tactile Texture Classification

Brayshaw

Ward-Cherrier

Pearson

2022

Neuro-Inspired Computational Elements Conference

View full text Add to dashboard Cite

The development of upper limb prosthesis that are able to relay information on their status back to the user is an important step towards making this assistive technology more intuitive. Applied within this context, neuromorphic hardware has the potential to reduce processing time while simultaneously reducing power requirements. Towards this, we have begun a systematic evaluation of algorithms that best leverage rich neuromorphic data, and how such algorithms may be implemented. In this paper, we apply conventional machine learning techniques to temporal domain representations of textures derived from a neuromorphic tactile sensor. We then contrast these results with those from a novel spatio-temporal domain classification approach, the Hierarchy of Event-Based Time-Surfaces (HOTS). We achieved higher accuracies when classifying temporal data with our supervised learning methods (91% with a KNN) than when classifying with HOTS (76% with a single layer), indicating that simple temporal encoding is sufficient for the classification of texture.

show abstract

Section: A Texture Classificationmentioning

confidence: 99%

“…While the use of neuromorphic texture classification has been explored previously [5], [10], [11], a direct comparison of different classification methods for neuromorphic data, using the same tactile sensor for each, is yet to be presented. Within this paper we present the results from the following:…”

Section: Introductionmentioning

confidence: 99%

Temporal and Spatio-temporal domains for Neuromorphic Tactile Texture Classification

Brayshaw

Ward-Cherrier

Pearson

2022

Neuro-Inspired Computational Elements Conference

View full text Add to dashboard Cite

show abstract

“…However, the critical inversion operation is still calculated on a conventional computer. The feasibility of using the extracted spiking signal features was further explored for combination with unsupervised clustering [161] and a sparse coding classifier [162]. These studies realized the neuromorphic construction and integration of tactile sensors, encoding interfaces, and feature extraction, and took an important step for the online application of tactile neuromorphic processing.…”

Section: Tactile Neuromorphic Computingmentioning

confidence: 99%

Embodied tactile perception and learning

Liu

Guo

Sun

et al. 2020

Brain Science Advances

View full text Add to dashboard Cite

Various living creatures exhibit embodiment intelligence, which is reflected by a collaborative interaction of the brain, body, and environment. The actual behavior of embodiment intelligence is generated by a continuous and dynamic interaction between a subject and the environment through information perception and physical manipulation. The physical interaction between a robot and the environment is the basis for realizing embodied perception and learning. Tactile information plays a critical role in this physical interaction process. It can be used to ensure safety, stability, and compliance, and can provide unique information that is difficult to capture using other perception modalities. However, due to the limitations of existing sensors and perception and learning methods, the development of robotic tactile research lags significantly behind other sensing modalities, such as vision and hearing, thereby seriously restricting the development of robotic embodiment intelligence. This paper presents the current challenges related to robotic tactile embodiment intelligence and reviews the theory and methods of robotic embodied tactile intelligence. Tactile perception and learning methods for embodiment intelligence can be designed based on the development of new large‐scale tactile array sensing devices, with the aim to make breakthroughs in the neuromorphic computing technology of tactile intelligence.

show abstract

“…The dataset, composed of analog signals recorded from 9 different sensors, is converted to spikes using the Izhikevich neuron model [9]. A variety of algorithms was applied to the dataset: extreme learning machines [10], sparse representation classification [8], unsupervised clustering [11] and support vector machines [12]. The textures chosen in these works are coarse and therefore more suited for the approach related to the spatial distribution of the sensors on the artificial skin.…”

Section: Introductionmentioning

confidence: 99%

Texture Recognition Using a Biologically Plausible Spiking Phase-Locked Loop Model for Spike Train Frequency Decomposition

Mastella,

Tiemens,

Chicca

2024

Preprint

View full text Add to dashboard Cite

Neural spikes can encode a rich set of information, ranging from the perceived intensity of light sources to the likelihood associated with decisions made in the cortex. Among these capabilities, previous studies demonstrated that spikes can also encode in their activity multiple frequencies at the same time, such as those generated by skin vibrations during textures scanning. However, the mechanism responsible for decoding spikes containing multiple frequencies is yet to be uncovered. In this paper, we introduce a novel spiking neural network model tailored for frequency decomposition of spike trains. Our model mimics neural microcircuits hypothesized in the somatosensory cortex, making it a biologically plausible candidate for decoding spike trains observed in tactile peripheral nerves. We showcase the ability of simple neurons and synapses to replicate the functionality of a phase-locked loop (PLL) and delve into the emergent properties when multiple spiking phase-locked loops (sPLLs) interact with diverse inputs. Furthermore, we demonstrate how these sPLLs can decode textures by leveraging the spectral features of spike trains generated in peripheral nerves. By harnessing our model's frequency decomposition capabilities, we achieve significant performance enhancements over state-of-the-art approaches on a Multifrequency Spike Train (MST) dataset. Our findings underscore the potential of sPLLs in elucidating the mechanisms behind texture decoding in the brain, while also showcasing their potential to outperform conventional SNNs in handling spike trains with multiple frequencies. We believe this study sheds light into the neuronal mechanisms behind texture decoding, while presenting a practical framework for augmenting the capabilities of artificial neural networks in intricate pattern recognition tasks.

show abstract

Unsupervised Learning and Adaptive Classification of Neuromorphic Tactile Encoding of Textures

Cited by 17 publications

References 13 publications

Temporal and Spatio-temporal domains for Neuromorphic Tactile Texture Classification

Temporal and Spatio-temporal domains for Neuromorphic Tactile Texture Classification

Embodied tactile perception and learning

Texture Recognition Using a Biologically Plausible Spiking Phase-Locked Loop Model for Spike Train Frequency Decomposition

Contact Info

Product

Resources

About