Inspired by the way the human brain works, hyperdimensional computing (HDC) is an emerging computing paradigm that computes with patterns of neural activities instead of scalar numbers. Researchers have established the importance of sequential processing in the brain and the variable binding problems, whereby distinct sensory events are combined for perception, decision, and action. HDC is also inspired by such advances in modeling combinatorial binding within simulated neural systems using high-dimensional vector and related operations. Recently, HDC has demonstrated promising capabilities on various applications such as language classifications, biosignal processing, and sensory-motor active perception.
While HDC has been attracting attention globally in the research community, given its short history, HDC is still in its infancy and is expecting major advancements both theoretically and practically. This research topic aims to present the latest research work in hyperdimensional computing that is correlated with neuromorphic computing and/or neuroscience. Some major problems in the field include: (1) What are the killer applications of HDC, especially compared to traditional machine learning methods such as deep neural networks? What are the advantages of HDC that can be leveraged in different application domains? (2) What are the optimal encoding methods in HDC for arbitrary tasks of interests? How do the advances in neuroscience benefit the development of HDC? (3) How can HDC be combined, or augmented with alternative biologically-inspired computing paradigms such as deep neural networks, neuromorphic computing, or evolutionary algorithms, to create a highly accurate yet efficient intelligent machine?
We welcome manuscripts reporting novel research (experimental, numerical, and theoretical) on hyperdimensional computing that cater to the broad domain of neuromorphic computing. The topics of interest will comprise of (but will not be limited to) the following:
(1) Application of advancements in neuroscience and neuromorphic engineering in HDC algorithm design.
(2) Applications of HDC in critical fields such as healthcare, communications, robotics, transportation, computing, and consumer industry.
(3) Efficient integration of HDC with alternative paradigms for improved performance.
(4) Novel approaches at all levels in the learning pipeline of HD computing: encoding, training, and inference.
(5) Novel computing systems leveraging HDC.
(6) High-performance and/or energy-efficient circuit and/or architecture for HDC.
(7) Evaluation, characterization, and/or assessment of HDC in practical application domains, especially compared to traditional learning methods such as deep neural networks.
Topic editor Dr. Abbas Rahimi is employed by IBM Research. All other Topic Editors declare no competing interests with regards to the Research Topic subject.
Inspired by the way the human brain works, hyperdimensional computing (HDC) is an emerging computing paradigm that computes with patterns of neural activities instead of scalar numbers. Researchers have established the importance of sequential processing in the brain and the variable binding problems, whereby distinct sensory events are combined for perception, decision, and action. HDC is also inspired by such advances in modeling combinatorial binding within simulated neural systems using high-dimensional vector and related operations. Recently, HDC has demonstrated promising capabilities on various applications such as language classifications, biosignal processing, and sensory-motor active perception.
While HDC has been attracting attention globally in the research community, given its short history, HDC is still in its infancy and is expecting major advancements both theoretically and practically. This research topic aims to present the latest research work in hyperdimensional computing that is correlated with neuromorphic computing and/or neuroscience. Some major problems in the field include: (1) What are the killer applications of HDC, especially compared to traditional machine learning methods such as deep neural networks? What are the advantages of HDC that can be leveraged in different application domains? (2) What are the optimal encoding methods in HDC for arbitrary tasks of interests? How do the advances in neuroscience benefit the development of HDC? (3) How can HDC be combined, or augmented with alternative biologically-inspired computing paradigms such as deep neural networks, neuromorphic computing, or evolutionary algorithms, to create a highly accurate yet efficient intelligent machine?
We welcome manuscripts reporting novel research (experimental, numerical, and theoretical) on hyperdimensional computing that cater to the broad domain of neuromorphic computing. The topics of interest will comprise of (but will not be limited to) the following:
(1) Application of advancements in neuroscience and neuromorphic engineering in HDC algorithm design.
(2) Applications of HDC in critical fields such as healthcare, communications, robotics, transportation, computing, and consumer industry.
(3) Efficient integration of HDC with alternative paradigms for improved performance.
(4) Novel approaches at all levels in the learning pipeline of HD computing: encoding, training, and inference.
(5) Novel computing systems leveraging HDC.
(6) High-performance and/or energy-efficient circuit and/or architecture for HDC.
(7) Evaluation, characterization, and/or assessment of HDC in practical application domains, especially compared to traditional learning methods such as deep neural networks.
Topic editor Dr. Abbas Rahimi is employed by IBM Research. All other Topic Editors declare no competing interests with regards to the Research Topic subject.
