<> "The repository administrator has not yet configured an RDF license."^^ . <> . . "Harnessing function from form: towards\r\nbio-inspired artificial intelligence in\r\nneuronal substrates"^^ . "Despite the recent success of deep learning, the mammalian brain is still unrivaled when it comes\r\nto interpreting complex, high-dimensional data streams like visual, auditory and somatosensory stimuli.\r\nHowever, the underlying computational principles allowing the brain to deal with unreliable, high-dimensional\r\nand often incomplete data while having a power consumption on the order of a few watt are still mostly\r\nunknown.\r\nIn this work, we investigate how specific functionalities emerge from simple structures observed in the\r\nmammalian cortex, and how these might be utilized in non-von Neumann devices like “neuromorphic\r\nhardware”. Firstly, we show that an ensemble of deterministic, spiking neural networks can be shaped by\r\na simple, local learning rule to perform sampling-based Bayesian inference. This suggests a coding scheme\r\nwhere spikes (or “action potentials”) represent samples of a posterior distribution, constrained by sensory\r\ninput, without the need for any source of stochasticity. Secondly, we introduce a top-down framework where\r\nneuronal and synaptic dynamics are derived using a least action principle and gradient-based minimization.\r\nCombined, neurosynaptic dynamics approximate real-time error backpropagation, mappable to mechanistic\r\ncomponents of cortical networks, whose dynamics can again be described within the proposed framework.\r\nThe presented models narrow the gap between well-defined, functional algorithms and their biophysical\r\nimplementation, improving our understanding of the computational principles the brain might employ.\r\nFurthermore, such models are naturally translated to hardware mimicking the vastly parallel neural\r\nstructure of the brain, promising a strongly accelerated and energy-efficient implementation of powerful\r\nlearning and inference algorithms, which we demonstrate for the physical model system “BrainScaleS–1”."^^ . "2020" . . . . . . . . "Dominik"^^ . "Dold"^^ . "Dominik Dold"^^ . . . . . . "Harnessing function from form: towards\r\nbio-inspired artificial intelligence in\r\nneuronal substrates (Other)"^^ . . . . . . "medium.jpg"^^ . . . "Harnessing function from form: towards\r\nbio-inspired artificial intelligence in\r\nneuronal substrates (Other)"^^ . . . . . . "small.jpg"^^ . . . "Harnessing function from form: towards\r\nbio-inspired artificial intelligence in\r\nneuronal substrates (Other)"^^ . . . . . . "preview.jpg"^^ . . . "Harnessing function from form: towards\r\nbio-inspired artificial intelligence in\r\nneuronal substrates (PDF)"^^ . . . "Dissertation (1).pdf"^^ . . . "Harnessing function from form: towards\r\nbio-inspired artificial intelligence in\r\nneuronal substrates (Compressed Archive)"^^ . . . . "Videos.zip"^^ . . . "Harnessing function from form: towards\r\nbio-inspired artificial intelligence in\r\nneuronal substrates (Other)"^^ . . . . . . "indexcodes.txt"^^ . . . "Harnessing function from form: towards\r\nbio-inspired artificial intelligence in\r\nneuronal substrates (Other)"^^ . . . . . . "lightbox.jpg"^^ . . "HTML Summary of #28374 \n\nHarnessing function from form: towards \nbio-inspired artificial intelligence in \nneuronal substrates\n\n" . "text/html" . . . "000 Allgemeines, Wissenschaft, Informatik"@de . "000 Generalities, Science"@en . . . "310 Statistik"@de . "310 General statistics"@en . . . "500 Naturwissenschaften und Mathematik"@de . "500 Natural sciences and mathematics"@en . . . "530 Physik"@de . "530 Physics"@en . . . "570 Biowissenschaften, Biologie"@de . "570 Life sciences"@en . . . "600 Technik, Medizin, angewandte Wissenschaften"@de . "600 Technology (Applied sciences)"@en . .