Patent
A stochastic implementation method of an activation function for an artificial neural network and a system including the same
Yeo In June,Lee Byung Geun,Gi Sang Gyun +2 more
- 05 Nov 2018
2
TL;DR: In this paper, an activation function for an artificial neural network can be implemented without an analog-to-digital converter by adjusting comparator inputs constituting the neural network, which can implement four different types of active functions.
read more
Abstract: The present invention relates to implementing an activation function for an artificial neural network. Specifically, the present invention can implement four different types of active functions by adjusting comparator inputs constituting the artificial neural network. It is possible to implement the activation function without an analog-to-digital converter.
read more
Chat with Paper
AI Agents for this Paper
Find similar papers on Google Scholar, PubMed and Arxiv
Write a critical review of this paper
Analyze citations of this paper to find unaddressed research gaps
Citations
Patent
A configurable neural network activation function implementation device
Che Deliang,Li Na +1 more
- 28 May 2019
TL;DR: In this article, a configurable neural network activation function implementation device is presented, which consists of a controller, a symbol judgment module, a range detection module, an address generator, a parameter register, a floating point multiplier and a phase operand arithmetic unit.
2
Patent
Method to Improve Accuracy in Stochastic Computing for Deep Neural Networks
Lee Jongeun,Zhakatayev Aidyn +1 more
- 04 Nov 2020
TL;DR: In this paper, the authors present a device for increasing the accuracy of a stochastic computing circuit that simplifies a structure by implementing an SC multiplier that uses an SM-SC probability number (SN) and using AND gates for data bits.
References
Efficient hardware implementation of the hyperbolic tangent sigmoid function
Ashkan Hosseinzadeh Namin,Karl Leboeuf,Roberto Muscedere,Huapeng Wu,Majid Ahmadi +4 more
- 24 May 2009
TL;DR: This paper presents a simple and efficient architecture for digital hardware implementation of the hyperbolic tangent sigmoid function, which proves to be more efficient considering area × delay as a performance metric when compared to similar proposals.
140
A digital neural network architecture for VLSI
Jr. Tomlinson M.S.,D.J. Walker,M.A. Sivilotti +2 more
- 17 Jun 1990
TL;DR: The key innovation is the representation of neuron activations and synaptic weights as stochastic functions of time, leading to efficient implementations of the synapses, and a flexible architecture that permits the realization of arbitrary network topologies and dimensions.
100
A digital hardware pulse-mode neuron with piecewise linear activation function
TL;DR: Experimental results show that the piecewise-linear function of the proposed neuron is programmable and robust against the change in the number of input signals, and the convergence rate of the learning and generalization capability are improved.
79
Architecture and statistical model of a pulse-mode digital multilayer neural network
Young-Chul Kim,M.A. Shanblatt +1 more
TL;DR: A new architecture and a statistical model for a pulse-mode digital multilayer neural network (DMNN) are presented, which leads to a massively parallel yet compact and flexible network architecture, well suited for VLSI implementation.
51
A floating-gate programmable array of silicon neurons for central pattern generating networks
Francesco Tenore,R.J. Vogelstein,Ralph Etienne-Cummings,Gert Cauwenberghs,P. Hasler +4 more
- 21 May 2006
TL;DR: A new central pattern generator chip with 24 silicon neurons and reprogrammable connectivity is presented, enabling construction of a fully-interconnected network with sensory feedback from off-chip elements.