AxNN: Energy-efficient neuromorphic systems using approximate computing

By Swagath Venkataramani, Ashish Ranjan, Kaushik Roy, Anand Raghunathan. 2014

258 citations recorded by the source corpus.

Subfields: Electrical and Electronic Engineering

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Electrical and Electronic Engineering

1. Sparsh Mittal
2. Ala Al-Fuqaha
3. M. Mohammadi
4. Kaushik Roy
5. Bashir I. Morshed
6. Syed Kamrul Islam
7. Jianlin Cheng
8. Md Maruf Hossain Shuvo

Citations in the corpus

• Deep Learning for IoT Big Data and Streaming Analytics: A Survey (2017, methodology): “Approximate computing is another approach to both implement machine learning tools on IoT devices and contribute to the energy saving of their hosting devices [145], • Good for time-critical IoT apps • Energy-efficient • Provides more security and privacy for data • Special-purpose networks [146].”
• A Survey of Techniques for Approximate Computing (2016, methodology): “Further, once relaxable portions are identified, conveying this to the software or compiler requires source-code annotations; several programming frameworks provide support for this (Section 3.3).”
• A Low Effort Approach to Structured CNN Design Using PCA (2018, methodology): “The first are techniques that prune individual weights, such as in [5], [6], [14] and [13].”
• ApproxDBN: Approximate Computing for Discriminative Deep Belief Networks (2017, methodology): “To generalize the criticality analysis of feedforward deterministic neural network [7][12] to generative stochastic model DDBNs, there are two key steps: (i) choose a relevant loss function expressible as a function of the values of hidden neurons, and (ii) create a surrogate stochastic derivative s”
• Energy-efficient ConvNets through approximate computing (2016, background): “A first attempt in combining neural networks with approximate computing has been done in [19].”
• Efficient Acceleration of Deep Learning Inference on Resource-Constrained Edge Devices: A Review (2023, methodology): “An approximate NN (AxNN) can be obtained using a quality-configurable technique exploiting neuromorphic processing [174].”
• LightNN: Filling the Gap between Conventional Deep Neural Networks and Binarized Networks (2017, background): “propose an energy-efficient neural network design approach by reducing bit precision of resilient neurons [12], while Zhang et al.”
• Efficient embedded learning for IoT devices (2016, background): “AxNN leverages an operation from the training of deep learning networks viz. backpropagation to characterize the resilience of the network’s neurons.”
• A Survey of Neuromorphic Computing and Neural Networks in Hardware (2017, background): “data set [316], [2580], [2585], [2590], [2591], the CIFAR10”