Efficient modeling of complex Analog integrated circuits using neural networks

Hasani, Ramin; Haerle, Dieter; Grosu, Radu

doi:10.1109/prime.2016.7519486

Cited by 9 publications

(6 citation statements)

References 4 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For problems such as language modelling where a large amount of sequential data and substantial computational resources are available, transformers 36 and their variants are great choices of models. CfCs could bring value when: (1) data have limitations and irregularities (for example, medical data, financial time series, robotics 37 and closed-loop control, and multi-agent autonomous systems in supervised and reinforcement learning schemes 38 ), (2) the training and inference efficiency of a model is important (for example, embedded applications [39][40][41] ) and (3) when interpretability matters 42 .…”

Section: What Are the Limitations Of Cfcs?mentioning

confidence: 99%

Closed-form continuous-time neural networks

et al. 2022

View full text Add to dashboard Cite

Continuous-time neural networks are a class of machine learning systems that can tackle representation learning on spatiotemporal decision-making tasks. These models are typically represented by continuous differential equations. However, their expressive power when they are deployed on computers is bottlenecked by numerical differential equation solvers. This limitation has notably slowed down the scaling and understanding of numerous natural physical phenomena such as the dynamics of nervous systems. Ideally, we would circumvent this bottleneck by solving the given dynamical system in closed form. This is known to be intractable in general. Here, we show that it is possible to closely approximate the interaction between neurons and synapses—the building blocks of natural and artificial neural networks—constructed by liquid time-constant networks efficiently in closed form. To this end, we compute a tightly bounded approximation of the solution of an integral appearing in liquid time-constant dynamics that has had no known closed-form solution so far. This closed-form solution impacts the design of continuous-time and continuous-depth neural models. For instance, since time appears explicitly in closed form, the formulation relaxes the need for complex numerical solvers. Consequently, we obtain models that are between one and five orders of magnitude faster in training and inference compared with differential equation-based counterparts. More importantly, in contrast to ordinary differential equation-based continuous networks, closed-form networks can scale remarkably well compared with other deep learning instances. Lastly, as these models are derived from liquid networks, they show good performance in time-series modelling compared with advanced recurrent neural network models.

show abstract

Section: What Are the Limitations Of Cfcs?mentioning

confidence: 99%

Closed-form continuous-time neural networks

et al. 2022

View full text Add to dashboard Cite

show abstract

“…In [7], authors employed a NARX NN for modeling the power-up behavior of the BGR. In this paper we model the rest of the decomposed features and thus complete the behavioral modeling of the circuit.…”

Section: Compnn For Mimo System Modelingmentioning

confidence: 99%

“…A nonlinear auto-regressive neural network with exogenous input (NARX NN) appears to be a suitable framework for deriving approximations, up to a prescribed, maximum error, of the BGR. It has been previously demonstrated that a recurrent nature of the NARX NN topology consisting of only seven neurons and three three-time input-and-output delay components is able to precisely reproduce the turn-on behavior of the circuit [7].…”

Section: Narx Neural-network Architecturementioning

confidence: 99%

“…We aim to train a specific NARX network for each of the behavioral features where we use the input data as the exogenous input to the neural network and the output data as the target values to be learned. In order to gain high precision in the training process, we use the network in a feedforward topology in which the input of this topology consists of the original inputs and outputs, plus all the delayed inputs and outputs, up to their maximum input and output delays, respectively [7]. A Levenberg-Marquardt (LM) back-propagation algorithm is employed for training each network [14].…”

Section: Training Process and Network Performancementioning

confidence: 99%

“…The developed NN model has shown a reasonable time performance in transient simulations; however, since the model is coded in Verilog-A, simulation speed-up is limited. In [7], authors used a novel nonlinear autoregressive neuralnetwork with exogenous input (NARX) for modeling the power-up behavior of a BGR. They demonstrated attractive improvements in the time performance of the transient simulations of the analog circuit within the Cadence AMS simulator by using this NARX model.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Compositional neural-network modeling of complex analog circuits

Hasani

Haerle

Baumgartner

et al. 2017

2017 International Joint Conference on Neural Networks (IJCNN)

Self Cite

View full text Add to dashboard Cite

We introduce CompNN, a compositional method for the construction of a neural-network (NN) capturing the dynamic behavior of a complex analog multiple-input multiple-output (MIMO) system. CompNN first learns for each input/output pair (i, j), a small-sized nonlinear auto-regressive neural network with exogenous input (NARX) representing the transfer-function hij. The training dataset is generated by varying input i of the MIMO, only. Then, for each output j, the transfer functions hij are combined by a time-delayed neural network (TDNN) layer, fj. The training dataset for fj is generated by varying all MIMO inputs. The final output is f = (f1,. . ., fn). The NNs parameters are learned using Levenberg-Marquardt back-propagation algorithm. We apply CompNN to learn an NN abstraction of a CMOS band-gap voltage-reference circuit (BGR). First, we learn the NARX NNs corresponding to trimming, load-jump and linejump responses of the circuit. Then, we recompose the outputs by training the second layer TDNN structure. We demonstrate the performance of our learned NN in the transient simulation of the BGR by reducing the simulation-time by a factor of 17 compared to the transistor-level simulations. CompNN allows us to map particular parts of the NN to specific behavioral features of the BGR. To the best of our knowledge, CompNN is the first method to learn the NN of an analog integrated circuit (MIMO system) in a compositional fashion.

show abstract