Automated extraction of expert knowledge in analog topology selection and sizing

McConaghy,; Palmers,; Gielen, Georges; Steyaert, Michiel

doi:10.1109/iccad.2008.4681603

Cited by 9 publications

(8 citation statements)

References 9 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In [26], CAFFEINE extracted whitebox models relating performance-tradeoff objectives. The input dataset contained 1576 Pareto-optimal points in five objectives.…”

Section: Other Applicationsmentioning

confidence: 99%

“…Future CAFFEINE applications include the following. First is to extract the (nonlinear) sensitivity of the model output relative to each input variable, in a flow similar to [26]. Note that other sensitivity-extraction approaches have been proposed, but they were specific to the model template; e.g.…”

Section: Other Applicationsmentioning

confidence: 99%

“…From simulation data, a model y = f (x) is constructed, where y is typically a performance metric, x includes design, process, or environmental variables, and f is an approximation of the SPICE mapping. Models used include linear models [8], [9], [25], posynomials [10]- [12], polynomials [13], [14], [25], splines [15], [25], neural networks [16], [17], [25], boosted neural networks [18], [25], support vector machines [19]- [21], [25], latent variable regression (LVR) [22], [23], kriging [24], [25], and stochastic gradient boosting [26]. However, such models either follow an overly restrictive functional template which limits their applicability, or they are opaque and thus provide no insight to the designer.…”

mentioning

confidence: 99%

“…However, such models either follow an overly restrictive functional template which limits their applicability, or they are opaque and thus provide no insight to the designer. Less opaque flows exist, such as visualizing CART trees [26], nonlinear sensitivity analysis [26], or plotting the mapping from an LVR's first affine transform w 1 * x to the output y = f 1 (w 1 * x) [22], [23]. While useful, these approaches do not give the functional relations that symbolic models provide.…”

mentioning

confidence: 99%

See 3 more Smart Citations

Template-Free Symbolic Performance Modeling of Analog Circuits via Canonical-Form Functions and Genetic Programming

McConaghy

Gielen

2009

IEEE Trans. Comput.-Aided Des. Integr. Circuits Syst.

View full text Add to dashboard Cite

Abstract-This paper presents CAFFEINE, a method to automatically generate compact, interpretable symbolic performance models of analog circuits with no prior specification of an equation template. CAFFEINE uses SPICE simulation data, to model arbitrary nonlinear circuits and circuit characteristics. CAFFEINE expressions are canonical form functions: productof-sum layers alternating with sum-of-product layers, as defined by a grammar. Multi-objective genetic programming trades off error with model complexity. On test problems, CAFFEINE models demonstrate lower prediction error than posynomials, splines, neural networks, kriging, and support vector machines. This paper also demonstrates techniques to scale CAFFEINE to larger problems.

show abstract

“…In [26], CAFFEINE extracted whitebox models relating performance-tradeoff objectives. The input dataset contained 1576 Pareto-optimal points in five objectives.…”

Section: Other Applicationsmentioning

confidence: 99%

Section: Other Applicationsmentioning

confidence: 99%

mentioning

confidence: 99%

mentioning

confidence: 99%

See 2 more Smart Citations

Template-Free Symbolic Performance Modeling of Analog Circuits via Canonical-Form Functions and Genetic Programming

McConaghy

Gielen

2009

IEEE Trans. Comput.-Aided Des. Integr. Circuits Syst.

View full text Add to dashboard Cite

show abstract

“…Instead, we validate a circuit by meaningfully introducing the performance constraints over several stages to quickly converge to the final circuit. Recent approaches like [16] [17] are more about blind search over a set of topologies and verifying their performances through spice simulation.…”

Section: Introductionmentioning

confidence: 99%

A geometric programming aided knowledge based approach for analog circuit synthesis and sizing

Maji

Mandal

2011

Proceedings of the 21st Edition of the Great Lakes Symposium on Great Lakes Symposium on VLSI

View full text Add to dashboard Cite

A knowledge based approach empowered by geometric programming (GP) for analog circuit synthesis and sizing is presented. Analog circuit performance specification is mapped to various building blocks of a circuit topology. Then, the topology is modified according to the design rules in the library. Each modification is validated over two steps. In the first step, dc performance constraints are introduced. If qualified, ac performance constraints are introduced. Validation over two steps helps to gradually close in on input specifications removing any undesired correction made initially. It results in faster convergence.

show abstract

Automatic generation of macromodels and design equations for application to Op Amp design

Shi

2023

Circuit Theory & Apps

View full text Add to dashboard Cite

SummaryOperational amplifier (Op Amp) is a special class of analog integrated circuits. Multiple‐stage Op Amp design is of particular interest recently. However, analysis of this class of circuits has to be done manually and requires advanced skills. Automatic analysis of this class of circuits is studied in this paper, which takes a novel approach by circuit recognition and symbolic generation. Circuit recognition is deterministic, implemented by deterministic circuit partitioning and functional block extraction methods. Symbolic analysis of extracted circuit blocks can subsequently be performed with greatly reduced complexity while generated models and equations bear more useful information for circuit design. Symbolic results so generated can be applied in design space exploration, including tasks like finding the limitation in circuit topology, conducting initial sizing, and determining performance tradeoffs. This paper proposes the key algorithms for realizing the recognition of a few well‐known functional blocks frequently used in CMOS Op Amp circuits and further explores the possibility of using the recognized circuit cells in combination with the gm/ID method for circuit sizing. Preliminary tests show that this approach is a potential candidate for locating the performance boundaries of multiple‐stage Op Amps.

show abstract

Automated extraction of expert knowledge in analog topology selection and sizing

Cited by 9 publications

References 9 publications

Template-Free Symbolic Performance Modeling of Analog Circuits via Canonical-Form Functions and Genetic Programming

Template-Free Symbolic Performance Modeling of Analog Circuits via Canonical-Form Functions and Genetic Programming

A geometric programming aided knowledge based approach for analog circuit synthesis and sizing

Automatic generation of macromodels and design equations for application to Op Amp design

Contact Info

Product

Resources

About