Computer Vision and Pattern Recognition · arXiv:2003.01822 v2 · submitted 3 Mar 2020
Implicitly Defined Layers in Neural Networks
The paper says
“In this paper we demonstrate that defining individual layers in a neural network implicitly provide much richer representations over the standard explicit one, consequently enabling a vastly broader class of end-to-end trainable architectures.”
from the abstract · picked out as the main claim by a classifier
Reviews
Reviews by people
No person has reviewed this paper on Pith yet. A scientist listed on Pith can review it for a fee, and their review is published here whatever it concludes.
The paper
Abstract
In conventional formulations of multilayer feedforward neural networks, the individual layers are customarily defined by explicit functions. In this paper we demonstrate that defining individual layers in a neural network implicitly provide much richer representations over the standard explicit one, consequently enabling a vastly broader class of end-to-end trainable architectures. We present a general framework of implicitly defined layers, where much of the theoretical analysis of such layers can be addressed through the implicit function theorem. We also show how implicitly defined layers can be seamlessly incorporated into existing machine learning libraries. In particular with respect to current automatic differentiation techniques for use in backpropagation based training. Finally, we demonstrate the versatility and relevance of our proposed approach on a number of diverse example problems with promising results.
The paper's own sentences, arranged. Each label is drawn from a closed vocabulary; the number beside it is the probability the classifier gave that label. No sentence in this record was written by a model.
abstract compiled · 2026-09-23 · no language-model review
How the main claim is stated: asserted 1.00
What the abstract itself states, as probabilities: claims first 0.07 · states limitation 0.06 · human participants 0.04 · controlled comparison 0.35 · randomized assignment 0.01 · states quantitative result 0.04 · compares with prior results 0.82 · mentions code or data availability 0.07
The abstract, sentence by sentence
This table labels the abstract only (arXiv listing abstract v2). The number is the probability the classifier gave the label.
| Sentence | Role | The authors' own work | States the main conclusion | Strength |
|---|---|---|---|---|
| In conventional formulations of multilayer feedforward neural networks, the individual layers are customarily defined by explicit functions. | background 1.00 | 0.05 | 0.02 | asserted 1.00 |
| In this paper we demonstrate that defining individual layers in a neural network \emph{implicitly} provide much richer representations over the standard explicit one, consequently enabling a vastly broader class of end-to-end trainable architectures. | contribution 0.56 | 0.92 | 0.89 | asserted 1.00 |
| We present a general framework of implicitly defined layers, where much of the theoretical analysis of such layers can be addressed through the implicit function theorem. | contribution 0.64 | 0.87 | 0.11 | asserted 0.71 |
| We also show how implicitly defined layers can be seamlessly incorporated into existing machine learning libraries. | contribution 0.81 | 0.90 | 0.05 | asserted 1.00 |
| In particular with respect to current automatic differentiation techniques for use in backpropagation based training. | method 0.63 | 0.37 | 0.03 | unclear 0.86 |
| Finally, we demonstrate the versatility and relevance of our proposed approach on a number of diverse example problems with promising results. | result 0.70 | 0.95 | 0.17 | asserted 0.99 |
What kind of paper the classifier reports this is: method or tool 1.00 · evidence from simulation or computation 0.38
What was examined, and what was not
Sources read
- arXiv listing abstract v2
Not examined
- full text
- figures and tables
- references and citing sentences
- relations to other papers
Levels run
- P abstract
Not run
- S sections
- T body sentences
- C citations
- R cross-paper relations
How the labels were made
Compiled by classifier-1.13.0 via TypeSafe API, question bank abstract-p1-20260923, 2026-09-23: 1 request, 4,847 input tokens.
Every quoted sentence on this record is the authors'. Every label is from Pith's closed vocabulary, with the probability shown beside it. This record as JSON
Cited by 1 Pith paper
Reviewed papers on Pith that cite this one. The full citation record
- JFB: Jacobian-Free Backpropagation for Implicit Networks cs.LG · Mar 2021
How the literature uses it
5 arXiv papers cite this in 11 sentences
Papers citing it · 4 sentences without a confident label
Builds on it 1 paper
-
Our contributions are as follows: * We introduce Implicit Riemannian Concave Potential Maps, IRCPMs which extend Riemannian Convex Potential Maps, RCPMs from [ref] by allowing for much more general c-concave functions, <Ref>, and by making it easy to incorporate symmetries of the target density, <Ref>; * We extend results from implicit layers, [ref] , allowing higher-order implicit derivatives to be efficiently computed, <Ref>; * We discuss conditions under which the infimum problem necessary to define the potential $ψ^c$ can be simply solved by gradient descent, Section [ref].
Implicit Riemannian Concave Potential Maps · Introduction · 1.00
Uses its method or tool 2 papers
-
According to the implicit function theorem stated in [ref] , the derivatives of solution $U$ to the input predictions $Ŷ$ can be calculated by [equation]
End-to-End Learning-based Operation of Integrated Energy Systems for Buildings and Data Centers · Constrained Optimization as a Neural Layer · 0.94
-
As in [ref] , $u_i$ are obtained by solving a linear system of equations [equation] Second and higher-order vector-Jacobian products can be obtained by a recursive application of <Ref>.
Implicit Riemannian Concave Potential Maps · Implicit higher-order gradients · 0.94
Cites it as background 3 papers
-
A growing paradigm shift has emerged from explicit to implicit neural networks [ref] .
JFB: Jacobian-Free Backpropagation for Implicit Networks · Introduction · 0.98
-
In [ref] implicit layers defined via minimisation or fixed-point equations have been studied.
Implicit Riemannian Concave Potential Maps · Related Work · 0.94
-
Implicit models There has been growing interest in implicit models in recent years [ref] .
Alternating Differentiation for Optimization Layers · Related work · 1.00
What it cites · 3 sentences without a confident label
Uses data or values from 1 paper
-
Our network was evaluated with the proposed formulation on 8,000 3D models from four different categories ('cars', 'chairs', 'bottles' and 'sofas') in the ShapeNet dataset [ref] .
ShapeNet: An Information-Rich 3D Model Repository · Experiments - Level Set Layers · 0.97
Cites as background 1 paper
-
Differentiation through optimization problems, e.g., $2$ operators, has seen a number of advances in recent years, among which, there are techniques that come up in bi-level optimization [ref] and sensitivity analysis [ref] .
On Differentiating Parameterized Argmin and Argmax Problems with Application to Bi-level Optimization · Related Works · 0.98
Challenges and discussion
Know something about this paper? Leave a comment.
Come back by email
A short list of the week's newly reviewed papers. Nothing is sent until you confirm from your inbox, and every email has a link that stops it. No account.
No comments yet.
Sign in to comment