Hoeffding decomposition of black-box models with dependent inputs
- URL: http://arxiv.org/abs/2310.06567v2
- Date: Thu, 7 Mar 2024 09:14:24 GMT
- Title: Hoeffding decomposition of black-box models with dependent inputs
- Authors: Marouane Il Idrissi (EDF R\&D PRISME, IMT, SINCLAIR AI Lab), Nicolas
Bousquet (EDF R\&D PRISME, SINCLAIR AI Lab, LPSM (UMR\_8001)), Fabrice Gamboa
(IMT), Bertrand Iooss (EDF R\&D PRISME, IMT, SINCLAIR AI Lab, RT-UQ),
Jean-Michel Loubes (IMT)
- Abstract summary: One of the main challenges for interpreting black-box models is the ability to decompose square-integrable functions of non-independent random inputs into a sum of functions of every possible subset of variables.
We show that under two reasonable assumptions, it is always possible to decompose such a function uniquely.
The elements of this decomposition can be expressed using projections and allow for novel interpretability indices.
- Score: 22.519758624657644
- License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
- Abstract: One of the main challenges for interpreting black-box models is the ability
to uniquely decompose square-integrable functions of non-independent random
inputs into a sum of functions of every possible subset of variables. However,
dealing with dependencies among inputs can be complicated. We propose a novel
framework to study this problem, linking three domains of mathematics:
probability theory, functional analysis, and combinatorics. We show that, under
two reasonable assumptions on the inputs (non-perfect functional dependence and
non-degenerate stochastic dependence), it is always possible to decompose such
a function uniquely. This generalizes the well-known Hoeffding decomposition.
The elements of this decomposition can be expressed using oblique projections
and allow for novel interpretability indices for evaluation and variance
decomposition purposes. The properties of these novel indices are studied and
discussed. This generalization offers a path towards a more precise uncertainty
quantification, which can benefit sensitivity analysis and interpretability
studies whenever the inputs are dependent. This decomposition is illustrated
analytically, and the challenges for adopting these results in practice are
discussed.
Related papers
- Understanding Diffusion Models by Feynman's Path Integral [2.4373900721120285]
We introduce a novel formulation of diffusion models using Feynman's integral path.
We find this formulation providing comprehensive descriptions of score-based generative models.
We also demonstrate the derivation of backward differential equations and loss functions.
arXiv Detail & Related papers (2024-03-17T16:24:29Z) - Nonparametric Partial Disentanglement via Mechanism Sparsity: Sparse
Actions, Interventions and Sparse Temporal Dependencies [58.179981892921056]
This work introduces a novel principle for disentanglement we call mechanism sparsity regularization.
We propose a representation learning method that induces disentanglement by simultaneously learning the latent factors.
We show that the latent factors can be recovered by regularizing the learned causal graph to be sparse.
arXiv Detail & Related papers (2024-01-10T02:38:21Z) - Exactly Solvable Schr\"odinger equations with Singularities: A
Systematic Approach to Solving Complexified Potentials (part1) [0.0]
Extending the argument of the potential to a complex number leads to solving exactly the Schr"odinger equation.
This paper gives a new perspective on how to solve the second-order linear differential equation written in normal form.
arXiv Detail & Related papers (2023-01-10T05:39:43Z) - Mutual Exclusivity Training and Primitive Augmentation to Induce
Compositionality [84.94877848357896]
Recent datasets expose the lack of the systematic generalization ability in standard sequence-to-sequence models.
We analyze this behavior of seq2seq models and identify two contributing factors: a lack of mutual exclusivity bias and the tendency to memorize whole examples.
We show substantial empirical improvements using standard sequence-to-sequence models on two widely-used compositionality datasets.
arXiv Detail & Related papers (2022-11-28T17:36:41Z) - Weight-variant Latent Causal Models [79.79711624326299]
Causal representation learning exposes latent high-level causal variables behind low-level observations.
In this work we focus on identifying latent causal variables.
We show that the transitivity severely hinders the identifiability of latent causal variables.
We propose a novel method, termed Structural caUsAl Variational autoEncoder, which directly learns latent causal variables.
arXiv Detail & Related papers (2022-08-30T11:12:59Z) - Amortized Inference for Causal Structure Learning [72.84105256353801]
Learning causal structure poses a search problem that typically involves evaluating structures using a score or independence test.
We train a variational inference model to predict the causal structure from observational/interventional data.
Our models exhibit robust generalization capabilities under substantial distribution shift.
arXiv Detail & Related papers (2022-05-25T17:37:08Z) - Generalization Bounds via Convex Analysis [12.411844611718958]
We show that it is possible to replace the mutual information by any strongly convex function of the joint input-output distribution.
Examples include bounds stated in terms of $p$-norm divergences and the Wasserstein-2 distance.
arXiv Detail & Related papers (2022-02-10T12:30:45Z) - Partial Counterfactual Identification from Observational and
Experimental Data [83.798237968683]
We develop effective Monte Carlo algorithms to approximate the optimal bounds from an arbitrary combination of observational and experimental data.
Our algorithms are validated extensively on synthetic and real-world datasets.
arXiv Detail & Related papers (2021-10-12T02:21:30Z) - Disentangling Observed Causal Effects from Latent Confounders using
Method of Moments [67.27068846108047]
We provide guarantees on identifiability and learnability under mild assumptions.
We develop efficient algorithms based on coupled tensor decomposition with linear constraints to obtain scalable and guaranteed solutions.
arXiv Detail & Related papers (2021-01-17T07:48:45Z) - Understanding Double Descent Requires a Fine-Grained Bias-Variance
Decomposition [34.235007566913396]
We describe an interpretable, symmetric decomposition of the variance into terms associated with the labels.
We find that the bias decreases monotonically with the network width, but the variance terms exhibit non-monotonic behavior.
We also analyze the strikingly rich phenomenology that arises.
arXiv Detail & Related papers (2020-11-04T21:04:02Z) - Flexible mean field variational inference using mixtures of
non-overlapping exponential families [6.599344783327053]
I show that using standard mean field variational inference can fail to produce sensible results for models with sparsity-inducing priors.
I show that any mixture of a diffuse exponential family and a point mass at zero to model sparsity forms an exponential family.
arXiv Detail & Related papers (2020-10-14T01:46:56Z)
This list is automatically generated from the titles and abstracts of the papers in this site.
This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.