Adversarial training improves gradient interpretability, reducing misinterpretations.
problem Improving gradient interpretability in adversarially trained deep networks.
method Identified and demonstrated that adversarial training restricts gradients closer to the image manifold, making them more interpretable.
result Adversarial training leads to more meaningful loss gradients, aligning better with human perception.
Empirical study shows interpretable gradients improve adversarial robustness.
problem Connecting adversarial robustness and model interpretability.
method Introduced Interpretation Regularization (IR) to encourage interpretable gradients.
result Training models to have interpretable gradients improves adversarial robustness.
New method creates universal perturbations to fool neural network interpretations.
problem Vulnerability of gradient-based saliency maps to adversarial perturbations.
method Gradient-based optimization and PCA-based approach to create UPI.
result Existence and successful application of Universal Perturbation for Interpretation (UPI).
Gradient Weighted Superpixels improve CNN interpretability without sacrificing speed.
problem Efficiency vs. interpretability trade-off in CNNs, especially for large input volumes.
method Gradient-based pixel scoring techniques applied to superpixels.
result Superpixels approximate LIME in a fraction of the time, improving interpretability.
Optimized decision trees for reinforcement learning, improving sample complexity and interpretability.
problem Updating decision trees online in reinforcement learning.
method Gradient update over differentiable decision trees, including theoretical justification and empirical validation.
result Our approach outperforms neural networks in sample complexity and achieves higher rewards online.
New research shows input-gradients can be manipulated without changing model's core function, challenging their use for model interpretation.
problem Current methods for model interpretability using input-gradients are flawed due to their arbitrary manipulability.
method Investigated by reinterpreting logits as unnormalized log-densities, proposing novel approximations for score-matching.
result Improving alignment between implicit density model and data distribution enhances gradient structure and explanatory power.
Simplified SGD interpretation as Ito process for broader applicability.
problem Lack of generality in current SGD interpretation.
method Introduced a simplified scheme for discrete-time approximation of Ito process.
result Flexibly interprets SGD and SGLD, providing insights into their asymptotic properties.
The Volterra lattice is considered. New gradient interpretation for this dynamical system is proposed. This interpretation seems to be more natural than existing ones.
New strategy uses interpretability to improve adversarial learning.
problem Improving adversarial learning speed and effectiveness.
method Spatially constrained one-pixel adversarial perturbations guided by gradient-based interpretability.
result Spatially constrained one-pixel adversarial perturbations noticeably improve convergence and attack success.
Researchers interpret SGD using diffusion metrics for clearer geometric understanding.
problem Elusiveness of geometrical significance in stochastic gradient descent.
method Study a deterministic model with geodesics of diffusion metrics.
result Establishes parallel with General Relativity models.
A new Shapley value approach for neural networks interpretable and stable.
problem Neural networks' interpretability and training stability issues.
method Shapley value approximation for ReLU activation, globally continuous Shapley gradient, Shapley Activation function.
result SA consistently outperforms ReLU in training convergence, accuracy, and stability.
ParamBoost uses gradient boosting to create interpretable non-linear models with constraints.
problem Creating interpretable non-linear models with expert knowledge constraints.
method Gradient Boosting of cubic polynomials with specified constraints.
result ParamBoost outperforms state-of-the-art GAMs in real-world datasets.
A method interprets black-box models using an ensemble of gradient boosting machines.
problem Local and global interpretation of black-box models.
method An ensemble of gradient boosting machines (GBMs) to form a generalized additive model.
result Efficiency and properties demonstrated on synthetic and real datasets.
Proposes a model to interpret complex ML algorithms.
problem Complex ML models are hard to interpret.
method Uses model-based regression trees and interpretable main-effects models.
result Surrogate model provides interpretable results with good predictive performance.
MoreauGrad interprets neural nets robustly and sparsely.
problem Lack of robustness and sparsity in gradient-based interpretation methods.
method Moreau envelope for smooth and robust interpretation, combined with L1 regularization for sparsity.
result MoreauGrad provides a smooth, robust, and sparse explanation of neural nets.
The paper interprets policy-gradient algorithms using continuation theory.
problem Optimizing nonconvex functions in reinforcement learning.
method Formulates policy optimization as optimization by continuation, interprets policy-gradient algorithms as implicitly optimizing deterministic policies.
result Exploration in policy-gradient algorithms is seen as computing a continuation of the return of the policy.
A new gradient boosting method improves interpretability of probabilistic models.
problem Learning interpretable yet accurate probabilistic models with limited rule complexity.
method A new objective function that measures the angle between risk gradient and condition output vector projection.
result Significantly improves comprehensibility/accuracy trade-off of fitted ensemble.
In this paper, we provide an overview of first-order and second-order variants of the gradient descent method that are commonly used in machine learning. We propose a general framework in which 6 of these variants can be interpreted as different instances of the same approach. They are the vanilla gradient descent, the…
The paper interprets learned step sizes in deep-unfolded gradient descent.
problem Intuitive interpretation of learned non-constant step sizes in deep-unfolded gradient descent.
method Theoretical analysis and optimization of spectral radius.
result Chebyshev steps achieve the lower bound of convergence rate for first-order methods.
This manuscript proposes a probabilistic framework for algorithms that iteratively solve unconstrained linear problems Bx=b with positive definite B for x. The goal is to replace the point estimates returned by existing methods with a Gaussian posterior belief over the elements of the inverse of B, which can …
Study tests if input gradients highlight discriminative features, finds they often fail.
problem Validity of assumption that input gradients highlight discriminative features in model predictions.
method Developed DiffROAR framework and BlockMNIST dataset to test assumption on four benchmarks.
result Input gradients of standard models often fail to highlight discriminative features, while robust models do.
New methods accelerate gradient descent for convex and strongly convex functions.
problem Improving convergence rates of gradient-based optimization methods.
method Formulated two classes of first-order algorithms with Lyapunov analyses and Hamiltonian assisted gradient method.
result Achieved accelerated convergence rates matching Nesterov's methods in strongly and general convex settings.
SGD and stochastic gradient descent converge at optimal rates for certain non-convex functions.
problem Optimal convergence rates for non-convex functions under gradient noise.
method Geometric interpretation of the PL-condition to analyze convergence rates.
result Convergence rates of SGD and stochastic gradient descent match those of strongly convex quadratics.
The paper interprets diffusion models as gradient descent and proposes a new sampler.
problem Improving the efficiency and quality of diffusion models.
method Interprets diffusion models as gradient descent and proposes a new sampler.
result The new sampler achieves state-of-the-art FID scores and generates high quality samples.
Paper uses ML to predict SME defaults with interpretability.
problem Lack of interpretability in ML models for SME default prediction.
method Model-agnostic approach using Accumulated Local Effects and Shapley values.
result eXtreme Gradient Boosting algorithm provides highest classification power with interpretability.
Gradient descent optimization improved by circuit perspective.
problem Efficient training of large AI systems.
method Utilizing circuit perspective for mechanistic interpretability.
result Designing a curriculum for efficient learning.
Unified method for local GBDT feature contributions.
problem Need for local model interpretation for GBDT.
method Unified computation mechanism for instance-level feature contributions.
result Unified computation mechanism for GBDT feature contributions.
New tool for interpreting neural nets, explaining model behavior.
problem Interpreting neural network responses and explaining model behavior.
method Full-Gradient representation and FullGrad saliency map approximation.
result FullGrad method explains model behavior more comprehensively and accurately than other methods.
CBO interprets as SGD, leading to global convergence for nonconvex functions.
problem Understanding and improving gradient-based learning algorithms.
method Interpreting CBO as a stochastic relaxation of SGD.
result CBO provably converges globally to minimizers for nonsmooth nonconvex functions.
LIMIS improves locally interpretable models by selecting and distilling key instances.
problem Low fidelity of locally interpretable models.
method LIMIS uses instance-wise subsampling guided by policy gradient and reward to improve fidelity.
result LIMIS near-matches black-box model accuracy while significantly improving fidelity.
Sparse activations in neural models correlate with frequent words, suggesting sparsity is natural.
problem Interpretability and resource efficiency in neural language models.
method Used the Taxi-Euclidean norm to measure sparsity and analyzed gradients and activations of frequent words.
result Frequent input words are associated with sparse activations, while frequent target words are associated with dispersed activations.
Temporal difference learning explained through gradient splitting, improving convergence times.
problem Learning value functions in Markov Decision Processes with linear approximations.
method Interpreting TD learning as gradient splitting and applying convergence proofs from gradient descent.
result Improved convergence times for TD learning, especially with a minor variation.
ICCNLS models complex relationships as convex and concave components.
problem Complex input-output relationships with affine ambiguity.
method Sub-gradient constrained affine functions, global orthogonality constraints, L1, L2, and elastic net regularisation.
result Improved predictive accuracy and model simplicity compared to conventional methods.
Paper explores how feature interactions improve XGBoost models.
problem Improving XGBoost model performance and interpretability.
method Identifies feature interactions as constraints in XGBoost models.
result Accurate identification of feature interactions leads to significant model performance improvements.
Paper proposes HTAF for stable training of binary neural networks.
problem Challenges in training binary neural networks with gradient-based optimization.
method HTAF is a smooth approximation to the Heaviside function that enables stable training.
result HTAF enables stable training of various binary neural networks with gradient-based optimization.
Greedy PIG adapts integrated gradients for better feature attribution.
problem Interpreting deep learning model predictions.
method Unified discrete optimization framework for feature attribution and selection.
result Greedy PIG improves feature attribution on various tasks.
A new method relaxes Boolean Matrix Factorization to make it more efficient.
problem High computational cost of solving NP-hard combinatorial optimization problems in Boolean Matrix Factorization.
method Proposes a proximal gradient algorithm using an elastic-binary regularizer to relax BMF.
result Demonstrates improved runtime and better recall, loss, and interpretability on real-world data.
This paper addresses anisotropy in Transformer models, providing geometric insights and empirical support.
problem Anisotropy phenomenon in Transformer models, challenging their geometric interpretation.
method Derive geometric arguments and use concept-based mechanistic interpretability during training.
result Activation-derived directions capture large gradient energy and a larger share of gradient anisotropy than normal controls.
Transformers learn random walks optimally with gradient descent.
problem Transformer interpretability and learning random walks.
method Theoretical analysis and gradient descent training.
result Transformers can predict random walks optimally with gradient descent.
RCAV quantifies model sensitivity to semantic concepts, improving interpretability methods.
problem Lack of semantic interpretability in image classification models.
method RCAV calculates concept gradients and ascent steps to assess model sensitivity to semantic concepts.
result RCAV yields more accurate and robust interpretations of model behavior.
Introspects convolutional speech recognition models using Gradient-adjusted Neuron Activation Profiles.
problem Lack of interpretability in deep learning ASR models.
method Gradient-adjusted Neuron Activation Profiles (GradNAPs) for feature and representation visualization.
result Gains insight into how data is processed in convolutional ASR models.
Improved deep learning models using new attribution priors and expected gradients.
problem Improving interpretability and performance of deep learning models.
method Introducing new attribution priors and expected gradients method that satisfies interpretability axioms.
result Improves model performance across various real-world tasks.
FP uses random projections to train networks without feedback, achieving comparable performance to backpropagation.
problem Training neural networks without feedback from downstream layers.
method Forward Projection (FP) method that uses randomised nonlinear projections and closed-form regression.
result FP achieves comparable generalisation to backpropagation methods with a single forward pass, offering significant speedup.
Derives equations for deep learning biases and weights, showing data complexity reduction.
problem Understanding interpretability in supervised learning.
method Gradient flow equations and dynamical truncation of training data.
result Data complexity reduction at an exponential rate with training.
MALC combines interpretable linear models with black-box models for better predictions and transparency.
problem Combining interpretability with black-box models for better predictions.
method Formulates MALC as a convex optimization problem and uses accelerated proximal gradient method for training.
result MALC provides an efficient frontier balancing prediction accuracy and transparency.
Deep networks are mathematically equivalent to kernel machines learned by gradient descent.
problem Understanding the learned representations of deep learning models.
method Using gradient descent to learn deep networks, showing they are equivalent to kernel machines.
result Deep network weights are a superposition of training examples, revealing the learned function.
Dynamic CBDT improves treatment effect estimation in clinical data.
problem Estimating heterogeneous treatment effects in observational data with high accuracy and interpretability.
method Dynamic Regularized Causal Boosted Decision Trees (CBDT) integrating variance regularization and calibration.
result Significantly improved estimation accuracy and reliable coverage of true treatment effects.
We present a unifying framework for adapting the update direction in gradient-based iterative optimization methods. As natural special cases we re-derive classical momentum and Nesterov's accelerated gradient method, lending a new intuitive interpretation to the latter algorithm. We show that a new algorithm, which we …