New ODE solvers improve training efficiency and accuracy.
problem Training Neural ODEs requires efficient and accurate gradient calculation.
method Presented algebraically reversible ODE solvers that are time and memory efficient, calculate exact gradients, and are numerically stable.
result Reversible solvers strictly improve upon previous architectures in efficiency and accuracy.
The paper proves a reverse isoperimetric inequality and applies it to analyze surface flows.
problem Analyzing the negative gradient flow of the Willmore energy plus volume.
method Proved a quantitative reverse isoperimetric inequality and applied it to the flow.
result Initial surfaces converge to a round point in finite or infinite time.
Adaptive algorithm improves convergence rate of Langevin dynamics.
problem Improving convergence rate of Langevin dynamics.
method Adaptive non-reversible stochastic gradient Langevin dynamics algorithm.
result Improved convergence rate of the algorithm.
We study two procedures (reverse-mode and forward-mode) for computing the gradient of the validation error with respect to the hyperparameters of any iterative learning algorithm such as stochastic gradient descent. These procedures mirror two methods of computing gradients for recurrent neural networks and have differ…
NSGLD improves SGLD for non-convex optimization problems.
problem Optimizing non-convex objectives efficiently.
method Introducing non-reversible SGLD by adding an anti-symmetric matrix to the drift term of the Langevin diffusion.
result NSGLD converges faster to the same stationary distribution with non-asymptotic guarantees.
The paper analyzes variance reduction in stochastic gradient Langevin dynamics.
problem Reducing the variance of stochastic gradient estimators in Langevin dynamics.
method Central limit theorem and Poisson equation analysis for variance characterization.
result Anti-symmetric perturbations can reduce the variance of non-reversible Langevin dynamics.
Tuning hyperparameters of learning algorithms is hard because gradients are usually unavailable. We compute exact gradients of cross-validation performance with respect to all hyperparameters by chaining derivatives backwards through the entire training procedure. These gradients allow us to optimize thousands of hyper…
PETRA enables parallel training of deep models with reversible architectures.
problem Challenges in parallelizing deep model training.
method Introduces PETRA, a novel approach for parallelizing gradient computations in reversible architectures.
result Achieves competitive accuracies on CIFAR-10, ImageNet32, and ImageNet using ResNet models.
Stochastic gradient methods are the workhorse (algorithms) of large-scale optimization problems in machine learning, signal processing, and other computational sciences and engineering. This paper studies Markov chain gradient descent, a variant of stochastic gradient descent where the random samples are taken on the t…
RevDEQs improve performance on tasks with exact gradients and fewer function evaluations.
problem Inexact gradient calculation in DEQs leads to unstable training and requires regularisation or many function evaluations.
method Introduce Reversible Deep Equilibrium Models (RevDEQs) that allow for exact gradient calculation, no regularisation, and far fewer function evaluations.
result RevDEQs significantly improve performance on language modelling and image classification tasks.
Conformer encoder reverses sequence in time dimension, affecting decoder training.
problem Reversal of sequence in Conformer encoder impacts decoder training.
method Analyzed initial behavior of decoder cross-attention and proposed methods to avoid flipping.
result Self-attention module of Conformer starts dominating, allowing only reversed information to pass.
Improved KL divergence estimators for normalizing flows lead to faster convergence and better approximations.
problem Estimating KL divergences for normalizing flows efficiently and accurately.
method Path-gradient estimators for reverse and forward KL divergences.
result Path-gradient estimators lead to faster convergence and better approximation results.
A distributed system identification method for LTI systems using reverse experience replay.
problem Online system identification of LTI systems over multi-agent networks.
method DSGD-RER, a distributed variant of SGD-RER with backward updates.
result The estimation error decreases as the network size grows.
A new interpolation method speeds up neural ODE training.
problem Efficiently approximating gradients in neural ODEs.
method Interpolation-based technique to approximate gradients.
result Our method trains neural ODEs faster than the reverse dynamic method.
In reinforcement learning, temporal difference (TD) is the most direct algorithm to learn the value function of a policy. For large or infinite state spaces, exact representations of the value function are usually not available, and it must be approximated by a function in some parametric family. However, with \emph{no…
In his 2011 work, Maas has shown that the law of any time-reversible continuous-time Markov chain with finite state space evolves like a gradient flow of the relative entropy with respect to its stationary distribution. In this work we show the converse to the above by showing that if the relative law of a Markov chain…
The paper generalizes optimization algorithms using category theory.
problem Optimizing functions in a category-theoretic setting.
method Using the Cartesian reverse derivative to generalize gradient descent and Newton's method.
result Properties of optimization algorithms are preserved in the generalized setting, including invariances and convergence.
AEGR method improves anomaly detection in autoencoders without needing anomaly-free training data.
problem Challenges in anomaly detection, especially high dimensionality and noise in training sets.
method Gradient-reversal method for autoencoders, using reconstruction error and Local Outlier Factor.
result The proposed AEGR model outperforms other methods in detecting network anomalies.
The book covers scalable MCMC methods for Bayesian learning.
problem Scalability issues in Bayesian learning with large datasets.
method Advanced MCMC algorithms, including stochastic gradient, non-reversible, and continuous time methods.
result Substantial advances in practical and theoretical Bayesian computation.
Deep learning has seen tremendous success over the past decade in computer vision, machine translation, and gameplay. This success rests in crucial ways on gradient-descent optimization and the ability to learn parameters of a neural network by backpropagating observed errors. However, neural network architectures are …
We study some equivalent properties of the curvature-dimension conditions CD(n,K) inequality on infinite, but locally finite graph. These equivalences are gradient estimate, Poincaré type inequalities and reverse Poincaré inequalities. And we also obtain one equivalent property of gradient estimate for a new notion o…
We developed efficient methods to compute gradients for Neural SDEs, improving training speed and accuracy.
problem Training Neural SDEs requires accurate and efficient computation of gradients, which is challenging due to the complexity of SDEs.
method We introduced a reversible Heun method for solving backwards-in-time SDEs and a Brownian Interval for sampling and reconstructing Brownian motion.
result Our methods significantly improve training speed and accuracy for Neural SDEs, outperforming state-of-the-art techniques.
Adversarial domain adaptation reduces sample bias in high energy physics classifier.
problem Sample bias in high energy physics classifier training.
method Adversarial domain adaptation using neural networks with gradient reversal layer.
result Successful bias removal on simulated events at the LHC.
New method uses PDMPs with sub-sampling for efficient sampling from posterior distributions.
problem Efficient sampling from posterior distributions with limited data access.
method Approximate simulation of PDMPs with sub-sampling and stochastic gradient estimation.
result Stochastic-gradient PDMPs are efficient and robust compared to Langevin dynamics.
We consider the following problem: given two parallel and identically oriented bundles of light rays in n-dimensional Euclidean space and given a diffeomorphism between the rays of the former bundle and the rays of the latter one, is it possible to realize this diffeomorphism by means of several mirror reflections? We …
A new method computes gradients without backpropagation.
problem Optimization of machine learning models.
method Forward mode automatic differentiation to compute gradients.
result Forward gradient is an unbiased estimate of the gradient, eliminating the need for backpropagation.
CD learning is shown to be an adversarial game for fitting models.
problem Difficulty in understanding the convergence properties of CD learning.
method Presented an alternative derivation of CD without approximation, showing it as a time-reversal adversarial game.
result CD is an adversarial learning procedure where a discriminator tries to classify time-reversed Markov chains.
Unified framework for sampling from complex distributions, including discrete and mixed-variable systems.
problem Sampling from complex unnormalized distributions, especially in discrete or mixed-variable systems.
method Enforces time-reversibility using a prescribed physical transition kernel to minimize Maximum Mean Discrepancy (MMD).
result Demonstrates accurate reproduction of thermodynamic observables and mode-switching behavior across diverse systems.
Improved sampling via learned diffusions using variational losses.
problem Sampling from target distributions without direct access to samples.
method Generalized Schrödinger bridge problem, variational formulation, gradient-based optimization.
result Proposed log-variance loss leads to improved performance.
ACA method improves gradient estimation for neural ODEs, reducing error and training time.
problem Inaccurate gradient estimation methods hinder the performance of neural ODEs on benchmark tasks.
method Adaptive Checkpoint Adjoint (ACA) method that applies trajectory checkpointing, deletes redundant components, and supports adaptive solvers.
result ACA reduces error rate by half and training time by half compared to adjoint and naive methods on image classification tasks.
Let K be an irreducible and reversible Markov kernel on a finite set X. We construct a metric W on the set of probability measures on X and show that with respect to this metric, the law of the continuous time Markov chain evolves as the gradient flow of the entropy. This result is a discrete counterpart of the Wassers…
The paper derives new gradient and Hessian estimates for nonlinear parabolic equations.
problem Estimating solutions to nonlinear weighted parabolic equations.
method Derives Li-Yau and Hamilton type gradient estimates, and Hessian estimates.
result New gradient and Hessian estimates for positive solutions of nonlinear parabolic equations.
We develop the celebrated semigroup approach à la Bakry et al on Finsler manifolds, where natural Laplacian and heat semigroup are nonlinear, based on the Bochner-Weitzenböck formula established by Sturm and the author. We show the L1-gradient estimate on Finsler manifolds (under some additional assumptions in the n…
We introduce a general method for improving the convergence rate of gradient-based optimizers that is easy to implement and works well in practice. We demonstrate the effectiveness of the method in a range of optimization problems by applying it to stochastic gradient descent, stochastic gradient descent with Nesterov …
New PDMP samplers tackle variable selection in models.
problem Jointly explore model space and parameter space.
method Develop reversible jump PDMP samplers.
result New samplers mix better and are more efficient.
Locally private reinforcement learning protects individual environments from reverse engineering.
problem Protecting private information in distributed reinforcement learning environments.
method Locally differentially private algorithms that protect local agents' models from adversarial reverse engineering.
result Demonstrated that the proposed algorithm performs well under local differential privacy (LDP).
Classifies reversible and strongly reversible elements in quaternionic groups.
problem Classifying reversible and strongly reversible elements in quaternionic groups.
method Proves elements are reversible if and only if they are products of skew-involutions (resp. involutions).
result Proves elements are reversible if and only if they are products of skew-involutions (resp. involutions).
The paper classifies reversible and strongly reversible elements in Hermitian isometry groups.
problem Classifying reversible and strongly reversible elements in Hermitian isometry groups.
method Classification through group theory and algebraic manipulation.
result New classification of strongly reversible elements in Sp(n).
Memory-efficient learning for large-scale imaging systems.
problem Memory limitations in GPUs for real-world large-scale inverse problems.
method Exploits reversibility of network layers to enable data-driven design.
result Demonstrated on small-scale and large-scale real-world systems.
Variational Auto-Encoders (VAEs) have become very popular techniques to perform inference and learning in latent variable models as they allow us to leverage the rich representational power of neural networks to obtain flexible approximations of the posterior of latent variables as well as tight evidence lower bounds (…
This paper classifies reversible and strongly reversible elements in affine groups.
problem Classifying reversible and strongly reversible elements in affine groups.
method Identifying affine transformations and using conjugacy by involutions.
result Classification of reversible and strongly reversible elements in affine groups.
No methods currently exist for making arbitrary neural networks fair. In this work we introduce GRAD, a new and simplified method to producing fair neural networks that can be used for auto-encoding fair representations or directly with predictive networks. It is easy to implement and add to existing architectures, has…
BPN defends against adversarial attacks by generating beneficial perturbations.
problem Adversarial attacks cause deep neural networks to misclassify clean inputs.
method BPN generates beneficial perturbations during training to neutralize future adversarial attacks.
result BPN is robust to adversarial examples and more efficient than classical adversarial training.
New neural net learns time-reversible symplectic dynamics.
problem Lack of time-reversibility in neural networks for symplectic systems.
method Proposes a new neural network architecture for time-reversible symplectic systems.
result Demonstrates learning of time-reversible symplectic dynamics from data.
A new trading strategy using reinforcement learning for statistical arbitrage.
problem Traditional statistical arbitrage models rely on model assumptions and price deviations from a long-term mean.
method Empirical reversion time metric, reinforcement learning framework, and state space optimization.
result Optimal mean reversion strategy identified through reinforcement learning.
Algebraic method reveals criterion for quaternionic Möbius group reversibility.
problem Characterizing reversibility in quaternionic Möbius group elements.
method Purely algebraic approach using matrix entries and conjugacy invariants.
result Explicit criterion for reversibility in terms of matrix entries.
Recent studies have shown that online portfolio selection strategies that exploit the mean reversion property can achieve excess return from equity markets. This paper empirically investigates the performance of state-of-the-art mean reversion strategies on real market data. The aims of the study are twofold. The first…
A Finsler space is said to be geodesically reversible if each oriented geodesic can be reparametrized as a geodesic with the reverse orientation. A reversible Finsler space is geodesically reversible, but the converse need not be true. In this note, building on recent work of LeBrun and Mason, it is shown that a geodes…