We study the problem of machine unlearning and identify a notion of algorithmic stability, Total Variation (TV) stability, which we argue, is suitable for the goal of exact unlearning. For convex risk minimization problems, we design TV-stable algorithms based on noisy Stochastic Gradient Descent (SGD). Our key contribution is the design of corresponding efficient unlearning algorithms, which are based on constructing a (maximal) coupling of Markov chains for the noisy SGD procedure. To understand the trade-offs between accuracy and unlearning efficiency, we give upper and lower bounds on excess empirical and populations risk of TV stable algorithms for convex risk minimization. Our techniques generalize to arbitrary non-convex functions, and our algorithms are differentially private as well.

Nonconvex optimization has been widely adopted in various domains, including image recognition (Hinton et al., 2012; Krizhevsky et al., 2012), Bayesian graphical models (Jordan et al., 2004; Attias, 2000), recommendation systems (Salakhutdinov et al., 2007), etc. Despite the fact that solving a nonconvex problem is generally difficult, empirical evidences have shown that simple first order algorithms such as stochastic gradient descent (SGD), are able to solve a majority of the aforementioned nonconvex problems efficiently. The theory behind these empirical observations, however, is still largely unexplored. In classical optimization literature, there have been fruitful results on characterizing the convergence of SGD to first-order stationary points for nonconvex problems. However, these types of results fall short of explaining the empirical evidences that SGD often converges to global minima for a wide class of nonconvex problems used in practice. More recently, understanding the role of noise in the algorithmic behavior of SGD has received significant attention. For instance, Jin et al. (2017) show that a perturbed form of gradient descent is able to escape from strict saddle points and converge to second-order stationary points (i.e., local minima). Zhou et al. (2019) further show that noise in the update can help SGD to escape from spurious local minima and converge to the global minima.

Deep generative models have emerged as a powerful class of priors for signals in various inverse problems such as compressed sensing, phase retrieval and super-resolution. Here, we assume an unknown signal to lie in the range of some pre-trained generative model. A popular approach for signal recovery is via gradient descent in the low-dimensional latent space. While gradient descent has achieved good empirical performance, its theoretical behavior is not well understood. In this paper, we introduce the use of stochastic gradient Langevin dynamics (SGLD) for compressed sensing with a generative prior. Under mild assumptions on the generative model, we prove the convergence of SGLD to the true signal. We also demonstrate competitive empirical performance to standard gradient descent.

The integration of reasoning, learning, and decision-making is key to build more general AI systems. As a step in this direction, we propose a novel neural-logic architecture that can solve both inductive logic programming (ILP) and deep reinforcement learning (RL) problems. Our architecture defines a restricted but expressive continuous space of first-order logic programs by assigning weights to predicates instead of rules. Therefore, it is fully differentiable and can be efficiently trained with gradient descent. Besides, in the deep RL setting with actor-critic algorithms, we propose a novel efficient critic architecture. Compared to state-of-the-art methods on both ILP and RL problems, our proposition achieves excellent performance, while being able to provide a fully interpretable solution and scaling much better, especially during the testing phase.

Gradient descent is one of the most common machine learning algorithms used in neural networks [7], data science, optimization, and machine learning tasks. The gradient descent algorithm and its variants can be found in almost every machine learning model. Gradient descent is a popular optimization method of tuning the parameters in a machine learning model. Its goal is to apply optimization to find the least or minimal error value. It is mostly used to update the parameters of the model -- in this case, parameters refer to coefficients in regression and weights in a neural network.

An increasing number of machine learning applications employ flexible overparameterized models to fit the training data. Theoretical analysis of such'overfitted' solutions has been a recent focus of the learning community. It is conjectured that the use of large overparameterized neural networks makes the loss landscape amenable to optimization through local search methods, such as stochastic gradient descent. It is also hypothesized that implicit regularization, arising from the choice of the optimization algorithm and the neural network architecture, mitigates the large complexity and ensures that the'overfitted' solutions generalize. Suppose a'simple' class H of models captures the relationship between the covariates X and the response variable Y. Inspired by the use of overparameterized models, we may take a much larger class F H for computational or other purposes (such as lack of explicit description of H) and minimize training loss over this larger class.

Recent studies have provided both empirical and theoretical evidence illustrating that heavy tails can emerge in stochastic gradient descent (SGD) in various scenarios. Such heavy tails potentially result in iterates with diverging variance, which hinders the use of conventional convergence analysis techniques that rely on the existence of the second-order moments. In this paper, we provide convergence guarantees for SGD under a state-dependent and heavy-tailed noise with a potentially infinite variance, for a class of strongly convex objectives. In the case where the $p$-th moment of the noise exists for some $p\in [1,2)$, we first identify a condition on the Hessian, coined '$p$-positive (semi-)definiteness', that leads to an interesting interpolation between positive semi-definite matrices ($p=2$) and diagonally dominant matrices with non-negative diagonal entries ($p=1$). Under this condition, we then provide a convergence rate for the distance to the global optimum in $L^p$. Furthermore, we provide a generalized central limit theorem, which shows that the properly scaled Polyak-Ruppert averaging converges weakly to a multivariate $\alpha$-stable random vector. Our results indicate that even under heavy-tailed noise with infinite variance, SGD can converge to the global optimum without necessitating any modification neither to the loss function or to the algorithm itself, as typically required in robust statistics. We demonstrate the implications of our results to applications such as linear regression and generalized linear models subject to heavy-tailed data.

This paper studies fixed step-size stochastic approximation (SA) schemes, including stochastic gradient schemes, in a Riemannian framework. It is motivated by several applications, where geodesics can be computed explicitly, and their use accelerates crude Euclidean methods. A fixed step-size scheme defines a family of time-homogeneous Markov chains, parametrized by the step-size. Here, using this formulation, non-asymptotic performance bounds are derived, under Lyapunov conditions. Then, for any step-size, the corresponding Markov chain is proved to admit a unique stationary distribution, and to be geometrically ergodic. This result gives rise to a family of stationary distributions indexed by the step-size, which is further shown to converge to a Dirac measure, concentrated at the solution of the problem at hand, as the step-size goes to 0. Finally, the asymptotic rate of this convergence is established, through an asymptotic expansion of the bias, and a central limit theorem.

The Kalman filter is a fundamental filtering algorithm that fuses noisy sensory data, a previous state estimate, and a dynamics model to produce a principled estimate of the current state. It assumes, and is optimal for, linear models and white Gaussian noise. Due to its relative simplicity and general effectiveness, the Kalman filter is widely used in engineering applications. Since many sensory problems the brain faces are, at their core, filtering problems, it is possible that the brain possesses neural circuitry that implements equivalent computations to the Kalman filter. The standard approach to Kalman filtering requires complex matrix computations that are unlikely to be directly implementable in neural circuits. In this paper, we show that a gradient-descent approximation to the Kalman filter requires only local computations with variance weighted prediction errors. Moreover, we show that it is possible under the same scheme to adaptively learn the dynamics model with a learning rule that corresponds directly to Hebbian plasticity. We demonstrate the performance of our method on a simple Kalman filtering task, and propose a neural implementation of the required equations. The Bayesian Brain hypothesis has gained significant traction in cognitive neuroscience over the last two decades (Knill and Pouget, 2004; Doya et al., 2007; Pouget et al., 2013).

In this paper, we proposed a new technique, {\em variance controlled stochastic gradient} (VCSG), to improve the performance of the stochastic variance reduced gradient (SVRG) algorithm. To avoid over-reducing the variance of gradient by SVRG, a hyper-parameter $\lambda$ is introduced in VCSG that is able to control the reduced variance of SVRG. Theory shows that the optimization method can converge by using an unbiased gradient estimator, but in practice, biased gradient estimation can allow more efficient convergence to the vicinity since an unbiased approach is computationally more expensive. $\lambda$ also has the effect of balancing the trade-off between unbiased and biased estimations. Secondly, to minimize the number of full gradient calculations in SVRG, a variance-bounded batch is introduced to reduce the number of gradient calculations required in each iteration. For smooth non-convex functions, the proposed algorithm converges to an approximate first-order stationary point (i.e. $\mathbb{E}\|\nabla{f}(x)\|^{2}\leq\epsilon$) within $\mathcal{O}(min\{1/\epsilon^{3/2},n^{1/4}/\epsilon\})$ number of stochastic gradient evaluations, which improves the leading gradient complexity of stochastic gradient-based method SCS $(\mathcal{O}(min\{1/\epsilon^{5/3},n^{2/3}/\epsilon\})$. It is shown theoretically and experimentally that VCSG can be deployed to improve convergence.