Deep learning applications require optimization of nonconvex objective functions. These functions have multiple local minima and their optimization is a challenging problem. Simulated Annealing is a well-established method for optimization of such functions, but its efficiency depends on the efficiency of the adapted sampling methods. We explore relations between the Langevin dynamics and stochastic optimization. By combining the Momentum optimizer with Simulated Annealing, we propose CoolMomentum - a prospective stochastic optimization method. Empirical results confirm the efficiency of the proposed theoretical approach.

We describe a necessary and sufficient condition for the convergence to minimum Bayes risk when training two-layer ReLU-networks by gradient descent in the mean field regime with omni-directional initial parameter distribution. This article extends recent results of Chizat and Bach to ReLU-activated networks and to the situation in which there are no parameters which exactly achieve MBR. The condition does not depend on the initalization of parameters and concerns only the weak convergence of the realization of the neural network, not its parameter distribution.

Previous work has examined the ability of larger capacity neural networks to generalize better than smaller ones, even without explicit regularizers, by analyzing gradient based algorithms such as GD and SGD. The presence of noise and its effect on robustness to parameter perturbations has been linked to generalization. We examine a property of GD and SGD, namely that instead of iterating through all scalar weights in the network and updating them one by one, GD (and SGD) updates all the parameters at the same time. As a result, each parameter $w^i$ calculates its partial derivative at the stale parameter $\mathbf{w_t}$, but then suffers loss $\hat{L}(\mathbf{w_{t+1}})$. We show that this causes noise to be introduced into the optimization. We find that this noise penalizes models that are sensitive to perturbations in the weights. We find that penalties are most pronounced for batches that are currently being used to update, and are higher for larger models.

In this paper, we design and analyze a new family of adaptive subgradient methods for solving an important class of weakly convex (possibly nonsmooth) stochastic optimization problems. Adaptive methods that use exponential moving averages of past gradients to update search directions and learning rates have recently attracted a lot of attention for solving optimization problems that arise in machine learning. Nevertheless, their convergence analysis almost exclusively requires smoothness and/or convexity of the objective function. In contrast, we establish non-asymptotic rates of convergence of first and zeroth-order adaptive methods and their proximal variants for a reasonably broad class of nonsmooth \& nonconvex optimization problems. Experimental results indicate how the proposed algorithms empirically outperform stochastic gradient descent and its zeroth-order variant for solving such optimization problems.

This video will help you understand Stochastic Gradient Descent in Deep Neural Network in a very simplified manner. Deep learning is part of a broader family of machine learning methods based on artificial neural networks with representation learning. Learning can be supervised, semi-supervised or unsupervised. Get 10% flat off on the above complete course with certification - http://bit.ly/39trxCf (APPLY COUPON - YTDEG) Get 15% flat off on the these AI/ML courses with certification - (APPLY COUPON - YTEDU) 1.Learn Machine Learning By Building Projects - http://bit.ly/2MxMSSl 2.The Complete Web Development Course - Build 15 Projects - http://bit.ly/32Ah9oW 3.The Full Stack Web Development - http://bit.ly/2MZDBRV 4.Projects In Laravel: Learn Laravel Building 10 Projects - http://bit.ly/2MAiHtH 5.Mathematical Foundation For Machine Learning and AI - http://bit.ly/2N23Eb1 Get 10% flat off on the Below full E-Degree with certification - (APPLY COPOUN - YTDEG) Advance Artificial Intelligence & Machine Learning E-Degree - http://bit.ly/38mbiXm

Many statistical estimators are defined as the fixed point of a data-dependent operator, with estimators based on minimizing a cost function being an important special case. The limiting performance of such estimators depends on the properties of the population-level operator in the idealized limit of infinitely many samples. We develop a general framework that yields bounds on statistical accuracy based on the interplay between the deterministic convergence rate of the algorithm at the population level, and its degree of (in)stability when applied to an empirical object based on $n$ samples. Using this framework, we analyze both stable forms of gradient descent and some higher-order and unstable algorithms, including Newton's method and its cubic-regularized variant, as well as the EM algorithm. We provide applications of our general results to several concrete classes of models, including Gaussian mixture estimation, single-index models, and informative non-response models. We exhibit cases in which an unstable algorithm can achieve the same statistical accuracy as a stable algorithm in exponentially fewer steps---namely, with the number of iterations being reduced from polynomial to logarithmic in sample size $n$.

In this paper, novel gradient based online learning algorithms are developed to investigate an important environmental application: real-time river pollution source identification, which aims at estimating the released mass, the location and the released time of a river pollution source based on downstream sensor data monitoring the pollution concentration. The problem can be formulated as a non-convex loss minimization problem in statistical learning, and our online algorithms have vectorized and adaptive step-sizes to ensure high estimation accuracy on dimensions having different magnitudes. In order to avoid gradient-based method sticking into the saddle points of non-convex loss, the "escaping from saddle points" module and multi-start version of algorithms are derived to further improve the estimation accuracy by searching for the global minimimals of the loss functions. It can be shown theoretically and experimentally $O(N)$ local regret of the algorithms, and the high probability cumulative regret bound $O(N)$ under particular error bound condition on loss functions. A real-life river pollution source identification example shows superior performance of our algorithms than the existing methods in terms of estimating accuracy. The managerial insights for decision maker to use the algorithm in reality are also provided.

This paper tackles the age-old question of derivative free optimization in neural networks. This paper introduces AdaSwarm, a novel derivative-free optimizer to have similar or better performance to Adam but without "gradients". To support the AdaSwarm, a novel Particle Swarm Optimization Exponentially weighted Momentum PSO (EM-PSO), a derivative-free optimizer, is also proposed which tackles constrained and unconstrained single objective optimization problems and looks at applying the proposed momentum particle swarm optimization on benchmark test functions, engineering optimization problems and habitability scores for exoplanets which show speed and convergence of the technique. The EM-PSO is extended by approximating the gradient of a function at any point using the parameters of the particle swarm optimization. This is a novel technique to simulate gradient descent, an extremely popular method in the back-propagation algorithm, using the approximated gradients from the particle swarm optimization parameters. Mathematical proofs of gradient approximation by EM-PSO, thereby bypassing the gradient computation, are presented. The AdaSwarm is compared with various optimizers and the theory and algorithmic performance are supported by promising results.

Many aspects of the geometry of loss functions in deep learning remain mysterious. In this paper, we work toward a better understanding of the geometry of the loss function $L$ of overparameterized feedforward neural networks. In this setting, we identify several components of the critical locus of $L$ and study their geometric properties. For networks of depth $\ell \geq 4$, we identify a locus of critical points we call the star locus $S$. Within $S$ we identify a positive-dimensional sublocus $C$ with the property that for $p \in C$, $p$ is a degenerate critical point, and no existing theoretical result guarantees that gradient descent will not converge to $p$. For very wide networks, we build on earlier work and show that all critical points of $L$ are degenerate, and give lower bounds on the number of zero eigenvalues of the Hessian at each critical point. For networks that are both deep and very wide, we compare the growth rates of the zero eigenspaces of the Hessian at all the different families of critical points that we identify. The results in this paper provide a starting point to a more quantitative understanding of the properties of various components of the critical locus of $L$.

In this paper we propose an efficient stochastic optimization algorithm to search for Bayesian experimental designs such that the expected information gain is maximized. The gradient of the expected information gain with respect to experimental design parameters is given by a nested expectation, for which the standard Monte Carlo method using a fixed number of inner samples yields a biased estimator. In this paper, applying the idea of randomized multilevel Monte Carlo methods, we introduce an unbiased Monte Carlo estimator for the gradient of the expected information gain with finite expected squared $\ell_2$-norm and finite expected computational cost per sample. Our unbiased estimator can be combined well with stochastic gradient descent algorithms, which results in our proposal of an optimization algorithm to search for an optimal Bayesian experimental design. Numerical experiments confirm that our proposed algorithm works well not only for a simple test problem but also for a more realistic pharmacokinetic problem.