Deep neural networks (DNNs) have recently been achieving state-of-the-art performance on a variety of pattern-recognition tasks, most notably visual classification problems. Given that DNNs are now able to classify objects in images with near-human-level performance, questions naturally arise as to what differences remain between computer and human vision. A recent study revealed that changing an image (e.g. of a lion) in a way imperceptible to humans can cause a DNN to label the image as something else entirely (e.g. mislabeling a lion a library). Here we show a related result: it is easy to produce images that are completely unrecognizable to humans, but that state-of-the-art DNNs believe to be recognizable objects with 99.99% confidence (e.g. labeling with certainty that white noise static is a lion). Specifically, we take convolutional neural networks trained to perform well on either the ImageNet or MNIST datasets and then find images with evolutionary algorithms or gradient ascent that DNNs label with high confidence as belonging to each dataset class. It is possible to produce images totally unrecognizable to human eyes that DNNs believe with near certainty are familiar objects, which we call "fooling images" (more generally, fooling examples). Our results shed light on interesting differences between human vision and current DNNs, and raise questions about the generality of DNN computer vision.
The Java implementation of a portfolio of parameter-less evolutionary algorithms is presented. The Parameter-less Evolutionary Portfolio implements a heuristic that performs adaptive selection of parameter-less evolutionary algorithms in accordance with performance criteria that are measured during running time. At present time, the portfolio includes three parameter-less evolutionary algorithms: Parameter-less Univariate Marginal Distribution Algorithm, Parameter-less Extended Compact Genetic Algorithm, and Parameter-less Hierarchical Bayesian Optimization Algorithm. Initial experiments showed that the parameter-less portfolio can solve various classes of problems without the need for any prior parameter setting technique and with an increase in computational effort that can be considered acceptable.
The Parameter-less Genetic Algorithm was first presented by Harik and Lobo in 1999 as an alternative to the usual trial-and-error method of finding, for each given problem, an acceptable set-up of the parameter values of the genetic algorithm. Since then, the same strategy has been successfully applied to create parameter-less versions of other population-based search algorithms such as the Extended Compact Genetic Algorithm and the Hierarchical Bayesian Optimization Algorithm. This report describes a Java implementation, Parameter-less Evolutionary Algorithm (P-EAJava), that integrates several parameter-less evolutionary algorithms into a single platform. Along with a brief description of P-EAJava, we also provide detailed instructions on how to use it, how to implement new problems, and how to generate new parameter-less versions of evolutionary algorithms. At present time, P-EAJava already includes parameter-less versions of the Simple Genetic Algorithm, the Extended Compact Genetic Algorithm, the Univariate Marginal Distribution Algorithm, and the Hierarchical Bayesian Optimization Algorithm. The source and binary files of the Java implementation of P-EAJava are available for free download at https://github.com/JoseCPereira/2015ParameterlessEvolutionaryAlgorithmsJava.
The Simple Genetic Algorithm, the Univariate Marginal Distribution Algorithm, the Extended Compact Genetic Algorithm, and the Hierarchical Bayesian Optimization Algorithm are all well known Evolutionary Algorithms. In this report we present a Java implementation of these four algorithms with detailed instructions on how to use each of them to solve a given set of optimization problems. Additionally, it is explained how to implement and integrate new problems within the provided set. The source and binary files of the Java implementations are available for free download at https://github.com/JoseCPereira/2015EvolutionaryAlgorithmsJava.
It is quite usual when an evolutionary algorithm tool or library uses a language other than C, C++, Java or Matlab that a reviewer or the audience questions its usefulness based on the speed of those other languages, purportedly slower than the aforementioned ones. Despite speed being not everything needed to design a useful evolutionary algorithm application, in this paper we will measure the speed for several very basic evolutionary algorithm operations in several languages which use different virtual machines and approaches, and prove that, in fact, there is no big difference in speed between interpreted and compiled languages, and that in some cases, interpreted languages such as JavaScript or Python can be faster than compiled languages such as Scala, making them worthy of use for evolutionary algorithm experimentation.
Perturb and Combine (P&C) group of methods generate multiple versions of the predictor by perturbing the training set or construction and then combining them into a single predictor (Breiman, 1996b). The motive is to improve the accuracy in unstable classification and regression methods. One of the most well known method in this group is Bagging. Arcing or Adaptive Resampling and Combining methods like AdaBoost are smarter variants of P&C methods. In this extended abstract, we lay the groundwork for a new family of methods under the P&C umbrella, known as Evolutionary Sampling (ES). We employ Evolutionary algorithms to suggest smarter sampling in both the feature space (sub-spaces) as well as training samples. We discuss multiple fitness functions to assess ensembles and empirically compare our performance against randomized sampling of training data and feature sub-spaces.
Multi-Objective Optimization Problems (MOPs) have attracted growing attention during the last decades. Multi-Objective Evolutionary Algorithms (MOEAs) have been extensively used to address MOPs because are able to approximate a set of non-dominated high-quality solutions. The Multi-Objective Quadratic Assignment Problem (mQAP) is a MOP. The mQAP is a generalization of the classical QAP which has been extensively studied, and used in several real-life applications. The mQAP is defined as having as input several flows between the facilities which generate multiple cost functions that must be optimized simultaneously. In this study, we propose PasMoQAP, a parallel asynchronous memetic algorithm to solve the Multi-Objective Quadratic Assignment Problem. PasMoQAP is based on an island model that structures the population by creating sub-populations. The memetic algorithm on each island individually evolve a reduced population of solutions, and they asynchronously cooperate by sending selected solutions to the neighboring islands. The experimental results show that our approach significatively outperforms all the island-based variants of the multi-objective evolutionary algorithm NSGA-II. We show that PasMoQAP is a suitable alternative to solve the Multi-Objective Quadratic Assignment Problem.
The concept of gray-box optimization, in juxtaposition to black-box optimization, revolves about the idea of exploiting the problem structure to implement more efficient evolutionary algorithms (EAs). Work on factorized distribution algorithms (FDAs), whose factorizations are directly derived from the problem structure, has also contributed to show how exploiting the problem structure produces important gains in the efficiency of EAs. In this paper we analyze the general question of using problem structure in EAs focusing on confronting work done in gray-box optimization with related research accomplished in FDAs. This contrasted analysis helps us to identify, in current studies on the use problem structure in EAs, two distinct analytical characterizations of how these algorithms work. Moreover, we claim that these two characterizations collide and compete at the time of providing a coherent framework to investigate this type of algorithms. To illustrate this claim, we present a contrasted analysis of formalisms, questions, and results produced in FDAs and gray-box optimization. Common underlying principles in the two approaches, which are usually overlooked, are identified and discussed. Besides, an extensive review of previous research related to different uses of the problem structure in EAs is presented. The paper also elaborates on some of the questions that arise when extending the use of problem structure in EAs, such as the question of evolvability, high cardinality of the variables and large definition sets, constrained and multi-objective problems, etc. Finally, emergent approaches that exploit neural models to capture the problem structure are covered.
We propose a new method for learning the structure of convolutional neural networks (CNNs) that is more efficient than recent state-of-the-art methods based on reinforcement learning and evolutionary algorithms. Our approach uses a sequential model-based optimization (SMBO) strategy, in which we search for structures in order of increasing complexity, while simultaneously learning a surrogate model to guide the search through structure space. Direct comparison under the same search space shows that our method is up to 5 times more efficient than the RL method of Zoph et al. (2018) in terms of number of models evaluated, and 8 times faster in terms of total compute. The structures we discover in this way achieve state of the art classification accuracies on CIFAR-10 and ImageNet.
We introduce a novel approach for discriminative classification using evolutionary algorithms. We first propose an algorithm to optimize the total loss value using a modified 0-1 loss function in a one-dimensional space for classification. We then extend this algorithm for multi-dimensional classification using an evolutionary algorithm. The proposed evolutionary algorithm aims to find a hyperplane which best classifies instances while minimizes the classification risk. We test particle swarm optimization, evolutionary strategy, and covariance matrix adaptation evolutionary strategy for optimization purpose. Finally, we compare our results with well-established and state-of-the-art classification algorithms, for both binary and multi-class classification, on 19 benchmark classification problems, with and without noise and outliers. Results show that the performance of the proposed algorithm is significantly (t-test) better than all other methods in almost all problems tested. We also show that the proposed algorithm is significantly more robust against noise and outliers comparing to other methods. The running time of the algorithm is within a reasonable range for the solution of real-world classification problems.
Recent advances in deep neuroevolution have demonstrated that evolutionary algorithms, such as evolution strategies (ES) and genetic algorithms (GA), can scale to train deep neural networks to solve difficult reinforcement learning (RL) problems. However, it remains a challenge to analyze and interpret the underlying process of neuroevolution in such high dimensions. To begin to address this challenge, this paper presents an interactive data visualization tool called VINE (Visual Inspector for NeuroEvolution) aimed at helping neuroevolution researchers and end-users better understand and explore this family of algorithms. VINE works seamlessly with a breadth of neuroevolution algorithms, including ES and GA, and addresses the difficulty of observing the underlying dynamics of the learning process through an interactive visualization of the evolving agent's behavior characterizations over generations. As neuroevolution scales to neural networks with millions or more connections, visualization tools like VINE that offer fresh insight into the underlying dynamics of evolution become increasingly valuable and important for inspiring new innovations and applications.
Deep Reinforcement Learning (DRL) algorithms have been successfully applied to a range of challenging control tasks. However, these methods typically suffer from three core difficulties: temporal credit assignment with sparse rewards, lack of effective exploration, and brittle convergence properties that are extremely sensitive to hyperparameters. Collectively, these challenges severely limit the applicability of these approaches to real-world problems. Evolutionary Algorithms (EAs), a class of black box optimization techniques inspired by natural evolution, are well suited to address each of these three challenges. However, EAs typically suffer from high sample complexity and struggle to solve problems that require optimization of a large number of parameters. In this paper, we introduce Evolutionary Reinforcement Learning (ERL), a hybrid algorithm that leverages the population of an EA to provide diversified data to train an RL agent, and reinserts the RL agent into the EA population periodically to inject gradient information into the EA. ERL inherits EA's ability of temporal credit assignment with a fitness metric, effective exploration with a diverse set of policies, and stability of a population-based approach and complements it with off-policy DRL's ability to leverage gradients for higher sample efficiency and faster learning. Experiments in a range of challenging continuous control benchmarks demonstrate that ERL significantly outperforms prior DRL and EA methods.
Deep neural networks are vulnerable to adversarial examples, even in the black-box setting, where the attacker is restricted solely to query access. Existing black-box approaches to generating adversarial examples typically require a significant number of queries, either for training a substitute network or performing gradient estimation. We introduce GenAttack, a gradient-free optimization technique that uses genetic algorithms for synthesizing adversarial examples in the black-box setting. Our experiments on different datasets (MNIST, CIFAR-10, and ImageNet) show that GenAttack can successfully generate visually imperceptible adversarial examples against state-of-the-art image recognition models with orders of magnitude fewer queries than previous approaches. Against MNIST and CIFAR-10 models, GenAttack required roughly 2,126 and 2,568 times fewer queries respectively, than ZOO, the prior state-of-the-art black-box attack. In order to scale up the attack to large-scale high-dimensional ImageNet models, we perform a series of optimizations that further improve the query efficiency of our attack leading to 237 times fewer queries against the Inception-v3 model than ZOO. Furthermore, we show that GenAttack can successfully attack some state-of-the-art ImageNet defenses, including ensemble adversarial training and non-differentiable or randomized input transformations. Our results suggest that evolutionary algorithms open up a promising area of research into effective black-box attacks.
Time series prediction typically consists of a data reconstruction phase where the time series is broken into overlapping windows known as the timespan. The size of the timespan can be seen as a way of determining the extent of past information required for an effective prediction. In certain applications such as the prediction of wind-intensity of storms and cyclones, prediction models need to be dynamic in accommodating different values of the timespan. These applications require robust prediction as soon as the event takes place. We identify a new category of problem called dynamic time series prediction that requires a model to give prediction when presented with varying lengths of the timespan. In this paper, we propose a co-evolutionary multi-task learning method that provides a synergy between multi-task learning and co-evolutionary algorithms to address dynamic time series prediction. The method features effective use of building blocks of knowledge inspired by dynamic programming and multi-task learning. It enables neural networks to retain modularity during training for making a decision in situations even when certain inputs are missing. The effectiveness of the method is demonstrated using one-step-ahead chaotic time series and tropical cyclone wind-intensity prediction.