There is, I think, something ethereal about i —the square root of minus one. I remember first hearing about it at school. It seemed an odd beast at that time—an intruder hovering on the edge of reality.

Usually familiarity dulls this sense of the bizarre, but in the case of i it was the reverse: over the years the sense of its surreal nature intensified. It seemed that it was impossible to write mathematics that described the real world in …

I and i

Machine Learning is the study of methods for programming computers to learn. Computers are applied to a wide range of tasks, and for most of these it is relatively easy for programmers to design and implement the necessary software. However, there are many tasks for which this is difficult or impossible. These can be divided into four general categories. First, there are problems for which there exist no human experts. For example, in modern automated manufacturing facilities, there is a need to predict machine failures before they occur by analyzing sensor readings. Because the machines are new, there are no human experts who can be interviewed by a programmer to provide the knowledge necessary to build a computer system. A machine learning system can study recorded data and subsequent machine failures and learn prediction rules. Second, there are problems where human experts exist, but where they are unable to explain their expertise. This is the case in many perceptual tasks, such as speech recognition, hand-writing recognition, and natural language understanding. Virtually all humans exhibit expert-level abilities on these tasks, but none of them can describe the detailed steps that they follow as they perform them. Fortunately, humans can provide machines with examples of the inputs and correct outputs for these tasks, so machine learning algorithms can learn to map the inputs to the outputs. Third, there are problems where phenomena are changing rapidly. In finance, for example, people would like to predict the future behavior of the stock market, of consumer purchases, or of exchange rates. These behaviors change frequently, so that even if a programmer could construct a good predictive computer program, it would need to be rewritten frequently. A learning program can relieve the programmer of this burden by constantly modifying and tuning a set of learned prediction rules. Fourth, there are applications that need to be customized for each computer user separately. Consider, for example, a program to filter unwanted electronic mail messages. Different users will need different filters. It is unreasonable to expect each user to program his or her own rules, and it is infeasible to provide every user with a software engineer to keep the rules up-to-date. A machine learning system can learn which mail messages the user rejects and maintain the filtering rules automatically. Machine learning addresses many of the same research questions as the fields of statistics, data mining, and psychology, but with differences of emphasis. Statistics focuses on understanding the phenomena that have generated the data, often with the goal of testing different hypotheses about those phenomena. Data mining seeks to find patterns in the data that are understandable by people. Psychological studies of human learning aspire to understand the mechanisms underlying the various learning behaviors exhibited by people (concept learning, skill acquisition, strategy change, etc.).

Machine learning

We will review some of the major results in random graphs and some of the more challenging open problems. We will cover algorithmic and structural questions. We will touch on newer models, including those related to the WWW.

Random graphs

Deep convolutional neural networks have performed remarkably well on many Computer Vision tasks. However, these networks are heavily reliant on big data to avoid overfitting. Overfitting refers to the phenomenon when a network learns a function with very high variance such as to perfectly model the training data. Unfortunately, many application domains do not have access to big data, such as medical image analysis. This survey focuses on Data Augmentation, a data-space solution to the problem of limited data. Data Augmentation encompasses a suite of techniques that enhance the size and quality of training datasets such that better Deep Learning models can be built using them. The image augmentation algorithms discussed in this survey include geometric transformations, color space augmentations, kernel filters, mixing images, random erasing, feature space augmentation, adversarial training, generative adversarial networks, neural style transfer, and meta-learning. The application of augmentation methods based on GANs are heavily covered in this survey. In addition to augmentation techniques, this paper will briefly discuss other characteristics of Data Augmentation such as test-time augmentation, resolution impact, final dataset size, and curriculum learning. This survey will present existing methods for Data Augmentation, promising developments, and meta-level decisions for implementing Data Augmentation. Readers will understand how Data Augmentation can improve the performance of their models and expand limited datasets to take advantage of the capabilities of big data.

/pdf/a-survey-on-image-data-augmentation-for-deep-learning-4ru1bj2hbb.pdf

A survey on Image Data Augmentation for Deep Learning

An overview of the self-organizing map algorithm, on which the papers in this issue are based, is presented in this article.

The Self-Organizing Map

In this study we develop an Agent-based Model (ABM), called Neo-COOP, to investigate the emergence and evolution of altruistic and selfish behaviour in Neolithic-inspired household agents under varying degrees of environmental stress. We conduct scenario experimentation where we track the evolution of the agents' resource trading preferences in scenarios with varying frequencies of environmental stress and agent types initialized to exhibit differing altruistic or selfish tendencies. Our results suggest that neither extreme selfishness or extreme altruism is desirable but rather, some middle-ground value is. Additionally, we find that the frequency of the environmental stress plays a significant role in the emergence of selfish behaviour amongst the social elite with higher frequency environmental stress scenarios resulting in a greater disparity of resource transfer beliefs between agents with equal social status.

Do harsher environments cause selfish or altruistic behavior?

Agent Based Modeling (ABM) is a bottom-up approach that has been used to study adaptive group (collective) behavior. ABM is an analogical system that aids ethologists in constructing novel hypotheses, and allows the investigation of emergent phenomena in experiments that could not be conducted in nature [15], [2], [12], [11]. Many studies in ethology have formalized mathematical models of collective migration behavior [1], but few have examined the impact of phenotypic traits (such as lifetime length) on the learning and evolution of collective migration behavior [9], [4]. The first objective of this research is to test the impact of agent lifetime length on the adaptation of collective migration behaviors in a virtual environment. Agent behavior is adapted with a hybrid Particle Swarm Optimization (PSO) method that integrates learning and evolution. Learning (lifetime learning) refers to a process whereby agents learn new behaviors during their lifetime [13], [3]. Evolution (genetic learning) refers to behavioral adaptation over successive lifetimes (generations) of an agent population [5]. The second objective is to demonstrate these hybrid PSO methods are appropriate for modeling the adaptation of collective migration behaviors in an ABM. The motivation is that PSO methods combined with evolution and learning approaches have received little attention as ABMs for potentially addressing (supporting or refuting) hypotheses posited in ethological literature. The task was for an agent group (flock) to locate a migration point during a simulated season in a virtual environment, where a season consisted of X simulation iterations.

Lifetimes of migration

In nature, morphological diversity enhances functional diversity, however, there is little swarm (collective) robotics research on the impact of morphological and behavioral (body-brain) diversity that emerges in response to changing environments. This study investigates the impact of increasingly complex task environments on the artificial evolution of body-brain diversity in simulated robot swarms. We investigate whether increasing task environment complexity (collective behavior tasks requiring increasing degrees of cooperative behavior) mandates concurrent increases in behavioral, morphological, or coupled increases in body-brain diversity in robotic swarms. Experiments compared three variants of collective behavior evolution across increasingly complex task environments: two behavioral diversity maintenance variants and body-brain diversity maintenance. Results indicate that body-brain diversity maintenance yielded a significantly higher behavioral and morphological diversity in evolved swarms overall, which was beneficial in the most complex task environment.

The Impact of Morphological Diversity in Robot Swarms

This study evaluates objective versus non-objective-based evolutionary search methods for behavior evolution in robot teams. The goal is to evaluate the morphological robustness of evolved controllers, where controllers are evolved for specific robot sensory-motor configurations (morphologies) but must continue to function as these morphologies degrade. Robots use artificial neural network controllers where behavior evolution is directed by developmental neuro-evolution. Guiding evolutionary controller design we use objective (fitness function) versus non-objective (novelty) search. The former optimizes for behavioral fitness and the latter for behavioral novelty. These methods are evaluated across varying robot morphologies and increasing task complexity. Results indicate that novelty search yields no benefits over objective search, in terms of evolving morphologically robust controllers. That is, both novelty and objective search evolve team controllers that are morphologically robust given varying robot morphologies and increasing task complexity. Results thus suggest behavioral diversity methods such as novelty search mat not be suitable for generating robot behaviors that can continue functioning given changing robot morphologies, for example, due to damaged or disabled sensors and actuators.

Objective versus Non-Objective Search in Evolving Morphologically Robust Robot Controllers

This study evaluates comparative behavioral search methods for evolutionary controller design in robot teams, where the goal is to evaluate the morphological robustness of evolved controllers. That is, where controllers are evolved for specific robot sensory-motor configurations (morphologies) but must continue to function as these morphologies degrade. Robots use neural controllers where behavior evolution is directed by developmental neuro-evolution (HyperNEAT). Guiding evolutionary controller design we use objective fitness function) versus non-objective (novelty) search. The former optimizes for behavioral fitness and the latter for behavioral novelty. These search methods are evaluated across varying robot morphologies and increasing task complexity. Results indicate that both novelty and objective search evolve team controllers (behaviors) that are morphologically robust given degrading robot morphologies and increasing task complexity. Results thus suggest that novelty search is not necessarily suitable for generating robot team behaviors that are robust to changes in robot morphologies (for example, due to damaged or disabled sensors and actuators).

Geoff Nitschke

Papers

Do harsher environments cause selfish or altruistic behavior?

Lifetimes of migration

The Impact of Morphological Diversity in Robot Swarms

Objective versus Non-Objective Search in Evolving Morphologically Robust Robot Controllers

Is Novelty Search Good for Evolving Morphologically Robust Robot Controllers