Joshua Clifford Bongard

Morphological change in machines accelerates the evolution of robust behavior

Most animals exhibit significant neurological and morphological change throughout their lifetime. No robots to date, however, grow new morphological structure while behaving. This is due to technological limitations but also because it is unclear that morphological change provides a benefit to the acquisition of robust behavior in machines. Here I show that in evolving populations of simulated robots, if robots grow from anguilliform into legged robots during their lifetime in the early stages of evolution, and the anguilliform body plan is gradually lost during later stages of evolution, gaits are evolved for the final, legged form of the robot more rapidly—and the evolved gaits are more robust—compared to evolving populations of legged robots that do not transition through the anguilliform body plan. This suggests that morphological change, as well as the evolution of development, are two important processes that improve the automatic generation of robust behaviors for machines. It also provides an experimental platform for investigating the relationship between the evolution of development and robust behavior in biological organisms.

Improving genetic programming based symbolic regression using deterministic machine learning

Symbolic regression (SR) is a well studied method in genetic programming (GP) for discovering free-form mathematical models from observed data. However, it has not been widely accepted as a standard data science tool. The reluctance is in part due to the hard to analyze random nature of GP and scalability issues. On the other hand, most popular deterministic regression algorithms were designed to generate linear models and therefore lack the flexibility of GP based SR (GP-SR). Our hypothesis is that hybridizing these two techniques will create a synergy between the GP-SR and deterministic approaches to machine learning, which might help bring the GP based techniques closer to the realm of big learning. In this paper, we show that a hybrid deterministic/GP-SR algorithm outperforms GP-SR alone and the state-of-the-art deterministic regression technique alone on a set of multivariate polynomial symbolic regression tasks as the system to be modeled becomes more multivariate.

Avoiding local optima with user demonstrations and low-level control

Interactive Evolutionary Algorithms (IEAs) use human input to help drive a search process. Traditionally, IEAs allow the user to exhibit preferences among some set of individuals. Here we present a system in which the user directly demonstrates what he or she prefers. Demonstration has an advantage over preferences because the user can provide the system with a solution that would never have been presented to a user who can only provide preferences. However, demonstration exacerbates the user fatigue problem because it is more taxing than exhibiting preferences. The system compensates for this by retaining and reusing the user demonstration, similar in spirit to user modeling. The system is exercised on a robot locomotion and obstacle avoidance task that has an obvious local optimum. The user demonstration is provided through low-level control. The system is compared against a high-level fitness function that is susceptible to becoming trapped by a local optimum and a mid-level fitness function designed to remove the local optimum. We show that our proposed system outperforms most variants of these completely automatic methods, providing further evidence that Evolutionary Robotics (ER) can benefit by combining the intuitions of inexpert human users with the search capabilities of computers.

Exploiting the Relationship Between Structural Modularity and Sparsity for Faster Network Evolution

A network is structurally modular if it can be divided into tightly connected groups which are weakly connected or disconnected from each other. Such networks are known to be capable of fast evolutionary adaptation, which makes modularity a highly desired property of networks involved in evolutionary computation. Modularity is known to correlate positively with another network parameter, sparsity. Based on this relationship, we hypothesize that starting evolution with a population of sparse networks should increase the modularity of the evolved networks. We find that this technique can enhance the performance of an existing technique for modularity evolution, multiobjective performance-connection cost (P&CC) optimization, and enable a multiobjective algorithm which maximizes performance and minimizes time since last mutation to produce modular solutions almost as efficiently as the P&CC optimization does.

Modularity and Sparsity: Evolution of Neural Net Controllers in Physically Embodied Robots

While modularity is thought to be central for the evolution of complexity and evolvability, it remains unclear how systems boot-strap themselves into modularity from random or fully integrated starting conditions. Clune et al. (2013) suggested that a positive correlation between sparsity and modularity is the prime cause of this transition. We sought to test the generality of this modularity-sparsity hypothesis by testing it for the first time in physically embodied robots. A population of ten Tadros — autonomous, surface-swimming robots propelled by a flapping tail — was used. Individuals varied only in the structure of their neural net control, a 2 x 6 x 2 network with recurrence in the hidden layer. Each of the 60 possible connections was coded in the genome, and could achieve one of three states: -1, 0, 1. Inputs were two light-dependent resistors and outputs were two motor control variables to the flapping tail, one for the frequency of the flapping and the other for the turning offset. Each Tadro was tested separately in a circular tank lit by a single overhead light source. Fitness was the amount of light gathered by a vertically oriented sensor that was disconnected from the controller net. Reproduction was asexual, with the top performer cloned and then all individuals entered into a roulette wheel selection process, with genomes mutated to create the offspring. The starting population of networks was randomly generated. Over ten generations, the population’s mean fitness increased two-fold. This evolution occurred in spite of an unintentional integer overflow problem in recurrent nodes in the hidden layer that caused outputs to oscillate. Our investigation of the oscillatory behavior showed that the mutual information of inputs and outputs was sufficient for the reactive behaviors observed. While we had predicted that both modularity and sparsity would follow the same trend as fitness, neither did so. Instead, selection gradients within each generation showed that selection directly targeted sparsity of the connections to the motor outputs. Modularity, while not directly targeted, was correlated with sparsity, and hence was an indirect target of selection, its evolution a “by-product” of its correlation with sparsity.

A Deterministic and Symbolic Regression Hybrid Applied to Resting-State fMRI Data

Symbolic regression (SR) is one the most popular applications of genetic programming (GP) and an attractive alternative to the standard deterministic regression approaches due to its flexibility in generating free-form mathematical models from observed data without any domain knowledge. However, GP suffers from various issues hindering the applicability of the technique to real-life problems. In this paper, we show that a hybrid deterministic regression (DR)/genetic programming based symbolic regression (GP-SR) algorithm outperforms GP-SR alone on a brain imaging dataset.

Understanding Climate-Vegetation Interactions in Global Rainforests Through a GP-Tree Analysis

The tropical rainforests are the largest reserves of terrestrial carbon and therefore, the future of these rainforests is a question that is of immense importance in the geoscience research community. With the recent severe Amazonian droughts in 2005 and 2010 and on-going drought in the Congo region for more than two decades, there is growing concern that these forests could succumb to precipitation reduction, causing extensive carbon release and feedback to the carbon cycle. However, there is no single ecosystem model that quantifies the relationship between vegetation health in these rainforests and climatic factors. Small scale studies have used statistical correlation measure and simple linear regression to model climate-vegetation interactions, but suffer from the lack of comprehensive data representation as well as simplistic assumptions about dependency of the target on the covariates. In this paper we use genetic programming (GP) based symbolic regression for discovering equations that govern the vegetation climate dynamics in the rainforests. Expecting micro-regions within the rainforests to have unique characteristics compared to the overall general characteristics, we use a modified regression-tree based hierarchical partitioning of the space to build individual models for each partition. The discovery of these equations reveal very interesting characteristics about the Amazon and the Congo rainforests. Our method GP-tree shows that the rainforests exhibit tremendous resiliency in the face of extreme climatic events by adapting to changing conditions.

Epigenetic Operators and the Evolution of Physically Embodied Robots

The genetic operators of recombination, mutation, and selection are commonly included in studies of evolution and evolvability, but they are not the only operators that can affect the genotype-to-phenotype (G→P) map and thus the outcomes of evolution. In this paper, we present experiments with an epigenetic operator (EO, for short), interactive wiring of a circuit, alongside common genetic operators, investigating both epigenetic and genetic operator effects on the evolution of both simulated and physically embodied Braitenberg-inspired robots. As a platform for our experiments, we built a system that encoded the genetics for the physical circuitry of the analog robots and made explicit rules for how that circuitry would be constructed; phenotypic expression consisted of the placement of wires to form the circuitry and thus govern robot behavior. We then varied the presence of gene interactions across populations of robots, studying how the EO — and its effects on G→P maps — affected the results of evolution over several generations. Additionally, a variant of these experiments was run in simulation to provide an independent test of the evolution impact of this EO. Our results demonstrate that robot populations with the EO had quantitatively different and potentially less adaptive evolution than populations without it. For example, selection increased the rate at which functional circuitry was lost in the population with the EO, compared to the population without it. In addition, in simulation EO populations were significantly less fit than populations without it. More generally, results such as these demonstrate the interaction of genetic and epigenetic operators during evolution, suggesting the broad importance of including EOs in investigations of evolvability. To our knowledge, our work represents the first physically embodied EO to be used in the evolution of physically embodied robots.

What we write about when we write about causality: features of causal statements across large-scale social discourse

Identifying and communicating relationships between causes and effects is important for understanding our world, but is affected by language structure, cognitive and emotional biases, and the properties of the communication medium. Despite the increasing importance of social media, much remains unknown about causal statements made online. To study real-world causal attribution, we extract a large-scale corpus of causal statements made on the Twitter social network platform as well as a comparable random control corpus. We compare causal and control statements using statistical language and sentiment analysis tools. We find that causal statements have a number of significant lexical and grammatical differences compared with controls and tend to be more negative in sentiment than controls. Causal statements made online tend to focus on news and current events, medicine and health, or interpersonal relationships, as shown by topic models. By quantifying the features and potential biases of causality communication, this study improves our understanding of the accuracy of information and opinions found online.

Automatic identification of hierarchy in multivariate data

Given n variables to model, symbolic regression (SR) returns a flat list of n equations. As the number of state variables to be modeled scales, it becomes increasingly difficult to interpret such a list. Here we present a symbolic regression method that detects and captures hidden hierarchy in a given system. The method returns the equations in a hierarchical dependency graph, which increases the interpretability of the results. We demonstrate two variations of this hierarchical modeling approach, and show that both consistently outperform non-hierarchical symbolic regression on a number of synthetic data sets.