Risto Miikkulainen

Evolving neural networks through augmenting topologies

An important question in neuroevolution is how to gain an advantage from evolving neural network topologies along with weights. We present a method, NeuroEvolution of Augmenting Topologies (NEAT), which outperforms the best fixed-topology method on a challenging benchmark reinforcement learning task. We claim that the increased efficiency is due to (1) employing a principled method of crossover of different topologies, (2) protecting structural innovation using speciation, and (3) incrementally growing from minimal structure. We test this claim through a series of ablation studies that demonstrate that each component is necessary to the system as a whole and to each other. What results is signicantly faster learning. NEAT is also an important contribution to GAs because it shows how it is possible for evolution to both optimize and complexify solutions simultaneously, offering the possibility of evolving increasingly complex solutions over generations, and strengthening the analogy with biological evolution.

Intensive training induces longitudinal changes in meditation state related EEG oscillatory activity

The capacity to focus ones attention for an extended period of time can be increased through training in contemplative practices. However, the cognitive processes engaged during meditation that support trait changes in cognition are not well characterized. We conducted a longitudinal wait-list controlled study of intensive meditation training. Retreat participants practiced focused attention (FA) meditation techniques for three months during an initial retreat. Wait-list participants later undertook formally identical training during a second retreat. Dense-array scalp-recorded electroencephalogram (EEG) data were collected during 6 min of mindfulness of breathing meditation at three assessment points during each retreat. Second-order blind source separation, along with a novel semi-automatic artifact removal tool (SMART), was used for data preprocessing. We observed replicable reductions in meditative state-related beta-band power bilaterally over anteriocentral and posterior scalp regions. In addition, individual alpha frequency (IAF) decreased across both retreats and in direct relation to the amount of meditative practice. These findings provide evidence for replicable longitudinal changes in brain oscillatory activity during meditation and increase our understanding of the cortical processes engaged during meditation that may support long-term improvements in cognition.

Competitive coevolution through evolutionary complexification

Two major goals in machine learning are the discovery and improvement of solutions to complex problems. In this paper, we argue that complexification, i.e. the incremental elaboration of solutions through adding new structure, achieves both these goals. We demonstrate the power of complexification through the NeuroEvolution of Augmenting Topologies (NEAT) method, which evolves increasingly complex neural network architectures. NEAT is applied to an open-ended coevolutionary robot duel domain where robot controllers compete head to head. Because the robot duel domain supports a wide range of strategies, and because coevolution benefits from an escalating arms race, it serves as a suitable testbed for studying complexification. When compared to the evolution of networks with fixed structure, complexifying evolution discovers significantly more sophisticated strategies. The results suggest that in order to discover and improve complex solutions, evolution, and search in general, should be allowed to complexify as well as optimize.

A Taxonomy for artificial embryogeny

A major challenge for evolutionary computation is to evolve phenotypes such as neural networks, sensory systems, or motor controllers at the same level of complexity as found in biological organisms. In order to meet this challenge, many researchers are proposing indirect encodings, that is, evolutionary mechanisms where the same genes are used multiple times in the process of building a phenotype. Such gene reuse allows compact representations of very complex phenotypes. Development is a natural choice for implementing indirect encodings, if only because nature itself uses this very process. Motivated by the development of embryos in nature, we define artificial embryogeny (AE) as the subdiscipline of evolutionary computation (EC) in which phenotypes undergo a developmental phase. An increasing number of AE systems are currently being developed, and a need has arisen for a principled approach to comparing and contrasting, and ultimately building, such systems. Thus, in this paper, we develop a principled taxonomy for AE. This taxonomy provides a unified context for long-term research in AE, so that implementation decisions can be compared and contrasted along known dimensions in the design space of embryogenic systems. It also allows predicting how the settings of various AE parameters affect the capacity to efficiently evolve complex phenotypes.

Forming neural networks through efficient and adaptive coevolution

This article demonstrates the advantages of a cooperative, coevolutionary search in difficult control problems. The symbiotic adaptive neuroevolution (SANE) system coevolves a population of neurons that cooperate to form a functioning neural network. In this process, neurons assume different but overlapping roles, resulting in a robust encoding of control behavior. SANE is shown to be more efficient and more adaptive and to maintain higher levels of diversity than the more common network-based population approaches. Further empirical studies illustrate the emergent neuron specializations and the different roles the neurons assume in the population.

Computational Maps in the Visual Cortex

Biological Background.- Computational Foundations.- LISSOM: A Computational Map Model of V1.- Development of Maps and Connections.- Understanding Plasticity.- Understanding Visual Performance: The Tilt Aftereffect.- HLISSOM: A Hierarchical Model.- Understanding Low-Level Development: Orientation Maps.- Understanding High-Level Development: Face Detection.- PGLISSOM: A Perceptual Grouping Model.- Temporal Coding.- Understanding Perceptual Grouping: Contour Integration.- Computations in Visual Maps.- Scaling LISSOM simulations.- Discussion: Biological Assumptions and Predictions.- Future Work: Computational Directions.- Conclusion.

Natural Language Processing With Modular Pdp Networks and Distributed Lexicon

An approach to connectionist natural language processing is proposed, which is based on hierarchically organized modular Parallel Distributed Processing (PDP) networks and a central lexicon of distributed input/output representations. The modules communicate using these representations, which are global and publicly available in the system. The representations are developed automatically by all networks while they are learning their processing tasks. The resulting representations re ect the regularities in the subtasks, which facilitates robust processing in the face of noise and damage, supports improved generalization, and provides expectations about possible contexts. The lexicon can be extended by cloning new instances of the items, that is, by generating a number of items with known processing properties and distinct identities. This technique combinatorially increases the processing power of the system. The recurrent FGREP module, together with a central lexicon, is used as a basic building block in modeling higher level natural language tasks. A single module is used to form case-role representations of sentences from word-by-word sequential natural language input. A hierarchical organization of four recurrent FGREP modules (the DISPAR system) is trained to produce fully expanded paraphrases of script-based stories, where unmentioned events and role llers are inferred.

Incremental grid growing: encoding high-dimensional structure into a two-dimensional feature map

Ordinary feature maps with fully connected, fixed grid topology cannot properly reflect the structure of clusters in the input space. Incremental feature map algorithms, where nodes and connections are added to or deleted from the map according to the input distribution can overcome this problem. Such algorithms have been limited to maps that can be drawn in 2-D only in the case of two-dimensional input space. In the proposed approach, nodes are added incrementally to a regular two-dimensional grid, which is drawable at all times, irrespective of the dimensionality of the input space. The process results in a map that explicitly represents the cluster structure of the high-dimensional input.<<ETX>>

Topographic receptive fields and patterned lateral interaction in a self-organizing model of the primary visual cortex

This article presents a self-organizing neural network model for the simultaneous and cooperative development of topographic receptive fields and lateral interactions in cortical maps. Both afferent and lateral connections adapt by the same Hebbian mechanism in a purely local and unsupervised learning process. Afferent input weights of each neuron self organize into hill-shaped profiles, receptive fields organize topographically across the network, and unique lateral interaction profiles develop for each neuron. The model demonstrates how patterned lateral connections develop based on correlated activity and explains why lateral connection patterns closely follow receptive field properties such as ocular dominance.

Script recognition with hierarchical feature maps

The hierarchical feature map system recognizes an input story as an instance of a particular script by classifying it at three levels: scripts, tracks and role bindings. The recognition taxonomy, i.e. the breakdown of each script into the tracks and roles, is extracted automatically and independently for each script from examples of script instantiations in an unsupervised self-organizing process. The process resembles human learning in that the differentiation of the most frequently encountered scripts become gradually the most detailed. The resulting structure is a hierarchical pyramid of feature maps. The hierarchy visualizes the taxonomy and the maps lay out the topology of each level. The number of input lines and the self-organization time are considerably reduced compared to the ordinary single-level feature mapping. The system can recognize incomplete stories and recover the missing events. The taxonomy also serves as memory organization for script-based episodic memory. The maps assign a unique memory location for each script instantiation. The most salient parts of the input data are separated and most resources are concentrated on representing them accurately.