oliver.obst.eu » paper

Relating Fisher information to order parameters

oliver — Fri, 11 Nov 2011 05:22:46 +0000

In our new paper (Relating Fisher information to order parameters. Physical Review E, 84(4):041116, 2011), we study phase transitions and relevant order parameters via statistical estimation theory using the Fisher information matrix. The assumptions that we make limit our analysis to order parameters representable as a negative derivative of thermodynamic potential over some thermodynamic variable. Nevertheless, the resulting representation is sufficiently general and explicitly relates elements of the Fisher information matrix to the rate of change in the corresponding order parameters. The obtained relationships allow us to identify, in particular, second-order phase transitions via divergences of individual elements of the Fisher information matrix. A computational study of random Boolean networks (RBNs) supports the derived relationships, illustrating that Fisher information of the magnetization bias (that is, activity level) is peaked in finite-size networks at the critical points, and the maxima increase with the network size. The framework presented here reveals the basic thermodynamic reasons behind similar empirical observations reported previously. The study highlights the generality of Fisher information as a measure that can be applied to a broad range of systems, particularly those where the determination of order parameters is cumbersome.

Phase transitions in least-effort communications

oliver — Tue, 16 Nov 2010 05:31:13 +0000

Our new paper Phase transitions in least-effort communications has been published in J. Stat. Mech. In this work, we critically examine a model that attempts to explain the emergence of power laws (e.g., Zipf’s law) in human language. The model is based on the principle of least effort in communications—specifically, the overall effort is balanced between the speaker effort and listener effort, with some trade-off. It has been shown that an information-theoretic interpretation of this principle is sufficiently rich to explain the emergence of Zipf’s law in the vicinity of the transition between referentially useless systems (one signal for all referable objects) and indexical reference systems (one signal per object). The phase transition is defined in the space of communication accuracy (information content) expressed in terms of the trade-off parameter. Our study explicitly solves the continuous optimization problem, subsuming a recent, more specific result obtained within a discrete space. The obtained results contrast Zipf’s law found by heuristic search (that attained only local minima) in the vicinity of the transition between referentially useless systems and indexical reference systems, with an inverse-factorial (sub-logarithmic) law found at the transition that corresponds to global minima. The inverse-factorial law is observed to be the most representative frequency distribution among optimal solutions.

Improving Recurrent Neural Network Performance using Transfer Entropy

oliver — Wed, 15 Sep 2010 13:04:52 +0000

Our new paper, Improving Recurrent Neural Network Performance using Transfer Entropy, has been accepted at the 17th International Conference on Neural Information Processing (ICONIP) in Sydney.

In this paper, we present an approach to improve the hidden layer of recurrent neural networks, guided by the learning goal of the system, and apply this new method to reservoir computing approaches. Reservoir computing uses, in general, a fixed, randomly initialised hidden layer. A consequence of this is that performance is usually quite good, but not necessarily optimal for the task at hand. There exist self-organised approaches – like intrinsic plasticity – that are able to improve performance of reservoir computing approaches, but usually, they just consider the input to the system, and don’t take the actual task of the system into account.

Our reservoir adaptation optimises the information transfer at each individual unit, dependent on properties of the information transfer between input and (desired) output of the system. Using synthetic data, we show that this reservoir adaptation improves the performance of offline echo state learning and Recursive Least Squares Online Learning.

ICONIP 2010 takes place from the 22nd–25th November 2010 in Sydney, Australia.

Initialization and self-organized optimization of recurrent neural network connectivity

oliver — Fri, 13 Nov 2009 12:18:54 +0000

Our new paper describes a mathematical model for generic neural microcircuits, with potential engineering applications, as well as implications to understand how networks in biology are shaped to be optimally adapted to requirements of their environment.

Reservoir computing (RC) is a recent paradigm in the field of recurrent neural networks. Networks in RC have a sparsely and randomly connected fixed hidden layer, and only output connections are trained. RC networks have recently received increased attention as a mathematical model for generic neural microcircuits to investigate and explain computations in neocortical columns. Applied to specific tasks, their fixed random connectivity, however, leads to significant variation in performance. Few problem-specific optimization procedures are known, which would be important for engineering applications, but also in order to understand how networks in biology are shaped to be optimally adapted to requirements of their environment. We study a general network initialization method using permutation matrices and derive a new unsupervised learning rule based on intrinsic plasticity (IP). The IP-based learning uses only local learning, and its aim is to improve network performance in a self-organized way. Using three different benchmarks, we show that networks with permutation matrices for the reservoir connectivity have much more persistent memory than the other methods but are also able to perform highly nonlinear mappings. We also show that IP-based on sigmoid transfer functions is limited concerning the output distributions that can be achieved.

neurophysiology, optimisation, physiological models, recurrent neural nets, unsupervised learning, reservoir computing, echo state networks

Origins of Scaling in Genetic Code

oliver — Mon, 21 Sep 2009 12:02:56 +0000

The principle of least effort in communications has been shown, by Ferrer i Cancho and Solé, to explain emergence of power laws (e.g., Zipf’s law) in human languages. In our new paper, Origins of Scaling in Genetic Code (O. Obst, D. Polani, M. Prokopenko), published on ECAL 2009, we apply the principle and the information-theoretic model of Ferrer i Cancho and Solé to genetic coding. The application of the principle is achieved via equating the ambiguity of signals used by “speakers” with codon usage, on the one hand, and the effort of “hearers” with needs of amino acid translation mechanics, on the other hand. The re-interpreted model captures the case of the typical (vertical) gene transfer, and confirms that Zipf’s law can be found in the transition between referentially useless systems (i.e., ambiguous genetic coding) and indexical reference systems (i.e., zero-redundancy genetic coding). As with linguistic symbols, arranging genetic codes according to Zipf’s law is observed to be the optimal solution for maximising the referential power under the effort constraints. Thus, the model identifies the origins of scaling in genetic coding — via a trade-off between codon usage and needs of amino acid translation. Furthermore, the paper extends Ferrer i Cancho Solé model to multiple inputs, reaching out toward the case of horizontal gene transfer (HGT) where multiple contributors may share the same genetic coding. Importantly, the extended model also leads to a sharp transition between referentially useless systems (ambiguous HGT) and indexical reference systems (zero-redundancy HGT). Zipf’s law is also observed to be the optimal solution in the HGT case.

genetic code, evolution, language, zipf’s law, horizontal gene transfer

Inverse Steering Behaviors

oliver — Wed, 20 May 2009 12:41:31 +0000

I was excited to find one of my approaches being used in a commercial product for emergency egress simulation, sold by a company in the US: Back in 2006, Heni, Jan and I published an approach we called Inverse Steering Behaviors in the chapter “Fast, Neat, and Under Control: Arbitrating Between Steering Behaviors” of AI Game Programming Wisdom 3. The technique builds on Steering Behaviors by Craig Reynolds – reactive procedures for physical agents (like robots or simulated creatures) to move in a lifelike way within dynamic environments. Developed in the late 80s, steering behaviors found applications for example in movies like Lord of the Rings. Our Inverse Steering Behaviors improve the arbitration between individual behaviors, which results in less collisions. Back when we did the work, we used the approach in our robotic soccer team for navigation and to dribble around opponents. The agent-based emergency evacuation simulation system sold by Thunderhead Engineering, is called Pathfinder.
There’s a free 30-day trial available, and also short video.

Watch the video.

steering behaviors, inverse steering, robotic soccer, emergency egress simulation

Studies on Reservoir Initialization and Dynamics Shaping in Echo State Networks

oliver — Thu, 19 Feb 2009 12:49:54 +0000

In a paper that was recently accepted at the European Symposium on Artificial Neural Networks (ESANN 2009), we look at different ways to influence the performance of echo state networks. Traditionally, echo state networks and other reservoir computing approaches use a fixed random connected reservoir, which leads to significant variation in performance. Only few problem specific optimisation procedures are known to date. We study a general initialization method using permutation matrices and derive a new unsupervised learning rule based on intrinsic plasticity (IP) for echo state networks. Using three different benchmarks, we show that networks with permutation matrices for the reservoir connectivity have much longer memory than the other methods, but are also able to perform highly non-linear mappings. We also show that IP based on sigmoid transfer functions is limited concerning the output distributions that can be achieved.

Studies on Reservoir Initialization and Dynamics Shaping in Echo State Networks,
J. Boedecker, O. Obst, N.M. Mayer, M. Asada. The full paper will be available after the conference (April) is now available.

neural networks, intrinsic plasticity, echo state networks, reservoir computing

Computers in Sport

oliver — Wed, 18 Jun 2008 23:01:18 +0000

Finally, the book Computers in Sport (edited by P. Dabnicki and A. Baca) appeared. In this book, my colleagues and I have a chapter “Approaching a Formal Soccer Theory from the Behavior Specification in Robotic Soccer“, where we discuss a top-down approach to modelling soccer knowledge, as it can be found in soccer theory books. The goal is to model soccer strategies and tactics in a way that they are usable for multiple robotic soccer leagues in the RoboCup. We investigate if and how soccer theory can be formalized such that specification and execution are possible. The advantage is clear: theory abstracts from hardware and from specific situations in different leagues. We introduce basic primitives compliant with the terminology known in soccer theory, discuss an example on an abstract level and formalize it. The formalization of soccer presented here is appealing. It goes beyond the behaviour specification of soccer playing robots. For sports science a unified formal soccer theory might help to better understand and to formulate basic concepts in soccer. The possibility of the formalization to develop computer programs, which allow to simulate and to reason about soccer moves, might also take sports science a step further.
Get your copy of the book at your local book shop or at Amazon .

Technorati Tags: book, sport, computers, robocup, robots, theory

New Paper on Echo State Networks

oliver — Fri, 18 Apr 2008 08:16:28 +0000

At IPSN 2008, I’m going to present our work “Using Echo State Networks for Anomaly Detection in Underground Coal Mines”. In this work, we investigate the problem of identifying anomalies in monitoring critical gas concentrations using a sensor network in an underground coal mine. In this domain, one of the main problems is a provision of mine specific anomaly detection, with cyclical (moving) instead of flatline (static) alarm threshold levels. An additional practical difficulty in modelling a specific mine is the lack of fully labelled data of normal and abnormal situations. We present an approach addressing these difficulties based on echo state networks learning mine specific anomalies when only normal data is available. Echo state networks utilize incremental updates driven by new sensor readings, thus enabling a detection of anomalies at any time during the sensor network operation. We evaluate this approach against a benchmark — Bayes Network based anomaly detection, and observe that the quality of the overall predictions is comparable to the benchmark. However, the echo state networks maintain the same level of predictive accuracy for data from multiple sources. Therefore, the ability of echo state networks to model dynamical systems make this approach more suitable for anomaly detection and predictions in sensor networks. Check out the details here.

Technorati Tags: Sensor networks, Echo State Networks, Neural Networks, Anomaly detection, IPSN, 2008, Paper, Computer Science

Paper: Spatiotemporal Anomaly Detection in Gas Monitoring Sensor Networks

oliver — Thu, 31 Jan 2008 06:56:38 +0000

Our paper “Spatiotemporal Anomaly Detection in Gas Monitoring Sensor Networks” is currently being presented at the European conference on Wireless Sensor Networks (EWSN’08) in Bologna, Italy. In this paper, we use Bayesian Networks as a means for unsupervised learning and anomaly (event) detection in gas monitoring sensor networks for underground coal mines. We show that the Bayesian Network model can learn cyclical baselines for gas concentrations, and by this reduce false alarms usually caused by ﬂatline thresholds. You can check out the details here.

Technorati Tags: paper, sensor networks, bayesian networks, anomaly detection, coal mines