Bayesian Learning of Noisy Markov Decision Processes

We consider the inverse reinforcement learning problem, that is, the problem of learning from, and then predicting or mimicking a controller based on state/action data. We propose a statistical model for such data, derived from the structure of a Markov decision process. Adopting a Bayesian approach to inference, we show how latent variables of the model can be estimated, and how predictions about actions can be made, in a unified framework. A new Markov chain Monte Carlo (MCMC) sampler is devised for simulation from the posterior distribution. This step includes a parameter expansion step, which is shown to be essential for good convergence properties of the MCMC sampler. As an illustration, the method is applied to learning a human controller.

Inclusive design and the bottom line: How can its value be proven to decision makers?

Designing technology products that embrace the needs and capabilities of heterogeneous users leads not only to increased customer satisfaction and enhanced corporate social responsibility, but also better market penetration. Yet, achieving inclusion in today's pressured and fast-moving markets is not straight-forward. For a time, inaccessible and unusable design was solely seen as the fault of designers and a whole line of research was dedicated to pinpointing their frailties. More recently, it has become progressively more recognised that it is not necessarily designers' lack of awareness, or unwillingness, that results in sub-optimal design, but rather there are multi-faceted organisational factors at play that seldom provide an adequate environment in which inclusive products could be designed. Through literature review, a detailed audit of inclusivity practice in a large global company and ongoing research regarding quantification of cost-effectiveness of inclusive design, this paper discusses the overarching operational problems that prevent organisations from developing optimally inclusive products and offers best-practice principles for the future. © 2013 Springer-Verlag Berlin Heidelberg.

Interactive multi-objective particle swarm optimisation using decision space interaction

The most common approach to decision making in multi-objective optimisation with metaheuristics is a posteriori preference articulation. Increased model complexity and a gradual increase of optimisation problems with three or more objectives have revived an interest in progressively interactive decision making, where a human decision maker interacts with the algorithm at regular intervals. This paper presents an interactive approach to multi-objective particle swarm optimisation (MOPSO) using a novel technique to preference articulation based on decision space interaction and visual preference articulation. The approach is tested on a 2D aerofoil design case study and comparisons are drawn to non-interactive MOPSO. © 2013 IEEE.

Decision making at the front end of innovation: The hidden influence of knowledge and decision criteria

Decision making at the front end of innovation is critical for the success of companies. This paper presents a method, called decision making based on knowledge (DeBK), which was created to analyze the decision-making process at the front end. The method evaluates the knowledge of project information and the importance of decision criteria, compiling a measure that indicates whether decisions are founded on available knowledge and what criteria are in fact being considered to delineate them. The potential contribution of DeBK is corroborated through two projects that faced decision-making issues at the front end of innovation. © 2014 RADMA and John Wiley & Sons Ltd.

Motor effort alters changes of mind in sensorimotor decision making.

After committing to an action, a decision-maker can change their mind to revise the action. Such changes of mind can even occur when the stream of information that led to the action is curtailed at movement onset. This is explained by the time delays in sensory processing and motor planning which lead to a component at the end of the sensory stream that can only be processed after initiation. Such post-initiation processing can explain the pattern of changes of mind by asserting an accumulation of additional evidence to a criterion level, termed change-of-mind bound. Here we test the hypothesis that physical effort associated with the movement required to change one's mind affects the level of the change-of-mind bound and the time for post-initiation deliberation. We varied the effort required to change from one choice target to another in a reaching movement by varying the geometry of the choice targets or by applying a force field between the targets. We show that there is a reduction in the frequency of change of mind when the separation of the choice targets would require a larger excursion of the hand from the initial to the opposite choice. The reduction is best explained by an increase in the evidence required for changes of mind and a reduced time period of integration after the initial decision. Thus the criteria to revise an initial choice is sensitive to energetic costs.

Delayed-decision making in noisy bistable switches

Bistable switches are frequently encountered in biological systems. Typically, a bistable switch models a binary decision where each decision corresponds to one of the two stable equilibria. Recently, we showed that the global decision-making process in bistable switches strongly depends on a particular equilibrium point of these systems, their saddle point. In particular, we showed that a saddle point with a time-scale separation between its attractive and repulsive directions can delay the decision-making process. In this paper, we study the effects of white Gaussian noise on this mechanism of delayed decision-making induced by the saddle point. Results show that the mean decision-time strongly depends on the balance between the initial distance to the separatrix and the noise strength. © IFAC.

A derivation of the source-channel error exponent using nonidentical product distributions

This paper studies the random-coding exponent of joint source-channel coding for a scheme where source messages are assigned to disjoint subsets (referred to as classes), and codewords are independently generated according to a distribution that depends on the class index of the source message. For discrete memoryless systems, two optimally chosen classes and product distributions are found to be sufficient to attain the sphere-packing exponent in those cases where it is tight. © 2014 IEEE.

Compact Reconfigurable Binary-Decision-Diagram Logic Circuit on a GaAs Nanowire Network

We describe a reconfigurable binary-decision-diagram logic circuit based on Shannon's expansion of Boolean logic function and its graphical representation on a semiconductor nanowire network. The circuit is reconfigured by using programmable switches that electrically connect and disconnect a small number of branches. This circuit has a compact structure with a small number of devices compared with the conventional look-up table architecture. A variable Boolean logic circuit was fabricated on an etched GaAs nanowire network having hexagonal topology with Schottky wrap gates and SiN-based programmable switches, and its correct logic operation together with dynamic reconfiguration was demonstrated.

Correcting the systematic error of the density functional theory calculation: the alternate combination approach of genetic algorithm and neural network

The alternate combinational approach of genetic algorithm and neural network (AGANN) has been presented to correct the systematic error of the density functional theory (DFT) calculation. It treats the DFT as a black box and models the error through external statistical information. As a demonstration, the AGANN method has been applied in the correction of the lattice energies from the DFT calculation for 72 metal halides and hydrides. Through the AGANN correction, the mean absolute value of the relative errors of the calculated lattice energies to the experimental values decreases from 4.93% to 1.20% in the testing set. For comparison, the neural network approach reduces the mean value to 2.56%. And for the common combinational approach of genetic algorithm and neural network, the value drops to 2.15%. The multiple linear regression method almost has no correction effect here.