Faecal pellets in deep marine soft clay crusts: Implications for hot-oil pipeline design

In recent years, the presence of crusts within near surface sediments found in deep water locations off the west coast of Angola has been of interest to hot-oil pipeline designers. The origin for these crusts is considered to be of biological origin, based on the observation of thousands of faecal pellets in natural crust core samples. This paper presents the results of laboratory tests undertaken on natural and faecal pellet-only samples. These tests investigate the role faecal pellets play in modifying the gemechanical behaviour of clayey sediments. It is found that faecal pellets are able to significantly alter both the strength and the average grain-size of natural sediments, and therefore, influence the permeability and stiffness. Hot-oil pipelines self-embed into and subsequent shear on crusts containing faecal pellets. Being able to predict the time required for installed pipelines to consolidate the underlying sediment and thus, how soon after pipe-laying, the interface strength will develop is of great interest to pipeline designers. It is concluded from wet-sieving samples before and after oedometer tests, that the process of pipe laying is unlikely to destroy pellets. They will therefore, be a major constituent of the sediment subject to soil-pipeline shearing behaviour during axial pipe-walking and lateral buckling. Based on the presented results, a discussion highlighting the key implications for pipeline design is therefore provided. Copyright © 2011 by ASME.

Gaussian processes for fast policy optimisation of POMDP-based dialogue managers

Modelling dialogue as a Partially Observable Markov Decision Process (POMDP) enables a dialogue policy robust to speech understanding errors to be learnt. However, a major challenge in POMDP policy learning is to maintain tractability, so the use of approximation is inevitable. We propose applying Gaussian Processes in Reinforcement learning of optimal POMDP dialogue policies, in order (1) to make the learning process faster and (2) to obtain an estimate of the uncertainty of the approximation. We first demonstrate the idea on a simple voice mail dialogue task and then apply this method to a real-world tourist information dialogue task. © 2010 Association for Computational Linguistics.

On-line policy optimisation of spoken dialogue systems via live interaction with human subjects

Statistical dialogue models have required a large number of dialogues to optimise the dialogue policy, relying on the use of a simulated user. This results in a mismatch between training and live conditions, and significant development costs for the simulator thereby mitigating many of the claimed benefits of such models. Recent work on Gaussian process reinforcement learning, has shown that learning can be substantially accelerated. This paper reports on an experiment to learn a policy for a real-world task directly from human interaction using rewards provided by users. It shows that a usable policy can be learnt in just a few hundred dialogues without needing a user simulator and, using a learning strategy that reduces the risk of taking bad actions. The paper also investigates adaptation behaviour when the system continues learning for several thousand dialogues and highlights the need for robustness to noisy rewards. © 2011 IEEE.