Effective handling of dialogue state in the hidden information state POMDP-based dialogue manager

Effective dialogue management is critically dependent on the information that is encoded in the dialogue state. In order to deploy reinforcement learning for policy optimization, dialogue must be modeled as a Markov Decision Process. This requires that the dialogue statemust encode all relevent information obtained during the dialogue prior to that state. This can be achieved by combining the user goal, the dialogue history, and the last user action to form the dialogue state. In addition, to gain robustness to input errors, dialogue must be modeled as a Partially Observable Markov Decision Process (POMDP) and hence, a distribution over all possible states must be maintained at every dialogue turn. This poses a potential computational limitation since there can be a very large number of dialogue states. The Hidden Information State model provides a principled way of ensuring tractability in a POMDP-based dialogue model. The key feature of this model is the grouping of user goals into partitions that are dynamically built during the dialogue. In this article, we extend this model further to incorporate the notion of complements. This allows for a more complex user goal to be represented, and it enables an effective pruning technique to be implemented that preserves the overall system performance within a limited computational resource more effectively than existing approaches. © 2011 ACM.

Handling uncertainty in housing stock models

Housing stock models can be useful tools in helping to assess the environmental and socio-economic impacts of retrofits to residential buildings; however, existing housing stock models are not able to quantify the uncertainties that arise in the modelling process from various sources, thus limiting the role that they can play in helping decision makers. This paper examines the different sources of uncertainty involved in housing stock models and proposes a framework for handling these uncertainties. This framework involves integrating probabilistic sensitivity analysis with a Bayesian calibration process in order to quantify uncertain parameters more accurately. The proposed framework is tested on a case study building, and suggestions are made on how to expand the framework for retrofit analysis at an urban-scale. © 2011 Elsevier Ltd.

Improved handling and preservation of freshwater giant prawn, Macrobrachium rosenbergii for producing safe and wholesome products

Studies were undertaken to evaluate the quality changes in freshwater giant prawn, Macrobrachium rosenbergii during various storage conditions of handling and preservation and producing safe and quality products. The samples kept in ice immediately after catch with head-on and head-less condition were found to be acceptable for 6 days and 7 days, respectively. Delaying of icing considerably shortened the shelf-life. The pH value increased from 6.36 to 8.0 after 10 days in ice. The initial average TVB-N value of sample increased from below 10 mg/100 g to 25 mg/100 g with the lapse of storage period. The Ca++ ATPase activity in presence of 0.1M KCl slightly decreased at the end of 10 days of ice storage. Immediately after harvest, initial aerobic plate count (APC) was 2.88x10^6 CFU/g which gradually increased to 1.12x10^8 CFU/g after 6 days in ice storage and showed early signs of spoilage. Initial bacterial genera in the prawn iced at 0 hours were comprised of Coryneform (22.21 %), Bacillus (7.40%), Micrococcus (11.11 %), Achromobacter (48.14%), Flavobacterium/Cytophaga (7.40%), Pseudomonas (3.70%) and Aeromonas (3.70%). During ice storage Coryneforms and Bacillus were always dominating along with less prominent ones - Micrococcus, Achromobacter and Flavobacterium. Studies were conducted on the stability of myofibrillar protein of M. rosenbergii under different storage and pH conditions. The influence of a wide range of pH on the remaining Ca++ ATPase activity of M. rosenbergii muscle myofibrils after storage at -20°C for 2 days, at 0°C for 2 days and at 35°C for 30 minutes demonstrated that ATPase activities were lower in acidic and alkaline pH regions and the activity remained relatively high. Mg++ ATPase activities both in presence and absence of Ca++ remained high at neutral pH compared to those of acidic and alkaline region. The solubility of myofibrillar protein decreased gradually both in acidic and alkaline pH regions. The study also examined the bacteriological quality of freshly harvested M. rosenbergii, pond sediment and pond water from four commercial freshwater prawn farms at Fulpur and Tarakanda upazilas in the district of Mymensingh. The study included aerobic plate count (APC), total coliform count, detection, isolation and identification of suspected public health hazard bacteria and their seasonal variation, salt tolerance test, antibiotic sensitivity test of the isolates and washing effect of chlorinated water on the bacterial load in the prawn samples. APC in sediment soil and water of the farm and gill and hepatopancreas of freshly harvested prawns varied considerably among the farms and between summer and winter season. The range of coliform count in water, gill and hepatopancreas ranged between 6 - 2.8x10^2 CFU/ml, 1.2x10^2 - 3.32x10^2 CFU/g and 1.43x10^2 - 3.89 x10^3 CFU/g, respectively. No coliform was detected in pond sediment sample. Suspected health hazard bacteria isolated and identified from pond sediment, water, gill and hepatopancreas included Streptococcus, Bacillus, Escherichia coli, Klebsialla, Salmonella, Staphylococcus, Pseudomonas and Aeromonas. Bacillus, Salmonella and Staphyloccus [sic], and were found to be highly salt tolerant and capable of growing at 10% NaCl. The antibiotic discs with different concentration of antibiotics were used for the sensitivity test. The organisms were found to be most sensitive against Tetracyclin and Gentamycin.

An assessment of handling and processing methods used for the shrimp fishery by-catch in Kalpitiya, Sri Lanka

The by-catch from the shrimp trawl fishery in Kalpitiya is mainly used for the production of dried fish, which provides an additional source of income for fishermen in the area. It has been observed that current handling practices along the value addition chain are responsible for the poor quality and low price of the end product. This study was aimed at identifying the shortcomings in such handling practices by fishermen and dried fish producers and assessing the quality of shrimp fishery by-catch along the processing chain in order to recommend more efficient utilization methods that will improve the quality of the end product. Fresh fish, dried fish and harbour water samples were tested for total coli forms, faecal coliforms, E. coli and Salmonella in order to assess their microbial quality: In addition, standard plate counts (SPC) of fish samples were also carried out. A survey was carried out from July-October 2006 at Kalpitiya, using a pre-tested questionnaire to collect information from individuals who have been engaged in dried fish processing. Average values obtained for freshly landed and dried fish respectively, were, SPC 9.88x10 super(5) CFU/g and 30.43x10 super(5) CFU/g, total coliforms 23.05 and 24.23 MPN/g and fecal coliforms 8.28 and 9.00 MPN/g. These values exceed the recommendations in the SL standards. A quarter of the landed fresh fish and 38% of dried fish from the producers were positive for E. coli and thus failed to show required end product quality. SPC of harbour water was 14.35x10 super(6) CFU/ml and all samples were found to be contaminated with E. coli. None of the fishermen and dried fish producers were satisfied with the quality of the end product. The reasons for poor quality as indicated by them were: limited availability of ice (75%), lack of infrastructure facilities (65%), uncertainty of markets (52%), lack of emphasis on quality (47%) and poor access to available technologies (41%). Respondents to the questionnaire also identified: unavailability of potable water, insulated boxes, good landing jetty, racks for drying fish, poor cold storage facilities and limitations in dried fish storage facilities, as further factors leading to the loss of quality in their products. Results demonstrate that improvements to the infrastructure facilities and conducting of proper awareness programmes on handling practices could lead for improvements in the quality of value added products prepared from the shrimp fishery by-catch at Kalpitiya.

Physiological responses to capture and handling of free-ranging male Yangtze finless porpoises (Neophocaena phocaenoides asiaeorientalis)

We captured free-ranging male Yangtze finless porpoises over three seasons and assayed leukocytes and serum biochemistry to investigate physiological responses to the capture and handlings. Serum thyroid hormones (THs) declined sharply in those porpoises compared with hormone variation in a captive male finless porpoise. Hypernatremia and hypokalemia were also significant in the free-ranging animals suggesting that conservation of serum sodium might be acutely vital for this freshwater subspecies. The animals captured in spring showed more significant neutrophilia and eosinopenia than those captured in autumn suggesting that they may be more affected by capture during the breeding season. Furthermore, physical examination of porpoises when out of the water was apparently stressful, particularly when they were kept out of the water for longer periods. However, an increase in circulating THs may be an adaptive response to accommodate these short-term stresses.

The effect of the product cost factor on error handling in industrial robots

N.W. Hardy and M.H. Lee. The effect of the product cost factor on error handling in industrial robots. In Maria Gini, editor, Detecting and Resolving Errors in Manufacturing Systems. Papers from the 1994 AAAI Spring Symposium Series, pages 59-64, Menlo Park, CA, March 1994. The AAAI Press. Technical Report SS-94-04, ISBN 0-929280-60-1.

Gestão construtiva de conflitos em contexto de saúde

Dissertação apresentada à Universidade Fernando Pessoa como parte dos requisitos para a obtenção do grau de Mestre em Psicologia, ramo de Psicologia Clínica e da Saúde

Learning to drive: from hazard detection to hazard handling

Hazard perception has been found to correlate with crash involvement, and has thus been suggested as the most likely source of any skill gap between novice and experienced drivers. The most commonly used method for measuring hazard perception is to evaluate the perception-reaction time to filmed traffic events. It can be argued that this method lacks ecological validity and may be of limited value in predicting the actions drivers’ will take to hazards encountered. The first two studies of this thesis compare novice and experienced drivers’ performance on a hazard detection test, requiring discrete button press responses, with their behaviour in a more dynamic driving environment, requiring hazard handling ability. Results indicate that the hazard handling test is more successful at identifying experience-related differences in response time to hazards. Hazard detection test scores were strongly related to performance on a driver theory test, implying that traditional hazard perception tests may be focusing more on declarative knowledge of driving than on the procedural knowledge required to successfully avoid hazards while driving. One in five Irish drivers crash within a year of passing their driving test. This suggests that the current driver training system does not fully prepare drivers for the dangers they will encounter. Thus, the third and fourth studies in this thesis focus on the development of two simulator-based training regimes. In the third study participants receive intensive training on the molar elements of driving i.e. speed and distance evaluation. The fourth study focuses on training higher order situation awareness skills, including perception, comprehension and projection. Results indicate significant improvement in aspects of speed, distance and situation awareness across training days. However, neither training programme leads to significant improvements in hazard handling performance, highlighting the difficulties of applying learning to situations not previously encountered.

The use of fault detection and diagnostics to reduce air handling unit energy consumption

The contribution of buildings towards total worldwide energy consumption in developed countries is between 20% and 40%. Heating Ventilation and Air Conditioning (HVAC), and more specifically Air Handling Units (AHUs) energy consumption accounts on average for 40% of a typical medical device manufacturing or pharmaceutical facility’s energy consumption. Studies have indicated that 20 – 30% energy savings are achievable by recommissioning HVAC systems, and more specifically AHU operations, to rectify faulty operation. Automated Fault Detection and Diagnosis (AFDD) is a process concerned with potentially partially or fully automating the commissioning process through the detection of faults. An expert system is a knowledge-based system, which employs Artificial Intelligence (AI) methods to replicate the knowledge of a human subject matter expert, in a particular field, such as engineering, medicine, finance and marketing, to name a few. This thesis details the research and development work undertaken in the development and testing of a new AFDD expert system for AHUs which can be installed in minimal set up time on a large cross section of AHU types in a building management system vendor neutral manner. Both simulated and extensive field testing was undertaken against a widely available and industry known expert set of rules known as the Air Handling Unit Performance Assessment Rules (APAR) (and a later more developed version known as APAR_extended) in order to prove its effectiveness. Specifically, in tests against a dataset of 52 simulated faults, this new AFDD expert system identified all 52 derived issues whereas the APAR ruleset identified just 10. In tests using actual field data from 5 operating AHUs in 4 manufacturing facilities, the newly developed AFDD expert system for AHUs was shown to identify four individual fault case categories that the APAR method did not, as well as showing improvements made in the area of fault diagnosis.

Effect of cytochalasin B on 3-O-[(14)C]-methyl-D-glucose or D-[U-(14)C]glucose handling by BRIN-BD11 cells.

The present study aimed to investigate the effects of cytochalasin B (20 μM) on the uptake of 3-O-[(14)C]-methyl-D-glucose or D-[U-(14)C]glucose (8.3 mM each) by BRIN-BD11 cells. Taking into account the distribution space of tritiated water ((3)HOH), which was unexpectedly increased shortly after exposure of the cells to cytochalasin B and then progressively returned to its control values, and that of L-[1-(14)C]glucose, used as an extracellular marker, it was demonstrated that cytochalasin B caused a modest, but significant inhibition of the uptake of D-glucose and its non-metabolized analog by the BRIN-BD11 cells. These findings resemble those observed in acinar or ductal cells of the rat submaxillary gland and displayed a relative magnitude comparable to that found for the inhibition of D-glucose metabolism by cytochalasin B in purified pancreatic islet B cells. These findings reinforce the view that the primary site of action of cytochalasin B is located at the level of the plasma membrane.