Extending boosting for large scale spoken language understanding

We propose three methods for extending the Boosting family of classifiers motivated by the real-life problems we have encountered. First, we propose a semisupervised learning method for exploiting the unlabeled data in Boosting. We then present a novel classification model adaptation method. The goal of adaptation is optimizing an existing model for a new target application, which is similar to the previous one but may have different classes or class distributions. Finally, we present an efficient and effective cost-sensitive classification method that extends Boosting to allow for weighted classes. We evaluated these methods for call classification in the AT&;T VoiceTone^® spoken language understanding system. Our results indicate that it is possible to obtain the same classification performance by using 30% less labeled data when the unlabeled data is utilized through semisupervised learning. Using model adaptation we can achieve the same classification accuracy using less than half of the labeled data from the new application. Finally, we present significant improvements in the “important” (i.e., higher weighted) classes without a significant loss in overall performance using the proposed cost-sensitive classification method.     

MOBSY: Integration of vision and dialogue in service robots

Abstract. This contribution introduces MOBSY, a fully integrated, autonomous mobile service robot system. It acts as an automatic dialogue-based receptionist for visitors to our institute. MOBSY incorporates many techniques from different research areas into one working stand-alone system. The techniques involved range from computer vision over speech understanding to classical robotics. Along with the two main aspects of vision and speech, we also focus on the integration aspect, both on the methodological and on the technical level. We describe the task and the techniques involved. Finally, we discuss the experiences that we gained with MOBSY during a live performance at our institute.     

Protoplast from <Emphasis Type="Italic">β</Emphasis>-carotene-producing fungus <Emphasis Type="Italic">Blakeslea trispora</Emphasis>: Preparation,regeneration and validation

The effects of key parameters on the preparation and regeneration of protoplast from the β-carotene-producing fungus Blakeslea trispora were discussed in this paper, including the combination of various enzymes, mycelial age, digesting time and temperature, pH value, osmotic stabilizers, pretreatment, culture medium and culture method. Under the condition of mixed enzymes in osmotic stabilizer (0.6 M NaCl) combined with 2% lysozyme, 3% cellulase and 3% snailase, the highest protoplast yield, as high as 7.48×106 protoplasts/mL, was obtained when mycelial age was 60 h at pH 5.0–6.0 with digesting for 14–16 h at 28 °C. After purification of the obtained protoplasts, they were regenerated in PDA regenerative medium using bilayer plate culture method. To validate the usability of the protoplasts, a novel plasmid with green fluorescent protein (GFP) was used in transformation for easy visual observation. The results showed that the protoplasts prepared by the optimized method were active and applicable in further gene manipulation experiments. This work was presented at 13 ^th YABEC symposium held at Seoul, Korea, October 20–22, 2007.     

Onset of buoyancy-driven convection in melting from below

When a horizontal homogeneous solid is melted from below, convection can be induced in a thermally unstable melt layer. In this study the onset of buoyancy-driven convection during time-dependent melting is investigated by using similarly transformed disturbance equations. The critical Rayleigh numbers based on the melt-layer thickness are found numerically for various conditions. For small superheats, the present predictions approach the well known results of classical Rayleigh-Bénard problems, that is, critical Rayleigh numbers are located between 1,296 and 1,708, regardless of the Prandtl number. However, for high superheats the critical Rayleigh number increases with an increase in phase change rate but with decrease in Prandtl number.     

Microstructure and fatigue properties of hydroformed aluminum alloys 6063 and 5754

This study investigated the microstructure and fatigue properties of hydroformed sections of the 5754 and 6063 aluminum alloys. The second-phase particles in 6063-T7 are identified as a mixture of Al₁₂Fe₃Si and Al₉Fe₂Si₂, with a slightly higher fraction of the former. The constituent particles in the 5754 alloy are Al₄Mn-type hexagonal compounds, where Mn is partially substituted by various other elements, resulting in Al₄(Fe,Mn,Si,Cr). The results show that despite its lower yield strength, the hydroformed 5754 alloy has higher ultimate tensile strength, ductility, and, more importantly, higher fatigue resistance than the 6063 material. Both crystallographic stage I and noncrystallographic stage II cracking are found in the 6063-T7 samples, but only stage II cracking is observed in the 5754 alloy. This implies that the low fatigue strength of 6063-T7 is related to its relatively large grain size, resulting in rapid stage I crack propagation. The higher fatigue lives of the 5754 alloy compared to the 6063 alloy in both the low- and high-cycle life regimes are due to the increased fatigue-crack-initiation and propagation resistance of the 5754 alloy and its probable cyclic strain-hardening behavior.     

To treat or not to treat: comparison of different criteria used to determine whether weight loss is to be recommended

Background  

Excess body fat is a major risk factor for disease primarily due to its endocrine activity. In recent years several criteria have been introduced to evaluate this factor. Nevertheless, treatment need is currently assessed only on the basis of an individual's Body Mass Index (BMI), calculated as body weight (in kg) divided by height in m². The aim of our study was to determine whether application of the BMI, compared to adiposity-based criteria, results in underestimation of the number of subjects needing lifestyle intervention.     

The relationship between bioelectrical impedance phase angle and subjective global assessment in advanced colorectal cancer

Background  

Bioelectrical Impedance (BIA) derived phase angle is increasingly being used as an objective indicator of nutritional status in advanced cancer. Subjective Global Assessment (SGA) is a subjective method of nutritional status. The objective of this study was to investigate the association between BIA derived phase angle and SGA in advanced colorectal cancer.     

Optimization of the magnetic circuit in the MFL inspection system for storage-tank floors

Magnetization is the key to inspection of a tank floor via the magnetic-flux-leakage (MFL) technique. In order to optimize the magnetic circuit of the MFL detector and obtain the best detection effects, the influences of the magnet size on the floor magnetization condition, the gap magnetic flux density, and the magnetic force were studied with the help of the finite element method (FEM) and the effects of some other parameters, such as the magnet pole spacing and pole-piece thickness, on the signal-to-noise ratio were analyzed. The simulation results indicate that variation of the magnet width affects the magnetization much more than variation of the magnet thickness and that the detector can reach a trade-off between the magnetization effects and the driving force when the magnet is about 30 mm thick and 40 mm wide. On condition that the floor has reached its magnetizing saturation, an increase in the magnet-pole spacing and the pole-piece thickness can improve the testing sensitivity and the signal-to-noise ratio. The text was submitted by the authors in English.     

Joint Measurements on Qubits and Cloning of Observables

Cloning of observables, unlike standard cloning of states, aims at copying the information encoded in the statistics of a class of observables rather then on quantum states themselves. In such a process the emphasis is on the quantum operation (evolution plus measurement) necessary to retrieve the original information. We analyze, for qubit systems, the cloning of a class generated by two noncommuting observables, elucidating the relationship between such a process and joint measurements. This helps in establishing an optimality criterion for cloning of observables. We see that, even if the cloning machine is designed to act on the whole class generated by two noncommuting observables, the same optimal performances of a joint measurement can be attained. Finally, the connection with state dependent cloning is enlightened.