MiDTD: A Simple and Effective Distillation Framework for Distantly Supervised Relation Extraction

Relation extraction (RE), an important information extraction task, faced the great challenge brought by limited annotation data. To this end, distant supervision was proposed to automatically label RE data, and thus largely increased the number of annotated instances. Unfortunately, lots of noise relation annotations brought by automatic labeling become a new obstacle. Some recent studies have shown that the teacher-student framework of knowledge distillation can alleviate the interference of noise relation annotations via label softening. Nevertheless, we find that they still suffer from two problems: propagation of inaccurate dark knowledge and constraint of a unified distillation temperature . In this article, we propose a simple and effective Multi-instance Dynamic Temperature Distillation (MiDTD) framework, which is model-agnostic and mainly involves two modules: multi-instance target fusion (MiTF) and dynamic temperature regulation (DTR). MiTF combines the teacher’s predictions for multiple sentences with the same entity pair to amend the inaccurate dark knowledge in each student’s target. DTR allocates alterable distillation temperatures to different training instances to enable the softness of most student’s targets to be regulated to a moderate range. In experiments, we construct three concrete MiDTD instantiations with BERT, PCNN, and BiLSTM-based RE models, and the distilled students significantly outperform their teachers and the state-of-the-art (SOTA) methods.

Application of Dynamic Temperature-Humidity Chamber for Measuring Moisture Sorption Isotherms of Biomaterials as Compared to the Conventional Isopiestic Method

Measurement of water activity and moisture sorption isotherms of foods and biomaterials are important to determine the state of water. In this work, a dynamic temperature-humidity (DTH) controlled chamber was used to measure water sorption isotherm and compared with the conventional isopiestic method. Temperature and relative humidity of DTH chamber can be controlled in the range of -15 to 100°C and 0 to 98%, respectively; thus, measurement of water activity at any point can be measured within the above ranges. The DTH chamber method showed high reproducibility as compared with the conventional isopiestic method when measured isotherms of cellulose, lignin, and hemicellulase were compared at 30°C. Finally, isotherm data of cellulose, lignin, and hemicellulase were generated in the temperature range of 10-90°C using DTH chamber, and these were modelled by BET and GAB equations. The model parameters were correlated with the temperature.

Emotional Response to Vibrothermal Stimuli

Emotional response to haptic stimuli is a widely researched topic, but the combination of vibrotactile and thermal stimuli requires more attention. The purpose of this study is to investigate emotional response to vibrothermal stimulation by combining spatiotemporal vibrotactile stimulus with dynamic thermal stimulus (hot or cold). The vibrotactile and thermal stimuli were produced using the Haptic Chair and the Embr wave thermal bracelet, respectively. The results show that spatiotemporal vibrotactile patterns and their duration, and dynamic thermal stimulation, have an independent effect on the emotional response. Increasing duration generally increases the valence and arousal of emotional response. Shifting the dynamic temperature from cold to hot generally decreases the valence of emotional response but has no significant effect on arousal. Nevertheless, certain spatiotemporal patterns do exhibit unique responses to changes in dynamic temperature, although no interaction effects were found. The results show the potential of designing affective haptic interfaces using multimodal vibrothermal feedback.

Offshore Implementation of Temperature Microchip Under Critical Well Conditions

Drilling microchips are millimeter-size sensing devices, capable of measuring in-situ downhole temperature, and at the same time, withstanding harsh downhole conditions. In this work, 140 microchips were dropped from the drill pipe during the connections. The devices travel through the bottomhole assembly (BHA), drill bit, annulus, and eventually get recovered at the shale shaker. A total of 80 microchips were recovered at the shaker, which resulted in a physical recovery rate of 57%. The microchip recorded the dynamic temperature profile of the entire wellbore including a long openhole section only a few hours before the well turned into total loss. The data downloaded from the microchip shows an excellent consistency throughout the three tests. The measured dynamic bottomhole temperature provides a correction of 10 deg F to the best practice of the industry in terms of downhole thermal simulation, offering valuable measured input for the optimization of thermal activated LCMs or cementing job. To our best knowledge, it is the industry's first successful attempt in logging an openhole section in a highly loss zone. The microchip recorded the dynamic temperature profile of a long open hole only a few hours before the well turned into a total loss. Due to the lack of industrial solutions for downhole temperature measurement under such conditions, the microchip technology showed unique advantage for critical applications, especially in operations with highly valued assets.

Numerical Modeling of Temperature-Dependent Cell Membrane Permeability to Water Based on a Microfluidic System with Dynamic Temperature Control

In order to describe temperature-dependent cell osmotic behaviors in a more reliable method, a novel mathematical mass transfer model coupled with dynamic temperature change has been established based on the combination of a time domain to temperature domain transformation equation and a constant temperature mass transfer model. This novel model is numerically simulated under multiple temperature changing rates and extracellular osmolarities. A microfluidic system that can achieve single-cell osmotic behavior observation and provide dynamic and swift on-chip temperature control was built and tested in this paper. Utilizing the temperature control system, the on-chip heating processes are recorded and then described as polynomial time–temperature relationships. These dynamic temperature changing profiles were performed by obtaining cell membrane properties by parameter fitting only one set of testing experimental data to the mathematical model with a constant temperature changing rate. The numerical modeling results show that predicting the osmotic cell volume change using selected dynamic temperature profiles is more suitable for studies concerning cell membrane permeability determination and cryopreservation process than tests using constant temperature changing rates.