Unsupervised domain adaptation for lip reading based on cross-modal knowledge distillation

We present an unsupervised domain adaptation (UDA) method for a lip-reading model that is an image-based speech recognition model. Most of conventional UDA methods cannot be applied when the adaptation data consists of an unknown class, such as out-of-vocabulary words. In this paper, we propose a cross-modal knowledge distillation (KD)-based domain adaptation method, where we use the intermediate layer output in the audio-based speech recognition model as a teacher for the unlabeled adaptation data. Because the audio signal contains more information for recognizing speech than lip images, the knowledge of the audio-based model can be used as a powerful teacher in cases where the unlabeled adaptation data consists of audio-visual parallel data. In addition, because the proposed intermediate-layer-based KD can express the teacher as the sub-class (sub-word)-level representation, this method allows us to use the data of unknown classes for the adaptation. Through experiments on an image-based word recognition task, we demonstrate that the proposed approach can not only improve the UDA performance but can also use the unknown-class adaptation data.

A Research Agenda for Smarter Cyber-Physical Systems

With the advancing digitisation of society and industry we observe a progressing blending of computational, physical, and social processes. The trustworthiness and sustainability of these systems will be vital for our society. However, engineering modern computing systems is complex as they have to: i) operate in uncertain and continuously changing environments, ii) deal with huge amounts of data, and iii) require seamless interaction with human operators. To that end, we argue that both systems and the way we engineer them must become smarter. With smarter we mean that systems and engineering processes adapt and evolve themselves through a perpetual process that continuously improves their capabilities and utility to deal with the uncertainties and amounts of data they face. We highlight key engineering areas: cyber-physical systems, self-adaptation, data-driven technologies, and visual analytics, and outline key challenges in each of them. From this, we propose a research agenda for the years to come.

Optimising assessment of dark adaptation data using time to event analysis

In age-related macular degeneration (AMD) research, dark adaptation has been found to be a promising functional measurement. In more severe cases of AMD, dark adaptation cannot always be recorded within a maximum allowed time for the test (~ 20–30 min). These data are recorded either as censored data-points (data capped at the maximum test time) or as an estimated recovery time based on the trend observed from the data recorded within the maximum recording time. Therefore, dark adaptation data can have unusual attributes that may not be handled by standard statistical techniques. Here we show time-to-event analysis is a more powerful method for analysis of rod-intercept time data in measuring dark adaptation. For example, at 80% power (at α = 0.05) sample sizes were estimated to be 20 and 61 with uncapped (uncensored) and capped (censored) data using a standard t-test; these values improved to 12 and 38 when using the proposed time-to-event analysis. Our method can accommodate both skewed data and censored data points and offers the advantage of significantly reducing sample sizes when planning studies where this functional test is an outcome measure. The latter is important because designing trials and studies more efficiently equates to newer treatments likely being examined more efficiently.

Adaptive mesh refinement technique for continuum mechanics problems

Adaptive Mesh Refinement (AMR) is widely used today as a way to solve problems in the mechanics of continuous media, which makes it possible to increase the accuracy of the solution at an economical cost of computational resources. The paper presents new structures for storing and processing data for octree mesh adaptation in the leaf model paradigm. The current version of the technique implements a set of algorithms focused on applications to cluster-type parallel computing systems: dynamic mesh adaptation, data structure synchronization, and load balancing of the computing complex. The developed toolkit supports the functionality required for the implementation of various numerical models of continuum mechanics. As an example of possible applications, a difference schemes for heat conduction and gas dynamics using the developed AMR technique are discussed. The results of numerical experiments with model problems are presented.

Assessing the feasibility of adaptation options: methodological advancements and directions for climate adaptation research and practice

The Paris Agreement put adaptation prominently on the global climate action agenda. Despite a surge in research and praxis-based knowledge on adaptation, a critical policy roadblock is synthesizing and assessing this burgeoning evidence. We develop an approach to assess the multidimensional feasibility of adaptation options in a robust and transparent manner, providing direction for global climate policy and identifying knowledge gaps to further future climate research. The approach, which was tested in the IPCC Special Report on 1.5 °C (SR1.5) to assess 23 adaptation options, is underpinned by a systematic review of recent literature, expert elicitation, and iterative peer review. It responds to the challenge of limited agreement on adaptation indicators, lack of fine-scale adaptation data, and challenges of assessing synergies and trade-offs with mitigation. The findings offer methodological insights into how future assessments such as the IPCC Assessment Report (AR) six and regional, national, and sectoral assessment exercises could assess adaptation feasibility and synthesize the growing body of knowledge on climate change adaptation.

125 Effects of copper hydroxychloride and increasing concentrations of dietary fat on growth performance, total tract endogenous loss of fat, and apparent total tract digestibility of fat by growing pigs

Two experiments were conducted to test the hypothesis that Cu hydroxychloride (IntelliBond C, Micronutrients, LLC, Indianapolis, IN) improves gain:feed of pigs by increasing apparent total tract digestibility (ATTD) of fat. In experiment 1, 144 pigs (15.40 ± 2.39 kg) were allotted to 1 of 6 treatments. Four diets contained 0, 2, 4, or 6% choice white grease (CWG) and 2 diets contained 150 mg/kg Cu from Cu hydroxychloride and 0 or 2% CWG. Data for growth performance were analyzed by contrast statements, and Cu equivalency of CWG was determined using prediction equations derived from regression procedures using SAS. Results indicated that gain:feed linearly increased (P ≤ 0.05) as CWG concentration increased in diets (Table 1). Supplementation of Cu hydroxychloride to diets improved (P ≤ 0.05) gain:feed of pigs, which resulted in a CWG equivalence of 2.8 to 3.8% for 150 mg/kg Cu hydroxychloride. In experiment 2, 80 pigs (18.24 ± 1.81 kg) were housed individually in metabolism crates and randomly allotted to 1 of 10 diets. Eight diets contained 0, 15, 30, or 45% distillers dried grains with solubles (DDGS) and 0 or 150 mg/kg Cu hydroxychloride. Two additional diets contained 2% CWG and 15% DDGS and 0 or 150 mg/kg Cu hydroxychloride. Feces were collected for 4 d after 5 d adaptation. Data were analyzed by contrast statements and regression procedures using SAS. Supplementation of Cu to diets improved (P ≤ 0.05) ATTD of fat by 20% due to reduced (P ≤ 0.05) endogenous loss of fat (from 11.23 to 7.14 g/kg dry matter intake), but did not affect energy digestibility or true total tract digestibility of fat. In conclusion, supplementation of 150 mg/kg Cu hydroxychloride to diets improved gain:feed of pigs with a CWG equivalence of 2.8 to 3.8%, but this was not due to improved energy digestibility.

ANALISIS TEKNIK PENERJEMAHAN ADAPTASI DAN VARIASI PADA SUBTITLE FILM BATMAN VERSI BAHASA JAWA MATARAMAN

<p><em>The aim of this study is to identify and describe the cultural elements translated using adaptation technique and linguistic elements translated using variation technique in Batman movie subtitled into Mataraman Javanese. Then explain the translator’s reason of using the adaptation and variation technique, and describe the effect of both techniques towards the translation quality. This is a prescriptive qualitative study and categorized as an embedded research. The sources of data are 1) script of Batman movie and its translation in Mataraman Javanese, 2) the raters as informant who evaluate the translation in terms of accuracy and acceptability, and 3) the target audiences to rate the comprehension.</em></p><em>Based on the research result, the researcher collected 110 data which consist of 48 adaptation data and 62 variation data. The adaptation data in this study are ecology, material culture, social culture, procedure/ activity/ artistic term, economic system and language. Meanwhile, the variation data contain the personal pronouns for first person, second person, and third person translated into speech level of ngoko (showing low politeness), madya (showing middle politeness) and krama (showing high politeness). The use of adaptation and variation techniques by the translator is in line with the purpose of Mataraman Javanese program held by local TV station to accommodate the local wisdom. For that reason, the translation ideology of this subtitle is domestication by employing those translation techniques which are closes to the target language culture.</em> <em>Generally, the use of adaptation and variation techniques in this study is quite accurate, acceptable in the target culture and comprehensible to the target audience.</em>