A step towards unification of syntactic and statistical pattern recognition

The problem of pattern recognition is discussed in terms of single-entity representation versus multiple-entity representation. A combined syntactic-semantic approach based on attributed grammars is suggested. Syntax-semantics tradeoff in pattern representation is demonstrated. This approach is intended to be an initial step toward unification of syntactic and statistical approaches to pattern recognition.     

含砷氰化尾矿回填利用污染控制技术试验研究

针对某黄金生产企业含砷氰化尾矿污染特征,开展了搅拌洗涤法、臭氧氧化法、酸化溶砷法、铁盐固砷法等多种无害化方法联合处理试验研究,旨在将该含砷氰化尾矿处理至满足氰渣规范回填利用污染控制要求。结果表明：该含砷氰化尾矿回填利用污染控制技术工艺为压滤调浆搅拌洗涤+臭氧氧化+酸化溶砷+铁盐固砷,最佳参数为原矿浆压滤后加水调浆,矿浆浓度40%,臭氧投加量0.66 g/L,酸化溶砷pH值3、曝气量0.1 m³/h、反应时间2 h,铁盐固砷七水合硫酸亚铁投加量20.0 g/L、反应时间1 h。研究结果为该黄金生产企业含砷氰化尾矿回填利用提供了技术支撑。     

在Fe-Al-C系统中合成优质Ⅱa型宝石级金刚石单晶

和优质Ib型宝石级金刚石单晶生长相比，作为当今高温高压晶体生长的一种高精尖技术，优质1／a型宝石级金刚石单晶对合成技术提出了更高更苛刻的要求：枢研究从晶体的生长速度出发，发现一开始阶段（大约几个小时）的晶体生长速度对优质宝石级金刚石单晶的后期生长争关重要。埘Fe-Al-C系统（Al含量不大于2．5wt％）来说，采用多品种法将晶体开始阶段的晶体生长速度由1．5mg／h降至0．5mg／h后，对生长过程中金属包裹体的进入有了明显的抑制作用，晶体的质量有了很大提高。从晶体中包裹体的存任形式来看，为了获得优质Ⅱa型宝石级金刚石单晶，在触媒中人为地添加除氮剂给晶体生长过程中的排杂过程带来了很大的难度：为了更好的实现排杂，必须很好的处理晶体表面的径向平铺生长速度和晶体轴向的堆积生长速度之间的火系。     

红外测温技术在离心复合铸造轧辊生产上的应用

本文分析了离心复合铸造轧辊生产过程中离心机停机温度和时间对轧辊质量的影响 ,提出了用先进的红外测温技术来预报停机温度和时间 ,以实现微机监控适时停机 ,保证轧辊内外层熔合良好 ;同时指出了红外测温技术在离心复合铸造轧辊生产上的应用前景     

人胎盘泌乳素的提纯及裂解片断免疫学特性的研究

采用人胎盘泌乳素（hPL）-亲和层析和抗人血清抗体-亲和层析技术，从胎盘匀浆中提取hPL纯品，经圆盘电泳、免疫电泳及交叉试验证实为单一成分。溴化氰水解后的产物，以高效液相色谱分析。分离出7个肽链片断峰，分别用hPL和人生长激素（hGH）放免法检测其活性，其中第6肽链片断无hGH免疫活性，若用此片断制备特异抗体，可避免与hGH出现交叉反应。     

异丁烯两段催化氧化制甲基丙烯酸催化剂研究 Ⅰ．异丁烯气相氧化制甲基丙烯醛多元复合氧化物催化剂的研究

本文报导用于异丁烯（叔丁醇）选择氧化制ＭＡＬ（甲基丙烯醛）的多元复合氧化物催化剂的研究结果。同共沉淀法制备的催化剂的活性主要决定于其化学组成，同时受到制备条件的影响。最佳催化剂的主要物相组成为ＰｂＯ３，Ｂｉ２ＭｏＯ３．３、ＮｉＭｏＯ４、Ｓｂ２Ｏ３及ａ－Ｆｅ２Ｏ３小颗粒。在３５５℃，１０００～２５００ｈ－１空速范围，叔丁醇：水：空气＝１：３．３：７（摩尔比）的条件下可获得异丁烯转化率达９６％，ＭＡＬ选择性达８９％的结果。     

Exploiting Batch Processing on Streaming Architectures to Solve 2D Elliptic Finite Element Problems: A Hybridized Discontinuous Galerkin (HDG) Case Study

Numerical methods for elliptic partial differential equations (PDEs) within both continuous and hybridized discontinuous Galerkin (HDG) frameworks share the same general structure: local (elemental) matrix generation followed by a global linear system assembly and solve. The lack of inter-element communication and easily parallelizable nature of the local matrix generation stage coupled with the parallelization techniques developed for the linear system solvers make a numerical scheme for elliptic PDEs a good candidate for implementation on streaming architectures such as modern graphical processing units (GPUs). We propose an algorithmic pipeline for mapping an elliptic finite element method to the GPU and perform a case study for a particular method within the HDG framework. This study provides comparison between CPU and GPU implementations of the method as well as highlights certain performance-crucial implementation details. The choice of the HDG method for the case study was dictated by the computationally-heavy local matrix generation stage as well as the reduced trace-based communication pattern, which together make the method amenable to the fine-grained parallelism of GPUs. We demonstrate that the HDG method is well-suited for GPU implementation, obtaining total speedups on the order of 30–35 times over a serial CPU implementation for moderately sized problems.     

Generalized Transfer Subspace Learning Through Low-Rank Constraint

It is expensive to obtain labeled real-world visual data for use in training of supervised algorithms. Therefore, it is valuable to leverage existing databases of labeled data. However, the data in the source databases is often obtained under conditions that differ from those in the new task. Transfer learning provides techniques for transferring learned knowledge from a source domain to a target domain by finding a mapping between them. In this paper, we discuss a method for projecting both source and target data to a generalized subspace where each target sample can be represented by some combination of source samples. By employing a low-rank constraint during this transfer, the structure of source and target domains are preserved. This approach has three benefits. First, good alignment between the domains is ensured through the use of only relevant data in some subspace of the source domain in reconstructing the data in the target domain. Second, the discriminative power of the source domain is naturally passed on to the target domain. Third, noisy information will be filtered out during knowledge transfer. Extensive experiments on synthetic data, and important computer vision problems such as face recognition application and visual domain adaptation for object recognition demonstrate the superiority of the proposed approach over the existing, well-established methods.     

Increased α-linolenic acid intake increases tissue α-linolenic acid content and apparent oxidation with little effect on tissue docosahexaenoic acid in the guinea pig

The essential fatty acids do not have identical roles in nutrition. Linoleic acid (LA) accumulates throughout the body of most mammals, whereas α-linolenic acid (ALA) is rarely found in tissue lipids to the same extent as LA. It has been argued that this is the result of metabolism of ALA to docosahexaenoic acid (DHA) or that ALA is rapidly β-oxidized to acetyl CoA and CO₂. In this study, we consider the effect of high and low ALA levels on the tissue distribution of ALA and other n-3 polyunsaturated fatty acids (PUFA) in all tissues. Guinea pigs were fed one of two defined diets for 3 wk from wearning with both diets containing 1.8% (by weight) of LA and either 1.7% ALA or 0.03% ALA. The high ALA diet was associated with significantly increased ALA levels in all tissues except the brain and significantly increased levels of long-chain n-3 PUFA in all tissues except intestines, brain, carcass, and skin. The long-chain n-3 PUFA content of the whole body was less than 5% of that of the ALA content in both diet groups, and the major long-chain n-3 PUFA (>66% of total) in the body was 22∶5n−3. The brain was the only tissue where the DHA content exceeded that of 22∶5n−3. On the low ALA diet, there appeared to be conservation of ALA based on a comparison of the ratio of LA to ALA in the tissues compared with that in the diet. On the high ALA diet there was a loss of ALA relative to LA in the tissues compared with the diet. These studies suggest that the low levels of tissue ALA in the guinea pig are likely the result of β-oxidation or excretion via the skin and fur rather than metabolism to DHA.     

Study of collective user behaviour in Twitter: a fuzzy approach

The study of collective user behaviours in social networking sites has become an increasing important topic in social media mining. Understanding such behaviours has its potential to extract actionable patterns that can be beneficial to develop effective marketing strategies, optimise user experiences and maximise website revenues. With the rapid development of micro-blogging, Twitter has become a richer source of intelligence that can be used to study collective user behaviour, due to its efficient and meaningful user-to-user interactions. However, the classical statistical methods have some drawbacks in bridging the gap between user-generated data and human analysts who mostly use linguistic terms to analyse data and model/summarise knowledge learned. To address this gap, this work proposes a new approach, which employs the mass assignment theory-based fuzzy association rules algorithm (MASS-FARM), for the first time, to extract useful interaction behaviour of Twitter users. The influential factors (including activity time, number of friends/followers and the number of tweets) are represented as fuzzy granules, and the associations amongst are studied by employing MASS-FARM. The collective user behaviours are analysed in the Reply category and the Non-Reply category, respectively. The applicability and usefulness of the proposed method are demonstrated via an empirical study on a collected Twitter data set. The derived results are also discussed and compared with existing works.