Optimal pruning for tree-structured vector quantization     

Jianhua Lin  James A. Storer  Martin Cohn 《Information processing & management》1992,28(6)

Tree-structured vector quantization (VQ) is a technique designed to represent a codebook that simplifies encoding as well as vector quantizer design. Most design algorithms for tree-structured VQ used in the past are based on heuristics that successively partition the input space. Recently, Chou, Lookabaugh and Gray proposed a tree-pruning heuristic in which a given initial tree is pruned backwards according to certain optimization criterion. We define the notion of an optimal pruned tree subject to a cost constraint and study the computational complexity of finding such an optimal tree for various cost functions. Under the assumption that all trees are equally probable, we show that, on the average, the number of pruned trees in a given tree is exponential in the number of leaves. Furthermore, we prove that finding an optimal pruned tree subject to constraints such as entropy or the expected-depth is NP-hard. However, we show that when the constraint is the number of leaves, the problem can be solved in polynomial time. We develop an algorithm to find the optimal pruned tree in O(nk) time, where n is the size of the initial tree and kis the constraint size.  相似文献   

Assessing basic research: Some partial indicators of scientific progress in radio astronomy     

Ben R. Martin  John Irvine 《Research Policy》1983,12(2):61-90

As the costs of certain types of scientific research have escalated and as growth rates in overall national science budgets have declined, so the need for an explicit science policy has grown more urgent. In order to establish priorities between research groups competing for scarce funds, one of the most important pieces of information needed by science policy-makers is an assessment of those groups' recent scientific performance. This paper suggests a method for evaluating that performance.After reviewing the literature on scientific assessment, we argue that, while there are no simple measures of the contributions to scientific knowledge made by scientists, there are a number of ‘partial indicators’ — that is, variables determined partly by the magnitude of the particular contributions, and partly by ‘other factors’. If the partial indicators are to yield reliable results, then the influence of these ‘other factors’ must be minimised. This is the aim of the method of ‘converging partial indicators’ proposed in this paper. We argue that the method overcomes many of the problems encountered in previous work on scientific assessment by incorporating the following elements: (1) the indicators are applied to research groups rather than individual scientists; (2) the indicators based on citations are seen as reflecting the impact, rather than the quality or importance, of the research work; (3) a range of indicators are employed, each of which focusses on different aspects of a group's performance; (4) the indicators are applied to matched groups, comparing ‘like’ with ‘like’ as far as possible; (5) because of the imperfect or partial nature of the indicators, only in those cases where they yield convergent results can it be assumed that the influence of the ‘other factors’ has been kept relatively small (i.e. the matching of the groups has been largely successful), and that the indicators therefore provide a reasonably reliable estimate of the contribution to scientific progress made by different research groups.In an empirical study of four radio astronomy observatories, the method of converging partial indicators is tested, and several of the indicators (publications per researcher, citations per paper, numbers of highly cited papers, and peer evaluation) are found to give fairly consistent results. The results are of relevance to two questions: (a) can basic research be assessed? (b) more specifically, can significant differences in the research performance of radio astronomy centres be identified? We would maintain that the evidence presented in this paper is sufficient to justify a positive answer to both these questions, and hence to show that the method of converging partial indicators can yield information useful to science policy-makers.  相似文献   

CERN: Past performance and future prospects: II. The scientific performance of the CERN accelerators     

John Irvine  Ben R. Martin 《Research Policy》1984,13(5):247-284

In a series of three papers, we attempt to evaluate the past scientific performance of the three main particle accelerators at the Geneva-based European Organization for Nuclear Research (CERN) over the period since 1960, and to assess the future prospects for CERN and its users during the next ten to fifteen years.We concerned ourselves in a previous paper (Paper 1 - Martin and Irvine [51] with the position of the CERN accelerators in world high-energy physics relative to those at other large laboratories working in the field. We dealt primarily with the period from 1969 to 1978, and attempted to establish how the experimental output from the three principal CERN accelerators, taken as a whole, compared with that from other major facilities. In undertaking this comparative evaluation, we drew on the method of “converging partial indicators” used in previous studies of three Big Science specialties.In contrast, this paper (Paper II) focuses in detail on the scientific performance of each of the CERN accelerators taken individually. In particular, it asks, first, how the outputs from the CERN 28 GeV (giga or billion electron-volts) Proton Synchrotron compare with those from a very similar 33 GeV American accelerator at Brookhaven National Laboratory over the past two decades? Second, how great have been the experimental achievements of the Intersecting Storage Rings in world terms? And, third, how do the outputs from the CERN 400 GeV Super Proton Synchrotron and from a rival American machine at Fermi National Accelerator Laboratory compare? Attempts are then made to identify the main factors responsible for determining the relative scientific performance of each CERN machine.These factors are of relevance to the subject of a third paper (Paper III - Martin and Irvine [52]) which sets out to assess the future prospects for CERN and in particular for LEP, the large electron-positron collider scheduled for completion in the latter part of the 1980s. What are the construction requirements (financial and technical) associated with LEP, and how easily will they be met? How does the scientific potential of LEP compare with that of other major accelerators under construction around the world? And, in the light of the previous record of the CERN accelerators, to what extent is this potential likely to be realized? The paper concludes with a discussion of the extent to which predictive techniques can be utilized in the formulation of scientific priorities, and of the problems in current science policy-making that such techniques might help address.  相似文献   

The role of social embeddedness in professorial entrepreneurship: a comparison of electrical engineering and computer science at UC Berkeley and Stanford   总被引：1，自引：0，他引：1  

Martin Kenney  W Richard Goe 《Research Policy》2004,33(5):691-707

Professorial entrepreneurship has recently attracted much attention. This paper draws upon historical research, a survey of faculty, and an Internet-based methodology for identifying professorial affiliations with entrepreneurial firms at two of the premier electrical engineering and computer science departments in the US, The University of California, Berkeley and Stanford. We employ the concept of “nested embeddedness” to explain why the faculty members in these two institutions have different levels of entrepreneurship and corporate involvement. EE&CS faculty at both universities were found to be socially embedded in departments and disciplines that supported and placed value on entrepreneurial activities. However, while being embedded in a university environment with a history of success and high level of support for entrepreneurship, EE&CS faculty at Stanford had a significantly greater level of corporate involvement, including the founding of start-ups. Although significantly less than Stanford, the level of corporate involvement among EE&CS faculty at Berkeley was also substantial. This suggests that being embedded in an academic department and disciplines with cultures that are supportive of entrepreneurial activity can help counteract the disincentives created by a university environment that is not strongly supportive of these activities.  相似文献   

Mindset change in a cross-cultural context     

Martin Loeve 《Action Learning: Research and Practice》2007,4(2):211-218

Imagine you are Thai and a member of a management team from a European company that acts in an Asian emerging market. Imagine you are a European expat with the assignment to double the turnover from that Asian company in a few years. Imagine you are a Change Maker and they ask you to facilitate the (multinational) management team of that Asian company to develop a market strategy and translate it into a business plan, and you have to deal with cultural differences. How do you manage a small-scale change process based on the principles of action learning in that specific multinational context? That's the essence of this account of practice.  相似文献   

Marital conflict and conduct problems in Children of Twins   总被引：3，自引：0，他引：3  

Harden KP  Turkheimer E  Emery RE  D'Onofrio BM  Slutske WS  Heath AC  Martin NG 《Child development》2007,78(1):1-18

The Children-of-Twins design was used to test whether associations between marital conflict frequency and conduct problems can be replicated within the children of discordant twin pairs. A sample of 2,051 children (age 14-39 years) of 1,045 twins was used to estimate the genetic and environmental influences on marital conflict and determine whether genetic or environmental selection processes underlie the observed association between marital conflict and conduct problems. Results indicate that genetic and nonshared environmental factors influence the risk of marital conflict. Furthermore, genetic influences mediated the association between marital conflict frequency and conduct problems. These results highlight the need for quasiexperimental designs in investigations of intergenerational associations.  相似文献   

Together I Can! Joint Attention Boosts 3- to 4-Year-Olds’ Performance in a Verbal False-Belief Test     

Elia Psouni  Andreas Falck  Leni Boström  Martin Persson  Lisa Sidén  Maria Wallin 《Child development》2019,90(1):35-50

Effects of joint attention were addressed on 3- to 4-year-olds’ performance in a verbal false-Belief Test (FBT), featuring the experimenter as co-watcher rather than narrator. In two experiments, children (N = 183) watched a filmed-FBT jointly with a test leader, disjointed from a test leader, or alone. Children attending jointly with a test leader were more likely to pass the FBT compared with normative data and to spontaneously recall information indicating false-belief understanding, suggesting that joint attention strengthens the plausibility of the FBT and renders plot-critical information more salient. In a third experiment (N = 59), results were replicated using a typical, image-based FBT. Overall findings highlight the profound impact of experimenter as social context in verbal FBTs, and link recall of specific story features to false-belief understanding.  相似文献   

Interventions supporting baccalaureate achievement of Latinx STEM students matriculating at 2-year institutions: A systematic review     

Julie P. Martin  Nathan Hyungsok Choe  Jared Halter  Margaret Foster  Jeffrey Froyd  Maura Borrego  Erica R. Winterer 《科学教学研究杂志》2019,56(4):440-464

Numerous national reports have identified the importance of significantly improving pathways that begin with Latinx students enrolling in 2-year institutions and ultimately completing baccalaureate degrees in STEM fields at 4-year institutions. Many programs using multiple interventions have been designed, implemented, and studied to achieve this goal. To synthesize what has been learned from studies of these programs, this article presents a systematic review of published studies of programs designed to support Latinx student success in 2-year institutions and successful transfer to 4-year institutions, particularly for STEM majors. A total of 49 quantitative, 9 qualitative, and 16 mixed-methods studies published as reports, articles, or dissertations since 1980 were identified that met the criteria for the review. Studies covered a wide range of interventions, including mentoring, counseling, advising, study groups, tutoring, scholarships, orientations, career services, undergraduate research, articulation agreements, and transfer programs. Individually, these studies report positive influences on student success outcomes, including 2- and 4-year graduation, transfer to a 4-year institution, retention, and success in individual courses. However, the number of qualifying studies was surprisingly small, considering the importance of improving success of Latinx students and the length of time during which the problem has been repeatedly emphasized. Few interventions have been undertaken from explicitly assets-based perspectives or theoretical frameworks. The lack of explicit frameworks underlying interventions—combined with a sole/primary focus on students—suggests many interventions were approached from a deficit-based perspective. Further, the study found no pattern of replication studies that might confirm effectiveness of potentially promising interventions. Based on our analysis of evaluations presented in the studies, it does not appear that the research community has developed agreed-upon methods to evaluate commonly agreed-upon outcomes. Finally, no intervention has been sufficiently supported that widespread implementation could be recommended.  相似文献   

‘What does an O say when there's no E at the end?’ Parents' reading‐related knowledge and feedback during child‐to‐parent reading     

Aviva Segal  Sandra Martin‐Chang 《Journal of Research in Reading》2019,42(2):349-370

  相似文献   

Discourse forms in a classroom transitioning to student-centred scientific inquiry through co-teaching     

Carol A. B. Rees  Wolff-Michael Roth 《International Journal of Science Education》2019,41(5):586-606

Teacher–student discourse continues to be teacher-centred even though researchers and reform documents have recommended changes toward increased levels of student-centred discourse. In science education this situation is paralleled by effort to make scientific inquiry more student-centred. The purpose of this study was to investigate how discourse forms changed over time in a classroom where the regular teacher and his students were scaffolded in the transitioning to student-centred scientific inquiry. Video-recordings were collected at intervals over one academic year. Three prominent forms of discourse were identified: two teacher-authoritative forms and one more interactive, dialogic form. As the lessons increasingly turned into student-centred scientific inquiry, a shift to the dialogic discourse form was found. Co-teaching provided for (a) guidance towards an organisation of events in the classroom that included regular teacher–student dialogue (b) modelling of the more dialogic form of discourse.  相似文献