Search

The Normalized Cut (NCut) objective function, widely used in data clustering and image segmentation, quantifies the cost of graph partitioning in a way that biases clusters or segments that are balanced towards having lower values than unbalanced partitionings. However, this bias is so strong that it avoids any singleton partitions, even when vertices are very weakly connected to the rest of the graph. Motivated by the B\"uhler-Hein family of balanced cut costs, we propose the family of Compassionately Conservative Balanced (CCB) Cut costs, which are indexed by a parameter that can be used to strike a compromise between the desire to avoid too many singleton partitions and the notion that all partitions should be balanced. We show that CCB-Cut minimization can be relaxed into an orthogonally constrained $\ell_{\tau}$-minimization problem that coincides with the problem of computing Piecewise Flat Embeddings (PFE) for one particular index value, and we present an algorithm for solving the relaxed problem by iteratively minimizing a sequence of reweighted Rayleigh quotients (IRRQ). Using images from the BSDS500 database, we show that image segmentation based on CCB-Cut minimization provides better accuracy with respect to ground truth and greater variability in region size than NCut-based image segmentation., Comment: Long version of paper accepted at CVPR 2018

Functional connectivity analysis yields powerful insights into our understanding of the human brain. Group-wise functional community detection aims to partition the brain into clusters, or communities, in which functional activity is inter-regionally correlated in a common manner across a group of subjects. In this article, we show how to use multiple-view spectral clustering to perform group-wise functional community detection. In a series of experiments on 291 subjects from the Human Connectome Project, we compare three versions of multiple-view spectral clustering: MVSC (uniform weights), MVSCW (weights based on subject-specific embedding quality), and AASC (weights optimized along with the embedding) with the competing technique of Joint Diagonalization of Laplacians (JDL). Results show that multiple-view spectral clustering not only yields group-wise functional communities that are more consistent than JDL when using randomly selected subsets of individual brains, but it is several orders of magnitude faster than JDL., Comment: Presented at The MICCAI-BACON 16 Workshop (https://arxiv.org/abs/1611.03363)

We study how single-crystal chromium films of uniform thickness on W(110) substrates are converted to arrays of three-dimensional (3D) Cr islands during annealing. We use low-energy electron microscopy (LEEM) to directly observe a kinetic pathway that produces trenches that expose the wetting layer. Adjacent film steps move simultaneously uphill and downhill relative to the staircase of atomic steps on the substrate. This step motion thickens the film regions where steps advance. Where film steps retract, the film thins, eventually exposing the stable wetting layer. Since our analysis shows that thick Cr films have a lattice constant close to bulk Cr, we propose that surface and interface stress provide a possible driving force for the observed morphological instability. Atomistic simulations and analytic elastic models show that surface and interface stress can cause a dependence of film energy on thickness that leads to an instability to simultaneous thinning and thickening. We observe that de-wetting is also initiated at bunches of substrate steps in two other systems, Ag/W(110) and Ag/Ru(0001). We additionally describe how Cr films are converted into patterns of unidirectional stripes as the trenches that expose the wetting layer lengthen along the W[001] direction. Finally, we observe how 3D Cr islands form directly during film growth at elevated temperature. The Cr mesas (wedges) form as Cr film steps advance down the staircase of substrate steps, another example of the critical role that substrate steps play in 3D island formation.

This paper on recruiting and retaining minorities details the difficulties and successes of Gainesville College (Georgia) to induct and retain minorities from 1993 to 1998. During this five-year period, Black enrollment at the college decreased by 9.1% (from 84 to 77), while Hispanic and Asian enrollments increased by 61.2% (26 to 67) and 35.7% (27 to 42), respectively. Overall, White non-Hispanics accounted for 92.2% of the total enrollment in the fall of 1998, in contrast to 91.8% in fall 1997 and 93.7% in fall 1993. The minority population as a whole decreased slightly from 8.2% in fall 1997 to 7.8% in fall 1998. Included are summaries of changes in minority student enrollment, racial and ethnic distribution of the ten-county service area, recruitment and retention strategies for minority students, changes in graduation rates of minority students, and plans for improvement. The paper also details the racial and ethnic demographics of the college's service area, including K-12 enrollments, and the percentage of high school graduates with college-preparatory diplomas. Appended are various tables of data describing racial and ethnic percentages in various cohorts. (AS)

Gainesville College (GC) administered a Needs Assessment Survey in February 1998 to determine perceptions and attitudes about the college and areas that need improvement. Two groups were surveyed: high school counselors and principals, and community leaders such as Chamber of Commerce presidents and college trustees. There was a relatively high response rate from both groups. Among the high school counselors and principals, the highest ratings of satisfaction for GC occurred in areas such as financial aid services, administrators, guidance/counseling, and academic advisement. The lowest survey ratings concerned recreational facilities, student recruitment, and learning support services. Among the community leaders, highest ratings went to community and high school interaction, admissions office services, financial aid services, and preparation for transfer, while the lowest went to tutoring and academic support services, guidance/counseling services, academic advisement, student recruitment, and preparation for the job market. Includes data tables. Appendices include survey tabulations and forms for both groups. (YKH)

Gainesville College (GC) in Georgia conducted a study to determine how many traditional-aged students enter the school with General Education Development (GED) certificates and how well they progress academically. Students monitored were those who received their GED at 21 years of age or younger and entered GC within 4 years of receiving the certificate. Two hundred and seventy-six students who enrolled between fall 1991 and fall 1996 were monitored by transcript analysis to the end of fall 1997. The study group was overwhelmingly white with slightly more Asians than blacks and Hispanics. The percentage of females in the study group was only slightly higher than males at 53%. Only 15% of the students did not require any developmental studies courses, whereas 85% required between 1 and 3. Moreover, in the fall of 1997, more GED students were placed in developmental reading, English, and math courses than other first-time students. Results of the study showed a GED-student completion ratio of 90% for academic credit courses, slightly lower than that of non-GED first-time students. Similarly, GC grade point averages were slightly lower for students in the GED study group than the average college GPA. (Contains 7 pages of data tables.) (YKH)

In 1990, a study was conducted to investigate the five-year transfer patterns of students who entered Gainesville College, a public two-year unit of the University System of Georgia, during the fall of 1990 as they matriculated to public baccalaureate-granting institutions within the state. A variety of student cohorts were followed from fall 1990 to the end of fall 1995, including: students who started with remedial needs, male and female populations, students with and without associate degrees, traditional versus nontraditional age (25 and over) entering students, credit hours earned at the starting college, business administration majors, and high school college preparatory students. Highlights include the following: (1) 26% (n=150) earned associate degrees at the college with 91 of the degrees earned by females and 132 earned by students less than 25; (2) 35.8% (n=205) did not earn over 15 academic credits at the College within five years of entering the college; (3) of the 208 who transferred, 51% (n=106) of the transfer activity from the college occurred over the second year and 83.7% (n=174) occurred over the third year; and (4) the transfer rate for those with three remedial requirements were 15.3% (n=15). Extensive data tables are included. (HAA)

As part of an effort to evaluate the Academic Computing, Tutoring, and Testing Center, Gainesville College, in Georgia, surveyed 672 students in morning classes, 424 students in evening classes, and all faculty and staff to determine their use of and satisfaction with the Center. Faculty and staff responses were received from 79 individuals, including 50 faculty members, 11 administrative, and 18 staff members. Study findings included the following: (1) 50% of the student respondents spent 1 to 5 hours in the Center each week, while 8.7% indicated that they had never used it; (2) the most commonly reported use of the Center by students was for classroom assignments requiring the use of a computer, cited by 759 respondents; (3) all 38 dimensions of the Center included on the survey received ratings of "adequate" or "very adequate" by 84% or more of the students, while the highest rated dimensions were computer hardware and software; (4) the dimensions receiving the highest percentage of "very adequate" ratings by faculty were willingness to offer orientation classes, hours of operation, willingness of staff to discuss problems, and willingness to help special needs students; and (5) the dimensions receiving the lowest number of "very adequate" ratings by faculty were the condition of the photocopy equipment, the amount of staff, and the availability of professional staff. The survey instruments with tabulated responses and student comments to open-ended questions are included. (TGI)

Based on information from "Georgia Trend" magazine examining economic conditions across Georgia, Gainesville College (GC) is expected to experience an expanding base of students over the next 5 years. With respect to Hall County and the nine contiguous counties that make up GC's service area, data indicate a population growth in the region, growth in employment, increases in total buying power and per capita personal income, potential industrial and residential development, a diversified economy, and the possible need to establish a university system within the largest county of Gwinnett. With respect to employment trends, the globalization of markets will favor jobs linked to transportation; wholesale and retail trade; hotels and lodging places; and financial, legal and business services. Low-skilled positions and white collar middle management, however, will suffer reductions. Increased reliance on digital information, automation, and mechanization will create jobs for educated workers, help flatten the corporate structure, and also eliminate many repetitive unskilled jobs. In addition, occupations geared towards customizing markets for consumers who demand more choice, convenience, and low prices can be expected to flourish. Finally, the fastest growing job markets will be the professional specialty occupations, such as teaching, computer-related fields, and health care providers, growing at a rate of 2.4% annually. Data tables are included. (TGI)

In February 1996, Gainesville College, in Georgia, conducted a study of students in its Legal Assistant Management (LAM) Program to determine retention rates, numbers of graduates, and course pass rates. Retention and graduation rates were calculated for 175 students who enrolled in at least one LAM course from spring 1991 to fall 1995. In addition, the course-taking behavior was analyzed of 154 students who had passed at least one LAM course since spring 1991. Study findings included the following: (1) the overall retention rate for students who started the LAM sequence in the spring of 1993 was 78%; (2) of the 116 students who took at least one LAM course from spring 1991 to winter 1994, 24% (n=28) had received an Associate Degree in LAM as of fall 1995; (3) 50 of the 105 students who took at least 1 LAM course and who passed it with a "D" or better between spring 1991 and winter 1994 went on to complete the LAM course series with a "D" average or better by fall 1995; and (4) excluding the 16% of students who started the series between spring 1991 and winter 1994 and dropped out of the series by fall 1995, 65% of the students completed the sequence within 8 or more quarters as of fall 1995. The survey instrument and responses to open-ended questions are attached. (HAA)

Every 4 years, Gainesville College, in Georgia, conducts a survey of entering students to determine their demographic characteristics and reasons for enrolling. In fall 1995, 275 first-time freshmen students were surveyed, representing 55% of the total number of first-time freshmen, while the results were compared to findings from the Higher Education Research Institute's American Freshmen Survey (AFS) for fall 1995. Study findings, based on responses from 65% of the freshmen students (n=179), included the following: (1) 86% were 21 years old or under, 62% were female, and 95% were white; (2) 55% expected to live in their parents' home, while 20% expected to live in their own home; (3) almost 90% indicated they would be enrolled on a full-time basis; (4) parents or relatives was the most commonly cited source of funding, cited by 33%; (5) 49% felt that they were fair, poor, or very poor at communicating orally, another 49% felt the same about their mathematical skills, as did 39% about their ability to communicate in writing; and (6) 93% felt that they had a good chance of earning a "B" average at the college. Comparisons to findings from the fall 1995 AFS study indicated that, nationwide, students' ratings of their high school preparation have declined and that financial concerns have continued to escalate, mirroring GC students' experience. A separate report on AFS findings is attached. Data tables are included. (HAA)

A study was conducted at Gainesville College (GC) in Georgia to investigate the enrollment patterns, retention rates, and graduation rates of black students. Depending on the data element, the period of time covered extended from the late 1980's to winter 1995. In some cases, comparisons were drawn with other units of the University System of Georgia. Study findings included the following: (1) American black enrollments consistently fell below 5% of the total enrollment between fall 1990 and winter 1995, with reasons for this low enrollment rate being the small number of college-ready black students coming out of local high schools, increases in enrollments at Lanier Technical Institute, and local students with financial aid choosing to attend other colleges and universities; (2) 5-year systemwide retention rates for black students at GC were lower than for all other students, though similar to the average statewide experience of blacks who begin at other two-year colleges, but lower than the rates at traditionally black colleges; (3) of the 25 black students who enrolled at GC as first-time, full-time students during fall 1990 or fall 1991, 22 required remedial course work and only 1 received an associate degree within 3 years of arriving at GC; and (4) 88% of the seniors graduating from area high schools in 1993-94 were white, with the major feeder of black students to GC graduating only six black students in 1993-94. A review of the literature on topics related to the recruitment and retention of minority students is included. (KP)

In an effort to inform space allocation decisions, Gainesville College (GC) in Georgia, undertook a project to analyze classroom usage for fall 1995 and make projections to the year 2000 based on annual enrollment increases of 3%. Factors potentially affecting the use of space were determined to include the following: (1) conversion to the semester system, which would result in gains in classroom space since two three-credit hour courses can be held in a classroom on each day of the week; (2) increasing the number of course offerings during GC's Activity Period, a morning session used for campus-wide events; (3) construction of a chemistry wing; and (4) potential increases in average class size. An analysis conducted of classroom use in six GC buildings indicated that the Science and Technology and Humanities and Social Sciences buildings were at maximum capacity in fall 1995, while there were no apparent usage problems for the Mathematics, Foreign Language, and Business building; Music and Art building; gymnasium; or Performing Arts building. Finally, with respect to enrollment projections, its was determine that given GC's current space, the maximum number of additional students that the college could handle is 300 and that this limit would be reached by the year 2000. A memorandum regarding the effects of converting to the semester system is appended. (TGI)

The Fresh Start (FS) program was initiated at Gainesville College (GC), in Georgia, to target at-risk students and improve their overall academic performance and retention. The program provides immediate placement of students into remedial classes, academic support services, intrusive counseling, and long-term computerized tracking to assess student outcomes. A study was conducted to track academic outcomes for one academic year and determine any differences for the 68 FS students who entered in summer 1993 and 233 similarly at-risk students who started in fall 1990. Results of the analysis included the following: (1) while the two sample groups were similar for most characteristics, 50% of the 1990 sample were female, compared to 59% of the FS sample; (2) the FS group had more remedial education needs upon entering than the 1990 sample; (3) after one year, the FS students had an average grade point average of 2.22, compared to 1.70 for the 1990 students, while 19% (n=13) of the FS and 21% (n=49) of the 1990 group were candidates for academic probation; (4) 79% of the 1990 group returned for the subsequent quarter, while 69% of the FS students did; and (5) for both groups, less than 7% of the students who attempted academic credit courses received "A's," overall pass rates in the humanities were 11% better for FS students than the 1990 group. The results suggested that the FS program, as implemented, resulted in overall better academic outcomes for at-risk students. (KP)

A longitudinal study followed self-declared transfer students from time of entry at Gainesville College (GC) in fall 1988 through fall 1992, focusing on 4-year transfer rates to the University System of Georgia. Separate transfer rates were calculated for students who earned associate degrees from GC and for business administration majors, and students transferring to private and out-of-state institutions were excluded. Many GC students were in the high-risk category with respect to student persistence and transfer, as evidenced by the following student body characteristics: (1) a significant number of students attended GC part-time and worked while attending college; (2) in fall 1988, 33% of GC students were 22 or older, almost 20% were married, and many had young children; (3) a relatively large number of students entered GC without having taken the prescribed college preparatory curriculum (CPC) in high school, with 45% of the students entering with one or more CPC deficiencies in English, math, science, social science or foreign language; (4) one out of two students was placed in remedial courses; (5) many students were the first in their family to attend college; and (6) over 80% of the transfer students transferred to one of three senior colleges within commuting distance of GC. The transfer rate for all transfer majors was 35.7%; for students earning associate degrees, 74.5%; for business majors, 37.6%, for business majors with associate degrees, 81.6%. Transfer rates increased to 86.5% if calculated for associate degree transfers who earned 60 or more academic credits, and earned a "C" or better in Accounting II, Microeconomics, and Macroeconomics. (KP)

Two studies were performed to examine the transfer rates of first-time freshmen in the Business Administration (BA) major at a public two-year college to senior colleges or universities in the University System of Georgia (USG). The first study tracked students for 4 years, beginning in fall 1988, and the second study tracked students for 3 years, beginning in fall 1989. Students being tracked were divided into four groups: those who had declared BA as their major, those who had earned a BA associate degree over the 3- and 4-year study periods, and both of the previous groups further limited to those students who had completed 60 or more credit hours at a two-year college and had received a "C" or better in Accounting, Microeconomics, and Macroeconomics. Results of the study included: (1) transfer rates for BA majors were 37% for the 3-year study and 38% for the 4-year study; (2) within 3 years of starting at a two-year college, over 70% of BA majors with associate degrees had transferred to an institution in the USG, compared with 82% after 4 years; (3) 60% of BA students who had completed 60 or more units and had received a "C" or better in Accounting, Microeconomics, and Macroeconomics transferred after 3 years, and 68% had transferred after 4 years; and (4) BA graduates with a "C" or better average in their major classes reached a 4-year transfer rate of 87%. (Contains 14 references.) (MAB)