Whales and Sonar: Environmental Exemptions for the Navy's Mid-Frequency Active Sonar Training.

Mid-frequency active (MFA) sonar emits pulses of sound from an underwater transmitter to help determine the size, distance, and speed of objects. The sound waves bounce off objects and reflect back to underwater acoustic receivers as an echo. MFA sonar has been used since World War II, and the Navy indicates it is the only reliable way to track submarines, especially more recently designed submarines that operate more quietly, making them more difficult to detect. Scientists have asserted that sonar may harm certain marine mammals under certain conditions, especially beaked whales. Depending on the exposure, they believe that sonar may damage the ears of the mammals, causing hemorrhaging and/or disorientation. The Navy agrees that the sonar may harm some marine mammals, but says it has taken protective measures so that animals are not harmed. (PDF contains 20 pages)

Does training policy help to attract, retain and develop valuable human resources? Analysis from the Mondragon case

This is an electronic version of the accepted paper in the journal:Advances in the Economic Analysis of Participatory and Labor-Managed Firms. Volumen. 12

Effects of 12 weeks muscle hipetrophy training program in a young male athlete: stydy case

PURPOSE: The main goals of the present study were: 1) to review some recommendations about how to increase lean body mass; 2) to analyse whether following scientific sources of current recommendations, visible changes can be shown or not in a participant (body composition, strength and blood analyses). METHODS: One male athlete completed 12 weeks of resistance training program and following a diet protocol. Some test were determined such as, strength 6RM, blood analyses, skindfold measurements, body perimeters and impedance test. Body composition measurements were taken 3 times during the program (before-T1, after 6 weeks of intervention period-T2 and at the end of the program-T3). On the other hand, strength tests and blood analyses were performed twice (before and after the program). RESULTS: Strength was increased in general; blood analyses showed that Creatine kinase was increased a 104% and Triglycerides level was decreased a 22.5%; in the impedance test, body mass (1.6%), lean body mass (3.5%) and Body mass index (1.7%) were increased, whereas fat mass was decreased (15.5%); relaxed and contracted biceps perimeters were also increased. CONCLUSION: A muscle hypertrophy training program mixed with an appropriate diet during 12 weeks leads to interesting adaptations related to increase in body weight, lean body mass, biceps perimeters, strength and creatine kinase levels, and a decrease in fat mass.

The development of the South Carolina Coastal Information Network and portal site: Bringing training opportunities and educational resources to coastal community officials and the public

The South Carolina Coastal Information Network (SCCIN) emerged as a result of a number of coastal outreach institutions working in partnership to enhance coordination of the coastal community outreach efforts in South Carolina. This organized effort, led by the S.C. Sea Grant Consortium and its Extension Program, includes partners from federal and state agencies, regional government agencies, and private organizations seeking to coordinate and/or jointly deliver outreach programs that target coastal community constituents. The Network was officially formed in 2006 with the original intention of fostering intra-and inter- agency communication, coordination, and cooperation. Network partners include the S.C. Sea Grant Consortium, S.C. Department of Health and Environmental Control – Office of Ocean and Coastal Resource Management and Bureau of Water, S.C. Department of Natural Resources – ACE Basin National Estuarine Research Reserve, North Inlet-Winyah Bay National Estuarine Research Reserve, Clemson University Cooperative Extension Service and Carolina Clear, Berkeley-Charleston-Dorchester Council of Governments, Waccamaw Regional Council of Governments, Urban Land Institute of South Carolina, S.C. Department of Archives and History, the National Oceanic and Atmospheric Administration – Coastal Services Center and Hollings Marine Laboratory, Michaux Conservancy, Ashley-Cooper Stormwater Education Consortium, the Coastal Waccamaw Stormwater Education Consortium, the S.C. Chapter of the U.S. Green Building Council, and the Lowcountry Council of Governments. (PDF contains 3 pages)

A short training program improves the accuracy of portion-size estimates in future dietitians

[EN] The objective of this study was to determine whether a short training program, using real foods, would decreased their portion-size estimation errors after training. 90 student volunteers (20.18±0.44 y old) of the University of the Basque Country (Spain) were trained in observational techniques and tested in food-weight estimation during and after a 3-hour training period. The program included 57 commonly consumed foods that represent a variety of forms (125 different shapes). Estimates of food weight were compared with actual weights. Effectiveness of training was determined by examining change in the absolute percentage error for all observers and over all foods over time. Data were analyzed using SPSS vs. 13.0. The portion-size errors decreased after training for most of the foods. Additionally, the accuracy of their estimates clearly varies by food group and forms. Amorphous was the food type estimated least accurately both before and after training. Our findings suggest that future dietitians can be trained to estimate quantities by direct observation across a wide range of foods. However this training may have been too brief for participants to fully assimilate the application.

Optimal Data Distributions in Machine Learning

In the first part of the thesis we explore three fundamental questions that arise naturally when we conceive a machine learning scenario where the training and test distributions can differ. Contrary to conventional wisdom, we show that in fact mismatched training and test distribution can yield better out-of-sample performance. This optimal performance can be obtained by training with the dual distribution. This optimal training distribution depends on the test distribution set by the problem, but not on the target function that we want to learn. We show how to obtain this distribution in both discrete and continuous input spaces, as well as how to approximate it in a practical scenario. Benefits of using this distribution are exemplified in both synthetic and real data sets.

In order to apply the dual distribution in the supervised learning scenario where the training data set is fixed, it is necessary to use weights to make the sample appear as if it came from the dual distribution. We explore the negative effect that weighting a sample can have. The theoretical decomposition of the use of weights regarding its effect on the out-of-sample error is easy to understand but not actionable in practice, as the quantities involved cannot be computed. Hence, we propose the Targeted Weighting algorithm that determines if, for a given set of weights, the out-of-sample performance will improve or not in a practical setting. This is necessary as the setting assumes there are no labeled points distributed according to the test distribution, only unlabeled samples.

Finally, we propose a new class of matching algorithms that can be used to match the training set to a desired distribution, such as the dual distribution (or the test distribution). These algorithms can be applied to very large datasets, and we show how they lead to improved performance in a large real dataset such as the Netflix dataset. Their computational complexity is the main reason for their advantage over previous algorithms proposed in the covariate shift literature.

In the second part of the thesis we apply Machine Learning to the problem of behavior recognition. We develop a specific behavior classifier to study fly aggression, and we develop a system that allows analyzing behavior in videos of animals, with minimal supervision. The system, which we call CUBA (Caltech Unsupervised Behavior Analysis), allows detecting movemes, actions, and stories from time series describing the position of animals in videos. The method summarizes the data, as well as it provides biologists with a mathematical tool to test new hypotheses. Other benefits of CUBA include finding classifiers for specific behaviors without the need for annotation, as well as providing means to discriminate groups of animals, for example, according to their genetic line.