An Extended Kernel for Generalized Multiple-Instance Learning

The multiple-instance learning (MIL) model has been successful in areas such as drug discovery and content-based image-retrieval. Recently, this model was generalized and a corresponding kernel was introduced to learn generalized MIL concepts with a support vector machine. While this kernel enjoyed empirical success, it has limitations in its representation. We extend this kernel by enriching its representation and empirically evaluate our new kernel on data from content-based image retrieval, biological sequence analysis, and drug discovery. We found that our new kernel generalized noticeably better than the old one in content-based image retrieval and biological sequence analysis and was slightly better or even with the old kernel in the other applications, showing that an SVM using this kernel does not overfit despite its richer representation.

Paleoclimate Reconstruction Along The Pole-Equator-Pole Transect of the Americas (PEP 1)

Examples are presented of inter-hemispheric comparison of instrumental climate and paleoclimate proxy records from the Americas for different temporal scales. Despite a certain symmetry of seasonal precipitation patterns along the PEP 1 transect, decadal variability of winter precipitation shows different characteristics in terms of amplitude and frequency in both the last 100 and last 1000 years. Such differences in variability are also seen in a comparison of time series of different El Nino/Southern Oscillation proxy records from North and South America, however, these differences do not appear to affect the spatial correlation with Pacific sea surface temperature patterns. Local and regional differences in response to climate change are even more pronounced for records with lower temporal resolution, and inter-hemispheric synchroneity may or may not be indicative of the same forcing. This aspect is illustrated in an inter-hemispheric comparison of the last 1000 years of glacier variability, and of the full- and late-glacial lake level history.

Moving Under, Around, and Through the Miocene Reconstruction of Southern California

Modern southern California is fragmented by faults that juxtapose blocks with contrasting topographies and differing geologic histories. Many of the tectonic events that have shaped southern California were initiated during the Miocene, as subduction along the ancient trench margin off southern California was replaced by transform (strikeslip) faulting, such as that along the San Andreas fault.