We consider a learning setting of importance to large scale machine learning: potentially unlimited data arrives sequentially, but only a small fraction of it is labeled. The learner cannot store the data; it should learn from both labeled and unlabeled data, and it may also request labels for some of the unlabeled items. This setting is frequently encountered in real-world applications and has the characteristics of online, semi-supervised, and active learning. Yet previous learning models fail to consider these characteristics jointly. We present OASIS, a Bayesian model for this learning setting. The main contributions of the model include the novel integration of a semi-supervised likelihood function, a sequential Monte Carlo scheme for efficient online Bayesian updating, and a posterior-reduction criterion for active learning. Encouraging results on both synthetic and real-world optical character recognition data demonstrate the synergy of these characteristics in OASIS.

Metalle schleifen und polieren in der Antike – Alte Quellen und neue Experimente. Seit der Mensch seine Werkzeuge und Jagdwaffen herstellen kann, hat er sie auch geschliffen. Nach den ersten Werkstoffen Knochen, Geweih und Holz stellten später die Metalle besondere Anforderungen. Rohgüsse von Schmuck und Geräten sowie geschmiedete Werkzeuge und Waffen mussten überschliffen, Schneiden und Klingen geschärft und Oberflächen aus ästhetischen Gründen auch poliert werden. Welche Schleif- und Poliermittel die Handwerker der Antike hierfür verwendeten, ist kaum bekannt; die schriftlichen Quellen sind rar, und archäologische Befunde schwer zu erkennen und zu dokumentieren. Dieses Buch stellt zusammen, welche Abrasivmaterialien dem frühen Handwerk zur Verfügung standen, wo die Rohstoffe zu finden und wie die Materialien aufzubereiten sind. Die antiken Autoren, die diese Materialien erwähnten, werden zitiert, und mit Experimenteller Archäologie hat der Autor die Tauglichkeit und Effizienz aller Schleifgeräte, Schleifmittel und Polierstoffe getestet.

Fluctuations in the size of taurine cattle (Bos taurus) have been regularly demonstrated using archaeozoological data from across time and space in Europe, and have been linked to cultural, social and economic changes, but little is known about whether phenotypic change is accompanied by changes in genetic diversity. Here, we performed PCR-typed analysis of the partial mtDNA d-loop fragments of 99 cattle from the Neolithic to Early Medieval times from a number of different sites across Switzerland, combining newly presented data with previously published data (n = 20). We found that most cattle included (84) were members of the common European macro-haplogroup T3. However, cattle belonging to the haplogroups T1, T2, Q and P were identified as early as the Neolithic period, before 2690 cal. BCE. The highest diversity was found in the Neolithic period, during the 1st century CE and during the 7th–8th centuries CE. Bottleneck phases with low genetic diversity were detected during the Late Iron Age and from the fifth to the seventh century CE. Based on the FST values, Horgen, Corded Ware and cattle populations from the seventh to the ninth century CE were plotted away from the clusters of all other populations. The periods with larger-sized cattle correspond with those of high mtDNA d-loop diversity. Phenotype and genotype both appear to respond to the same socio-economic and cultural processes.

No abstract

No abstract