Object class recognition is an active topic in computer vision still presenting many challenges. In most approaches, this task is addressed by supervised learning algorithms that need a large quantity of labels to perform well. This leads either to small datasets (< 10, 000 images) that capture only a subset of the real-world class distribution (but with a controlled and verified labeling procedure), or to large datasets that are more representative but also add more label noise. Therefore, semi-supervised learning is a promising direction. It requires only few labels while simultaneously making use of the vast amount of images available today. We address object class recognition with semi-supervised learning. These algorithms depend on the underlying structure given by the data, the image description, and the similarity measure, and the quality of the labels. This insight leads to the main research questions of this thesis: “Is the structure given by labeled and unlabeled data more important than the algorithm itself?”, “Can we improve this neighborhood structure by a better similarity metric or with more representative unlabeled data?”, and “Is there a connection between the quality of labels and the overall performance and how can we get more representative labels?”. We answer all these questions, i.e., we provide an extensive evaluation, we propose several graph improvements, and we introduce a novel active learning framework to get more representative labels.

Category models for objects or activities typically rely on supervised learning requiring sufficiently large training sets. Transferring knowledge from known categories to novel classes with no or only a few labels is far less researched even though it is a common scenario. In this work, we extend transfer learning with semi-supervised learning to exploit unlabeled instances of (novel) categories with no or only a few labeled instances. Our proposed approach Propagated Semantic Transfer combines three techniques. First, we transfer information from known to novel categories by incorporating external knowledge, such as linguistic or expert-specified information, e.g., by a mid-level layer of semantic attributes. Second, we exploit the manifold structure of novel classes. More specifically we adapt a graph-based learning algorithm - so far only used for semi-supervised learning -to zero-shot and few-shot learning. Third, we improve the local neighborhood in such graph structures by replacing the raw feature-based representation with a mid-level object- or attribute-based representation. We evaluate our approach on three challenging datasets in two different applications, namely on Animals with Attributes and ImageNet for image classification and on MPII Composites for activity recognition. Our approach consistently outperforms state-of-the-art transfer and semi-supervised approaches on all datasets.

Object class recognition is an active topic in computer vision still presenting many challenges. In most approaches, this task is addressed by supervised learning algorithms that need a large quantity of labels to perform well. This leads either to small datasets (<10,000 images) that capture only a subset of the real-world class distribution (but with a controlled and verified labeling procedure), or to large datasets that are more representative but also add more label noise. Therefore, semi-supervised learning has been established as a promising direction to address object recognition. It requires only few labels while simultaneously making use of the vast amount of images available today. In this chapter, we outline the main challenges of semi-supervised object recognition, we review existing approaches, and we emphasize open issues that should be addressed next to advance this research topic.

Internet data sources provide us with large image datasets which are mostly without any explicit labeling. This setting is ideal for semi-supervised learning which seeks to exploit labeled data as well as a large pool of unlabeled data points to improve learning and classification. While we have made considerable progress on the theory and algorithms, we have seen limited success to translate such progress to the large scale datasets which these methods are inspired by. We investigate the computational complexity of popular graph-based semi-supervised learning algorithms together with different possible speed-ups. Our findings lead to a new algorithm that scales up to 40 times larger datasets in comparison to previous approaches and even increases the classification performance. Our method is based on the key insights that by employing a density-based measure unlabeled data points can be selected similar to an active learning scheme. This leads to a compact graph resulting in an improved performance up to 11.6% at reduced computational costs.

Popular visual representations like SIFT have shown broad applicability across many task. This great generality comes naturally with a lack of specificity when focusing on a particular task or a set of classes. Metric learning approaches have been proposed to tailor general purpose representations to the needs of more specific tasks and have shown strong improvements on visual matching and recognition benchmarks. However, the performance of metric learning depends strongly on the labels that are used for learning. Therefore, we propose to combine metric learning with an active sample selection strategy in order to find labels that are representative for each class as well as improve the class separation of the learnt metric. We analyze several active sample selection strategies in terms of exploration and exploitation trade-offs. Our novel scheme achieves on three different datasets up to 10

Active learning aims to reduce the amount of labels required for classification. The main difficulty is to find a good trade-off between exploration and exploitation of the labeling process that depends — among other things — on the classification task, the distribution of the data and the employed classification scheme. In this paper, we analyze different sampling criteria including a novel density-based criteria and demonstrate the importance to combine exploration and exploitation sampling criteria. We also show that a time-varying combination of sampling criteria often improves performance. Finally, by formulating the criteria selection as a Markov decision process, we propose a novel feedback-driven framework based on reinforcement learning. Our method does not require prior information on the dataset or the sampling criteria but rather is able to adapt the sampling strategy during the learning process by experience. We evaluate our approach on three challenging object recognition datasets and show superior performance to previous active learning methods.

This paper considers scalable and unobtrusive activity recognition using on-body sensing for context awareness in wearable computing. Common methods for activity recognition rely on supervised learning requiring substantial amounts of labeled training data. Obtaining accurate and detailed annotations of activities is challenging, preventing the applicability of these approaches in real-world settings. This paper proposes new annotation strategies that substantially reduce the required amount of annotation. We explore two learning schemes for activity recognition that effectively leverage such sparsely labeled data together with more easily obtainable unlabeled data. Experimental results on two public data sets indicate that both approaches obtain results close to fully supervised techniques. The proposed methods are robust to the presence of erroneous labels occurring in real-world annotation data.

Lyme neuroborreliosis (LNB) is the most frequent tick-borne infectious disease of the central nervous system. In acute LNB and the rare chronic state of infection, patients can experience cognitive deficits such as attention and memory disturbances. During LNB, single compounds of Borrelia burgdorferi sensu lato are released into the subarachnoid space. To investigate the pathogenesis of neurologic dysfunction in LNB, we determined that the outer surface protein C (OspC), a major virulence factor of B. burgdorferi, stimulated mouse microglial cells in a dose-dependent manner to release nitric oxide (EC50 = 0.24 mg/L) in vitro. To mimic pathophysiologic conditions of long-term release of this bacterial component in vivo, we treated C57BL/6 mice with recombinant OspC from Borrelia garinii or buffer by intraventricular infusion and tested them for behavioral deficits. After 4weeks, brains were examined by routine histology and immunohistochemistry. Assessment of spatial learning and memory of treated mice during OspC exposure did not reveal significant differences from controls. Continuous exposure to intrathecal B. burgdorferi OspC led to activation of microglia and axonal damage without demonstrable cognitive impairment in experimental mice. These results suggest that long-term intrathecal exposure to OspC resulted in axonal damage that may underlie the neurologic manifestations in chronic LNB.

Graph-based methods are very popular in semi-supervised learning due to their well founded theoretical background, intuitive interpretation of local neighborhood structure, and strong performance on a wide range of challenging learning problems. However, the success of these methods is highly dependent on the pre-existing neighborhood structure in the data used to construct the graph. In this paper, we use metric learning to improve this critical step by increasing the precision of the nearest neighbors and building our graph in this new metric space. We show that learning of neighborhood relations before constructing the graph consistently improves performance of two label propagation schemes on three different datasets – achieving the best performance reported on Caltech 101 to date. Furthermore, we question the predominant random draw of labels and advocate the importance of the choice of labeled examples. Orthogonal to active learning schemes, we investigate how domain knowledge can substantially increase performance in these semi-supervised learning settings.

An 85-year-old man with myasthenia gravis was successfully treated with methotrexate (10 mg/week), pyridostigmine and prednisolone (0-30 mg/day) for over 10 years. Then, he developed dysphagia and lost weight. Gastroscopy revealed Candida esophagitis. The patient received nystatin for 2 weeks. Methotrexate was stopped, and immunosuppressive therapy was continued with prednisolone alone. The patient has now remained in good condition for over 1 year. Although dysphagia is a typical symptom of myasthenia gravis, swallowing disturbances should not be attributed hastily to this disease, since they may also be a complication of therapy.

Many computer vision methods rely on annotated image sets without taking advantage of the increasing number of unlabeled images available. This paper explores an alternative approach involving unsupervised structure discovery and semi-supervised learning (SSL) in image collections. Focusing on object classes, the first part of the paper contributes with an extensive evaluation of state-of-the-art image representations. Thus, it underlines the decisive influence of the local neighborhood structure and its direct consequences on SSL results and the importance of developing powerful object representations. In a second part, we propose and explore promising directions to improve results by looking at the local topology between images and feature combination strategies.

The analysis of rare mitochondrial transcripts derived from the L-strand of the mitochondrial genome requires a sensitive method such as the S1-nuclease protection assay. We examined whether the ribosomal mitochon­drial transcript 16S is suitable as an internal standard in a multiplex S1-nuclease protection assay for the measurement of different mitochondrial transcripts. For reliable quantification of rare mitochondrial transcripts with the RNase protection assay, a minimum of 2 μg of total RNA is necessary. Standard curves of 16S RNA produced with total RNA from human kidney, liver, brain, and a human neuroblastoma cell line (SH-SY5Y) revealed dose-response relationships that were saturated already at less than 0.5 μg of total RNA. Therefore, 16S is inappropriate as an internal standard for analyzing mitochondrial transcripts with RNase protection assays when more than 0.5 μg of total RNA have to be analyzed.

Stressor presence during the last weeks of gestation has been associated with behavioral disorders in later life. In this study we support further research on the long term effects of prenatal stress on Swiss mice descendant's behavior. Prenatal stress procedure consisted on restraining the dams under bright light for 45 min, three times per day from the 15th day of pregnancy, until birth. After weaning, offspring's motor performance and spontaneous exploratory behavior were measured by the tight-rope and T-maze tests, respectively. We also evaluated anxiety behavior using elevated plus maze test. We found that maternal stress improves the performance of the animals in the tight rope test and that this effect was sex and age dependent: prenatal stressed males obtained the best scores during the first month of life, while in females the same was achieved at the second month. Spontaneous exploratory behavior analysis revealed that it was elevated in prenatal stressed males and that this effect persisted on time. However, we did not find significant differences on this behavioral response among both females groups. Finally, differences on anxiety behavior were found only in females: prenatally stressed animals showed a higher proportion of entries into the open arms of a plus maze (reduced anxiety) compared to the control group. Our results show that prenatal stress modifies the normal behavior of the progeny: prenatal stressed animals have a better performance in the carried out test. These notably results suggest the existence of an adaptive response to prenatal stress.

Stimulation of murine primary microglia with Toll-like receptor (TLR) agonists enhances their ability to phagocytose and kill bacteria. Here we show that the viral TLR3 agonist poly(I:C) stimulates the release of cyto-/chemokines and nitric oxide by microglia. Poly(I:C) increases microglial phagocytosis and intracellular killing of Escherichia coli K1, a pathogenic encapsulated bacterial strain, after 30 and 90 min of co-incubation. Stimulation with a viral epitope may strengthen the resistance of the brain to bacterial infections in vivo. Our data encourage animal experiments with poly(I:C) derivatives to assess whether this approach can increase the resistance of the CNS against bacterial infections.

Severe bacterial infections such as sepsis and meningitis still kill or severely injure people despite the use of bactericidal antibiotics. Therefore, new strategies for a better therapy are needed. Activin A, a member of the TGF-β superfamily and its binding protein follistatin (FS) are released by various cell types during acute and chronic inflammatory processes. Until now, a clear definition of conditions in which activin A exerts either its pro- or anti-inflammatory functions is lacking. The activin/FS-system participates in the fine-tuning of the host's inflammatory response upon infectious stimuli. This response is on the one hand necessary for fighting pathogens, but on the other hand can negatively affect the host. This article focuses on the role of activin A and FS in infection and after acute inflammatory stimuli. The therapeutic potentials of blocking or promoting activin actions are discussed.