Semi-supervised learning of WLAN radio maps

In this thesis a manifold learning method is applied to the problem of WLAN positioning and automatic radio map creation. Due to the nature of WLAN signal strength measurements, a signal map created from raw measurements results in non-linear distance relations between measurement points. These signal strength vectors reside in a high-dimensioned coordinate system. With the help of the so called Isomap-algorithm the dimensionality of this map can be reduced, and thus more easily processed. By embedding position-labeled strategic key points, we can automatically adjust the mapping to match the surveyed environment. The environment is thus learned in a semi-supervised way; gathering training points and embedding them in a two-dimensional manifold gives us a rough mapping of the measured environment. After a calibration phase, where the labeled key points in the training data are used to associate coordinates in the manifold representation with geographical locations, we can perform positioning using the adjusted map. This can be achieved through a traditional supervised learning process, which in our case is a simple nearest neighbors matching of a sampled signal strength vector. We deployed this system in two locations in the Kumpula campus in Helsinki, Finland. Results indicate that positioning based on the learned radio map can achieve good accuracy, especially in hallways or other areas in the environment where the WLAN signal is constrained by obstacles such as walls.

Genetically Engineered Oncolytic Adenoviruses for the Treatment of Kidney and Breast Cancer

Metastatic kidney and breast cancer are devastating diseases currently lacking efficient treatment options. One promising developmental approach in cancer treatment are oncolytic adenoviruses, which have demonstrated excellent safety in many clinical trials. However, antitumor efficacy needs to be improved in order to make oncolytic viruses a viable treatment alternative. To be able to follow oncolytic virus replication in vivo, we set up a non-invasive imaging system based on coinjection of a replication deficient luciferase expressing virus and a replication competent virus. The system was validated in vitro and in vivo and used in other projects of the thesis. In another study we showed that capsid modifications on adenoviruses result in enhanced gene transfer and increased oncolytic effect on renal cancer cells in vitro. Moreover, capsid modified oncolytic adenoviruses demonstrated significantly improved antitumor efficacy in murine kidney cancer models. To transcriptionally target kidney cancer tissue we evaluated two hypoxia response elements for their usability as tissue specific promoters using a novel dual luciferase imaging system. Based on the results of the promoter evaluation and the studies on capsid modifications, we constructed a transcriptionally and transductionally targeted oncolytic adenovirus armed with an antiangiogenic transgene for enhanced renal cell cancer specificity and improved antitumor efficacy. This virus exhibited kidney cancer specific replication and significantly improved antitumor effect in a murine model of intraperitoneal disseminated renal cell cancer. Cancer stem cells are thought to be resistant to conventional cancer drugs and might play an important role in breast cancer relapse and the formation of metastasis. Therefore, we examined if capsid modified oncolytic adenoviruses are able to kill these cells proposed to be breast cancer initiating. Efficient oncolytic effect and significant antitumor efficacy on tumors established with breast cancer initiating cells was observed, suggesting that oncolytic adenoviruses might be able to prevent breast cancer relapse and could be used in the treatment of metastatic disease. In conclusion, the results presented in this thesis suggest that genetically engineered oncolytic adenoviruses have great potential in the treatment of metastatic kidney and breast cancer.