The formation and invariance of the canine nose pattern is studied. Nose images of ten beagle dogs were collected for ten months from month two to month eleven. The nose patterns in these images are examined visually and by a biometric algorithm. It is found that the canine nose pattern is fully formed at the end of the second month since birth and remains invariant until the end of the eleventh month. This study also strongly indicates that the canine nose pattern can be used as a unique biometric marker for each individual dog.

The uniqueness of the canine nose pattern was studied. A total of 180 nose images of 60 dogs of diverse age, gender, and breed were collected. The canine nose patterns in these images were examined visually and by a biometric algorithm. It was found that the canine nose pattern remains invariant regardless of when the image is taken; and that the canine nose pattern is indeed unique to each dog. The same study was also performed on an enlarged dataset of 278 nose images of 70 dogs of 19 breeds. The study of the enlarged dataset also leads to the same conclusion. The result of this paper confirms and enhances the claims of earlier works by others that the canine nose pattern is indeed unique to each animal and serves as a unique biometric marker.

This paper proposes the continuous semantic topic embedding model (CSTEM) which finds latent topic variables in documents using continuous semantic distance function between the topics and the words by means of the variational autoencoder(VAE). The semantic distance could be represented by any symmetric bell-shaped geometric distance function on the Euclidean space, for which the Mahalanobis distance is used in this paper. In order for the semantic distance to perform more properly, we newly introduce an additional model parameter for each word to take out the global factor from this distance indicating how likely it occurs regardless of its topic. It certainly improves the problem that the Gaussian distribution which is used in previous topic model with continuous word embedding could not explain the semantic relation correctly and helps to obtain the higher topic coherence. Through the experiments with the dataset of 20 Newsgroup, NIPS papers and CNN/Dailymail corpus, the performance of the recent state-of-the-art models is accomplished by our model as well as generating topic embedding vectors which makes possible to observe where the topic vectors are embedded with the word vectors in the real Euclidean space and how the topics are related each other semantically.