In order to detect new events, a system must support on-line learning, adapting to pattern dynamic characteristics. Studies of such adaptation have originated the novelty detection area, which aims at identifying unexpected or unknown patterns. These researches have motivated this work to propose the on-line and unsupervised Self-Organizing Novelty Detection (SONDE) neural network. In this network, the creation of new neurons points out novelties. Experiments evaluated the influence of SONDE parameters and their capability to detect novelty events. These evaluations considered the datasets Biomed, ALL-AML Leukemia and DLBCL. Results are compared to others from GWR.
Breast cancer accounts for the highest number of female deaths worldwide. Early detection of the disease is essential to increase the chances of treatment and cure of patients. Infrared thermography has emerged as a promising technique for diagnosis of the disease due to its low cost and that it does not emit harmful radiation, and it gives good results when applied in young women. This work uses convolutional neural networks in a database of 440 infrared images of 88 patients, classifying them into two classes: normal and pathology. During the training of the networks, we use transfer learning of the following convolutional neural network architectures: AlexNet, GoogLeNet, ResNet-18, VGG-16, and VGG-19. Our results show the great potential of using deep learning techniques combined with infrared images in the aid of breast cancer diagnosis.
Estimating the correct respiratory rate (RR) is an essential technique for intensive care units, hospitals, geriatric hospital facilities, and home care services. Capnography is a standard methodology used to monitor carbon dioxide concentrations or partial pressures of respiratory gases to provide the most accurate RR measurements. However, it is inconvenient to use and has been primarily used while administering anesthesia and during intensive care. Many researchers now use electrocardiogram signals to estimate RR. Despite the recent developments, the current hospital environments suffer from inaccurate respiratory monitoring. While various machine learning techniques, including deep learning, have recently been applied to the medical processing sector, only a few studies have been conducted in the field of RR estimation. Therefore, using photoplethysmography, machine-learning techniques such as the ensemble gradient boosting algorithm are being employed in RR estimation. Multi-phases are used based on various feature extraction and selection methodology to improve the performance for RR estimation. In this study, the number of ensembles is increased, and the proposed ensemble methodology is effectively learned to estimate the RR. The proposed ensemble-based gradient boosting algorithm are compared with those of ensemble-based long-short memory network, and ensemble-based supported vector regression techniques, 3.30 breaths per min (bpm), 4.82 bpm and 5.83 bpm based on mean absolute errors. The proposed method shows a more accurate estimate of the respiration rate.
Behavior studies have been conducted by scientists and philosophers who approach subjects such as star and planet trajectories, society organizations, living beings evolution and human language. With the advent of computer, new challenges have been observed in order to explore and understand the behavior variations of interactions with systems. Motivated by those challenges, this work proposes a new approach to automatically cluster, detect and identify behavior patterns. In order to validate this approach, we have modeled the knowledge embedded in interactions of handwriting signatures. The generated knowledge models were, afterwards, employed to verify signatures. Obtained results were compared to other related approaches presented in SVC2004, the First International Signature Verification Competition.
Machine learning is a field of artificial intelligence which aims at developing techniques to automatically transfer human knowledge into analytical models. Recently, those techniques have been applied to time series with unknown dynamics and fluctuations in the established behavior patterns, such as humancomputer interaction, inspection robotics and climate change. In order to detect novelties in those time series, techniques are required to learn and update knowledge structures, adapting themselves to data tendencies. The learning and updating process should integrate and accommodate novelty events into the normal behavior model, possibly incurring the revaluation of long-term memories. This sort of application has been addressed by the proposal of incremental techniques based on unsupervised neural networks and regression techniques. Such proposals have introduced two new concepts in time-series novelty detection. The first defines the temporal novelty, which indicates the occurrence of unexpected series of events. The second measures how novel a single event is, based on the historical knowledge. However, current studies do not fully consider both concepts of detecting and quantifying temporal novelties. This motivated the proposal of the self-organizing novelty detection neural network architecture (SONDE) which incrementally learns patterns in order to represent unknown dynamics and fluctuation of established behavior. The knowledge accumulated by SONDE is employed to estimate Markov chains which model causal relationships. This architecture is applied to detect and measure temporal and nontemporal novelties. The evaluation of the proposed technique is carried out through simulations and experiments, which have presented promising results.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.