Felipe Leno da Silva

Evaluating classification and feature selection techniques for honeybee subspecies identification using wing images

We have compared the performance of 7 classifiers to bee species identification.The use of feature selection improves the performance of classification in this domain.The best classifier to Apis mellifera subspecies identification is the Naive Bayes.We have outlined the importance of Apis mellifera to agriculture. The main pollinator commercially available, i.e. Apis mellifera, is now facing a severe population decrease worldwide due to the so-called Colony Collapse Disorder. Measures to preserve this species are urgent. Honeybees inhabit several different environments, from swamps to deserts, from high mountains to the African savannah. They are classified into several different subspecies, each one adapted to a particular set of environmental characteristics. The identification of subspecies is based on morphometric features from the entire bee body, but in the last years features from the fore wings have proven to be very efficient for classification. Several methods have been developed to perform the automatic classification through images of bee wings, and geometric morphometrics has been reported to achieve good results in terms of consumed time and reliability of the results. However, there has been no study evaluating the impact of feature selection and new classification methods on the identification performance. We here evaluate seven combinations of feature selectors and classifiers by their hit ratio with real bee wing images. Feature selection proved to be beneficial to all the evaluated combinations and the Naive Bayes classifier combined with a correlation-based feature selector achieved the best results. These conclusions can benefit researches that rely on classification by geometric morphometrics features, both for bees and for other animal species.

Graft-versus-host disease after liver transplantation

Graft-versus-host disease (GVHD) following liver trans-plantation (LT) is an uncommon complication but has highmortality and represents a major diagnostic challenge.GVHD occurs when immunocompetent donor lymphocytesoriginating from the transplanted liver undergo activationand clonal expansion, allowing them to mount a destructivecellular immune response against recipient tissues.Humoral GVHD is usually seen after an ABO-mismatchedliver transplant, but cellular GVHD is directed against themajor histocompatibility complex and often results in severemultisystem disease with high mortality.

Performance Evaluation of Cryptographic Algorithms over IoT Platforms and Operating Systems

The deployment of security services over Wireless Sensor Networks (WSN) and IoT devices brings significant processing and energy consumption overheads. These overheads are mainly determined by algorithmic efficiency, quality of implementation, and operating system. Benchmarks of symmetric primitives exist in the literature for WSN platforms but they are mostly focused on single platforms or single operating systems. Moreover, they are not up to date with respect to implementations and/or operating systems versions which had significant progress. Herein, we provide time and energy benchmarks of reference implementations for different platforms and operating systems and analyze their impact. Moreover, we not only give the first benchmark results of symmetric cryptography for the Intel Edison IoT platform but also describe a methodology of how to measure energy consumption on that platform.

MOO-MDP: An Object-Oriented Representation for Cooperative Multiagent Reinforcement Learning

Reinforcement learning (RL) is a widely known technique to enable autonomous learning. Even though RL methods achieved successes in increasingly large and complex problems, scaling solutions remains a challenge. One way to simplify (and consequently accelerate) learning is to exploit regularities in a domain, which allows generalization and reduction of the learning space. While object-oriented Markov decision processes (OO-MDPs) provide such generalization opportunities, we argue that the learning process may be further simplified by dividing the workload of tasks amongst multiple agents, solving problems as multiagent systems (MAS). In this paper, we propose a novel combination of OO-MDP and MAS, called multiagent OO-MDP (MOO-MDP). Our proposal accrues the benefits of both OO-MDP and MAS, better addressing scalability issues. We formalize the general model MOO-MDP and present an algorithm to solve deterministic cooperative MOO-MDPs. We show that our algorithm learns optimal policies while reducing the learning space by exploiting state abstractions. We experimentally compare our results with earlier approaches in three domains and evaluate the advantages of our approach in sample efficiency and memory requirements.

Towards Knowledge Transfer in Deep Reinforcement Learning

Driven by recent developments in the area of Artificial Intelligence research, a promising new technology for building intelligent agents has evolved. The technology is termed Deep Reinforcement Learning (DRL) and combines the classic field of Reinforcement Learning (RL) with the representational power of modern Deep Learning approaches. DRL enables solutions for difficult and high dimensional tasks, such as Atari game playing, for which previously proposed RL methods were inadequate. However, these new solution approaches still take a long time to learn how to actuate in such domains and so far are mainly researched for single task scenarios. The ability to generalize gathered knowledge and transfer it to another task has been researched for classical RL, but remains an open problem for the DRL domain. Consequently, in this article we evaluate under which conditions the application of Transfer Learning (TL) to the DRL domain improves the learning of a new task. Our results indicate that TL can greatly accelerate DRL when transferring knowledge from similar tasks, and that the similarity between tasks plays a key role in the success or failure of knowledge transfer.

The best approach for diagnosing primary sclerosing cholangitis

Primary sclerosing cholangitis (PSC) is a chronic chole-static syndrome with autoimmune features and is associatedwith other immunological diseases such as autoimmunepancreatitis and inflammatory bowel disease. In addition toprimary biliary cirrhosis, PSC is one of the most commonchronic cholestatic liver diseases.

Autonomously Reusing Knowledge in Multiagent Reinforcement Learning.

Autonomous agents are increasingly required to solve complex tasks; hard-coding behaviors has become infeasible. Hence, agents must learn how to solve tasks via interactions with the environment. In many cases, knowledge reuse will be a core technology to keep training times reasonable, and for that, agents must be able to autonomously and consistently reuse knowledge from multiple sources, including both their own previous internal knowledge and from other agents. In this paper, we provide a literature review of methods for knowledge reuse in Multiagent Reinforcement Learning. We define an important challenge problem for the AI community, survey the existent methods, and discuss how they can all contribute to this challenging problem. Moreover, we highlight gaps in the current literature, motivating “low-hanging fruit” for those interested in the area. Our ambition is that this paper will encourage the community to work on this difficult and relevant research challenge.

Pairwise registration in indoor environments using adaptive combination of 2D and 3D cues

Abstract Pairwise frame registration of indoor scenes with sparse 2D local features is not particularly robust under varying lighting conditions or low visual texture. In this case, the use of 3D local features can be a solution, as such attributes come from the 3D points themselves and are resistant to visual texture and illumination variations. However, they also hamper the registration task in cases where the scene has little geometric structure. Frameworks that use both types of features have been proposed, but they do not take into account the type of scene to better explore the use of 2D or 3D features. Because varying conditions are inevitable in real indoor scenes, we propose a new framework to improve pairwise registration of consecutive frames using an adaptive combination of sparse 2D and 3D features. In our proposal, the proportion of 2D and 3D features used in the registration is automatically defined according to the levels of geometric structure and visual texture contained in each scene. The effectiveness of our proposed framework is demonstrated by experimental results from challenging scenarios with datasets including unrestricted RGB-D camera motion in indoor environments and natural changes in illumination.

Object-Oriented Reinforcement Learning in Cooperative Multiagent Domains

Although Reinforcement Learning methods have successfully been applied to increasingly large problems, scalability remains a central issue. While Object-Oriented Markov Decision Processes (OO-MDP) are used to exploit regularities in a domain, Multiagent System (MAS) methods are used to divide workload amongst multiple agents. In this work we propose a novel combination of OO-MDP and MAS, called Multiagent Object-Oriented Markov Decision Process (MOO-MDP), so as to accrue the benefits of both strategies and be able to better address scalability issues. We present an algorithm to solve deterministic cooperative MOO-MDPs, and prove that it learns optimal policies while reducing the learning space by exploiting state abstractions. We experimentally compare our results with earlier approaches and show advantages with regard to discounted cumulative reward, number of steps to fulfill the task, and Q-table size.