Hierarchical multi-label classification using local neural networks

Hierarchical multi-label classification is a complex classification task where the classes involved in the problem are hierarchically structured and each example may simultaneously belong to more than one class in each hierarchical level. In this paper, we extend our previous works, where we investigated a new local-based classification method that incrementally trains a multi-layer perceptron for each level of the classification hierarchy. Predictions made by a neural network in a given level are used as inputs to the neural network responsible for the prediction in the next level. We compare the proposed method with one state-of-the-art decision-tree induction method and two decision-tree induction methods, using several hierarchical multi-label classification datasets. We perform a thorough experimental analysis, showing that our method obtains competitive results to a robust global method regarding both precision and recall evaluation measures.

Methodologies for Synthesizable Programmable Devices based on Multi-Stage Switching Networks

Nowadays the rise of non-recurring engineering (NRE) costs associated with complexity is becoming a major factor in SoC design, limiting both scaling opportunities and the flexibility advantages offered by the integration of complex computational units. The introduction of embedded programmable elements can represent an appealing solution, able both to guarantee the desired flexibility and upgradabilty and to widen the SoC market. In particular embedded FPGA (eFPGA) cores can provide bit-level optimization for those applications which benefits from synthesis, paying on the other side in terms of performance penalties and area overhead with respect to standard cell ASIC implementations. In this scenario this thesis proposes a design methodology for a synthesizable programmable device designed to be embedded in a SoC. A soft-core embedded FPGA (eFPGA) is hence presented and analyzed in terms of the opportunities given by a fully synthesizable approach, following an implementation flow based on Standard-Cell methodology. A key point of the proposed eFPGA template is that it adopts a Multi-Stage Switching Network (MSSN) as the foundation of the programmable interconnects, since it can be efficiently synthesized and optimized through a standard cell based implementation flow, ensuring at the same time an intrinsic congestion-free network topology. The evaluation of the flexibility potentialities of the eFPGA has been performed using different technology libraries (STMicroelectronics CMOS 65nm and BCD9s 0.11μm) through a design space exploration in terms of area-speed-leakage tradeoffs, enabled by the full synthesizability of the template. Since the most relevant disadvantage of the adopted soft approach, compared to a hardcore, is represented by a performance overhead increase, the eFPGA analysis has been made targeting small area budgets. The generation of the configuration bitstream has been obtained thanks to the implementation of a custom CAD flow environment, and has allowed functional verification and performance evaluation through an application-aware analysis.

A smart multi-hop hierarchical routing protocol for efficient video communication over wireless multimedia sensor networks

For smart applications, nodes in wireless multimedia sensor networks (MWSNs) have to take decisions based on sensed scalar physical measurements. A routing protocol must provide the multimedia delivery with quality level support and be energy-efficient for large-scale networks. With this goal in mind, this paper proposes a smart Multi-hop hierarchical routing protocol for Efficient VIdeo communication (MEVI). MEVI combines an opportunistic scheme to create clusters, a cross-layer solution to select routes based on network conditions, and a smart solution to trigger multimedia transmission according to sensed data. Simulations were conducted to show the benefits of MEVI compared with the well-known Low-Energy Adaptive Clustering Hierarchy (LEACH) protocol. This paper includes an analysis of the signaling overhead, energy-efficiency, and video quality.

Efficient Monte Carlo optimization for multi-label classifier chains

Multi-label classification (MLC) is the supervised learning problem where an instance may be associated with multiple labels. Modeling dependencies between labels allows MLC methods to improve their performance at the expense of an increased computational cost. In this paper we focus on the classifier chains (CC) approach for modeling dependencies. On the one hand, the original CC algorithm makes a greedy approximation, and is fast but tends to propagate errors down the chain. On the other hand, a recent Bayes-optimal method improves the performance, but is computationally intractable in practice. Here we present a novel double-Monte Carlo scheme (M2CC), both for finding a good chain sequence and performing efficient inference. The M2CC algorithm remains tractable for high-dimensional data sets and obtains the best overall accuracy, as shown on several real data sets with input dimension as high as 1449 and up to 103 labels.

Expressive power of binary relevance and chain classifiers based on Bayesian Networks for multi-label classification

Bayesian network classifiers are widely used in machine learning because they intuitively represent causal relations. Multi-label classification problems require each instance to be assigned a subset of a defined set of h labels. This problem is equivalent to finding a multi-valued decision function that predicts a vector of h binary classes. In this paper we obtain the decision boundaries of two widely used Bayesian network approaches for building multi-label classifiers: Multi-label Bayesian network classifiers built using the binary relevance method and Bayesian network chain classifiers. We extend our previous single-label results to multi-label chain classifiers, and we prove that, as expected, chain classifiers provide a more expressive model than the binary relevance method.

Multi-dimensional classification of GABAergic interneurons with Bayesian network-modeled label uncertainty

Abstract Interneuron classification is an important and long-debated topic in neuroscience. A recent study provided a data set of digitally reconstructed interneurons classified by 42 leading neuroscientists according to a pragmatic classification scheme composed of five categorical variables, namely, of the interneuron type and four features of axonal morphology. From this data set we now learned a model which can classify interneurons, on the basis of their axonal morphometric parameters, into these five descriptive variables simultaneously. Because of differences in opinion among the neuroscientists, especially regarding neuronal type, for many interneurons we lacked a unique, agreed-upon classification, which we could use to guide model learning. Instead, we guided model learning with a probability distribution over the neuronal type and the axonal features, obtained, for each interneuron, from the neuroscientists’ classification choices. We conveniently encoded such probability distributions with Bayesian networks, calling them label Bayesian networks (LBNs), and developed a method to predict them. This method predicts an LBN by forming a probabilistic consensus among the LBNs of the interneurons most similar to the one being classified. We used 18 axonal morphometric parameters as predictor variables, 13 of which we introduce in this paper as quantitative counterparts to the categorical axonal features. We were able to accurately predict interneuronal LBNs. Furthermore, when extracting crisp (i.e., non-probabilistic) predictions from the predicted LBNs, our method outperformed related work on interneuron classification. Our results indicate that our method is adequate for multi-dimensional classification of interneurons with probabilistic labels. Moreover, the introduced morphometric parameters are good predictors of interneuron type and the four features of axonal morphology and thus may serve as objective counterparts to the subjective, categorical axonal features.

Multi-dimensional classification of GABAergic interneurons with Bayesian network-modeled label uncertainty

Interneuron classification is an important and long-debated topic in neuroscience. A recent study provided a data set of digitally reconstructed interneurons classified by 42 leading neuroscientists according to a pragmatic classification scheme composed of five categorical variables, namely, of the interneuron type and four features of axonal morphology. From this data set we now learned a model which can classify interneurons, on the basis of their axonal morphometric parameters, into these five descriptive variables simultaneously. Because of differences in opinion among the neuroscientists, especially regarding neuronal type, for many interneurons we lacked a unique, agreed-upon classification, which we could use to guide model learning. Instead, we guided model learning with a probability distribution over the neuronal type and the axonal features, obtained, for each interneuron, from the neuroscientists’ classification choices. We conveniently encoded such probability distributions with Bayesian networks, calling them label Bayesian networks (LBNs), and developed a method to predict them. This method predicts an LBN by forming a probabilistic consensus among the LBNs of the interneurons most similar to the one being classified. We used 18 axonal morphometric parameters as predictor variables, 13 of which we introduce in this paper as quantitative counterparts to the categorical axonal features. We were able to accurately predict interneuronal LBNs. Furthermore, when extracting crisp (i.e., non-probabilistic) predictions from the predicted LBNs, our method outperformed related work on interneuron classification. Our results indicate that our method is adequate for multi-dimensional classification of interneurons with probabilistic labels. Moreover, the introduced morphometric parameters are good predictors of interneuron type and the four features of axonal morphology and thus may serve as objective counterparts to the subjective, categorical axonal features.

Lightweight peer-to-peer secure multi-party VoIP protocol

Multi-party voice-over-IP (MVoIP) services provide economical and convenient group communication mechanisms for many emerging applications such as distance collaboration systems, on-line meetings and Internet gaming. In this paper, we present a light peer-to-peer (P2P) protocol to provide MVoIP services on small platforms like mobile phones and PDAs. Unlike other proposals, our solution is fully distributed and self-organizing without requiring specialized servers or IP multicast support.

Improving kNN multi-label classification in Prototype Selection scenarios using class proposals

Prototype Selection (PS) algorithms allow a faster Nearest Neighbor classification by keeping only the most profitable prototypes of the training set. In turn, these schemes typically lower the performance accuracy. In this work a new strategy for multi-label classifications tasks is proposed to solve this accuracy drop without the need of using all the training set. For that, given a new instance, the PS algorithm is used as a fast recommender system which retrieves the most likely classes. Then, the actual classification is performed only considering the prototypes from the initial training set belonging to the suggested classes. Results show that this strategy provides a large set of trade-off solutions which fills the gap between PS-based classification efficiency and conventional kNN accuracy. Furthermore, this scheme is not only able to, at best, reach the performance of conventional kNN with barely a third of distances computed, but it does also outperform the latter in noisy scenarios, proving to be a much more robust approach.

Hierarchical Multi-Label Text Classification in a Low-Resource Setting

In this thesis we address a multi-label hierarchical text classification problem in a low-resource setting and explore different approaches to identify the best one for our case. The goal is to train a model that classifies English school exercises according to a hierarchical taxonomy with few labeled data. The experiments made in this work employ different machine learning models and text representation techniques: CatBoost with tf-idf features, classifiers based on pre-trained models (mBERT, LASER), and SetFit, a framework for few-shot text classification. SetFit proved to be the most promising approach, achieving better performance when during training only a few labeled examples per class are available. However, this thesis does not consider all the hierarchical taxonomy, but only the first two levels: to address classification with the classes at the third level further experiments should be carried out, exploring methods for zero-shot text classification, data augmentation, and strategies to exploit the hierarchical structure of the taxonomy during training.

Establishment of a protocol for obtention of neuronal stem cells lineages from the dog olfactory epithelium

A morphological and cell culture study from nasal mucosa of dogs was performed in order to establish a protocol to obtain a cell population committed to neuronal lineage, as a proposal for the treatment of traumatic and degenerative lesions in these animals, so that in the future these results could be applied to the human species. Twelve mongrel dogs of 60-day aged pregnancy were collected from urban pound dogs in São Paulo. Tissue from cribriform ethmoidal lamina of the fetuses was collected at necropsy under sterile conditions around 1h to 2h postmortem by uterine sections and sections from the fetal regions described above. Isolated cells of this tissue were added in DMEM/F-12 medium under standard conditions of incubation (5% CO², >37ºC). Cell culture based on isolated cells from biopsies of the olfactory epithelium showed rapid growth when cultured for 24 hours, showing phase-bright sphere cells found floating around the fragments, attached on culture flasks. After 20 days, a specific type of cells, predominantly ellipsoids or fusiform cells was characterized in vitro. The indirect immunofluorescence examination showed cells expressing markers of neuronal precursors (GFAP, neurofilament, oligodendrocyte, and III â-tubulin). The cell proliferation index showed Ki67 immunostaining with a trend to label cell groups throughout the apical region, while PCNA immunostaining label predominantly cell groups lying above the basal lamina. The transmission electron microscopy from the olfactory epithelium of dogs revealed cells with electron-dense cytoplasm and preserving the same distribution as those of positive cell staining for PCNA. Metabolic activity was confirmed by presence of euchromatin in the greatest part of cells. All these aspects give subsidies to support the hypothesis about resident progenitor cells among the basal cells of the olfactory epithelium, committed to renewal of these cell populations, especially neurons.

Comparison Between Single-Diode Low-Level Laser Therapy (LLLT) and LED Multi-Diode (Cluster) Therapy (LEDT) Applications Before High-Intensity Exercise

Background Data and Objective: There is anecdotal evidence that low-level laser therapy (LLLT) may affect the development of muscular fatigue, minor muscle damage, and recovery after heavy exercises. Although manufacturers claim that cluster probes (LEDT) maybe more effective than single-diode lasers in clinical settings, there is a lack of head-to-head comparisons in controlled trials. This study was designed to compare the effect of single-diode LLLT and cluster LEDT before heavy exercise. Materials and Methods: This was a randomized, placebo-controlled, double-blind cross-over study. Young male volleyball players (n = 8) were enrolled and asked to perform three Wingate cycle tests after 4 x 30 sec LLLT or LEDT pretreatment of the rectus femoris muscle with either (1) an active LEDT cluster-probe (660/850 nm, 10/30mW), (2) a placebo cluster-probe with no output, and (3) a single-diode 810-nm 200-mW laser. Results: The active LEDT group had significantly decreased post-exercise creatine kinase (CK) levels (-18.88 +/- 41.48U/L), compared to the placebo cluster group (26.88 +/- 15.18U/L) (p < 0.05) and the active single-diode laser group (43.38 +/- 32.90U/L) (p<0.01). None of the pre-exercise LLLT or LEDT protocols enhanced performance on the Wingate tests or reduced post-exercise blood lactate levels. However, a non-significant tendency toward lower post-exercise blood lactate levels in the treated groups should be explored further. Conclusion: In this experimental set-up, only the active LEDT probe decreased post-exercise CK levels after the Wingate cycle test. Neither performance nor blood lactate levels were significantly affected by this protocol of pre-exercise LEDT or LLLT.

Comparison of the Salmonella/Microsome Microsuspension Assay with the new Microplate Fluctuation Protocol for Testing the Mutagenicity of Environmental Samples

The objective of this study was to compare the responses of the Salmonella/microsome microsuspension assay with the new microplate fluctuation protocol (MPF) for the evaluation of the mutagenic activity of environmental samples. Organic extracts of total particulate atmospheric air samples, surface waters, and effluents were tested in dose-response experiments. The assays were performed with strain TA98 in the absence and presence of S9 mix. Both protocols produced similar results, despite the fact that the maximum score of the MPF is limited to 48 wells, whereas in the regular plate assay it is possible to count up to 1,500 colonies using an automatic counter. Similar sensitivities based on the lowest dose that resulted in a positive response were obtained for both assays. The MPF procedure is less laborious (e.g., all-liquid format, use of multi-channel pipettors) and allows for automation of the pipetting and dispensing steps, thus, reducing time of the analysis which is particularly important in environmental quality monitoring programs or in effect-directed analysis. The results show that the MPF procedure is a promising tool to test environmental samples for mutagenic activity. Environ. Mol. Mutagen. 51:31-38, 2010. (C) 2009 Wiley-Liss, Inc.

Additive Intraocular Pressure Reduction Effect of Fixed Combination of Maleate Timolol 0.5%/Dorzolamide 2% (Cosopt) on Monotherapy With Latanoprost (Xalatan) in Patients With Elevated Intraocular Pressure A Prospective, 4-week, Open-label, Randomized, Controlled Clinical Trial

Purpose: To evaluate the additive effect of dorzolamide/timolol fixed combination in patients under monotherapy with latanoprost. Patients and Methods: In this prospective, 4-week, randomized, open-label controlled clinical trial, patients with open-angle glaucoma or ocular hypertension, which presented at least 15% intraocular pressure (IOP) reduction after a minimum period of 15 days of monotherapy with latanoprost and whose IOP level was considered above the established target-IOP level were randomized to receive fixed combination of timolol/dorzolamide twice daily in one of eyes. The fellow eye was kept under monotherapy and was included in the control group. A modified diurnal tension curve (mDTC) followed by the water drinking test were performed in the baseline and week 4 visits to evaluate IOP profile between groups. Results: Forty-nine per-protocol patients were analyzed. After latanoprost monotherapy run-in period, IOP levels were significantly reduced (P<0.001) in both control and study groups to 15.34 +/- 2.96 mm Hg and 15.24 +/- 2.84 mm Hg (30.8% and 32.2% IOP reduction, respectively; P=0.552). At week 4, mean baseline diurnal IOP levels were 15.60 +/- 3.09 and 14.44 +/- 3.03 (7.4% difference; P=0.01). Mean baseline IOP modified diurnal tension curve peak after latanoprost run-in period were 17.47 +/- 3.68 mm Hg and 17.02 +/- 3.35 mm Hg (control and study eyes, respectively; P=0.530). At week 4 visit, mean water-drinking test peaks were significantly reduced in the study eye group in comparison with the control group: 19.02 +/- 3.81 mm Hg and 20.39 +/- 4.19 mm Hg, respectively (6.7% reduction; P=0.039). Conclusions: In our sample, dorzolamide 2%/timolol 0.5% fixed combination as add-on therapy in patients with open-angle glaucoma or ocular hypertension under monotherapy with latanoprost with IOP already in mid-teens levels may further enhance pressure reduction.