Assessing the Role of General Chemistry Learning in Higher Education

The inclusion of General Chemistry (GC) in the curricula of higher education courses in science and technology aims, on the one hand, to develop students' skills necessary for further studies and, on the other hand, to respond to the need of endowing future professionals of knowledge to analyze and solve multidisciplinary problems in a sustainable way. The participation of students in the evaluation of the role played by the GC in their training is crucial, and the analysis of the results can be an essential tool to increase success in the education of students and improving practices in various professions. Undeniably, this work will be focused on the development of an intelligent system to assess the role of GC. The computational framework is built on top of a Logic Programming approach to Knowledge Representation and Reasoning, complemented with a problem solving methodology moored on Artificial Neural Networks. The results so far obtained show that the proposed model stands for a good start, being its overall accuracy higher than 95%.

Mutual Information-based RBM Neural Networks

(Deep) neural networks are increasingly being used for various computer vision and pattern recognition tasks due to their strong ability to learn highly discriminative features. However, quantitative analysis of their classication ability and design philosophies are still nebulous. In this work, we use information theory to analyze the concatenated restricted Boltzmann machines (RBMs) and propose a mutual information-based RBM neural networks (MI-RBM). We develop a novel pretraining algorithm to maximize the mutual information between RBMs. Extensive experimental results on various classication tasks show the eectiveness of the proposed approach.

Assessment of naproxen adsorption on bone char in aqueous solutions using batch and fixed-bed processes

An integrated analysis of naproxen adsorption on bone char in batch and packed-bed column conditions has been performed. Kinetic, thermodynamic and breakthrough parameters have been calculated using adsorption models and artificial neural networks. Results show that naproxen removal using bone char in batch conditions is a feasible and effective process, which could involve electrostatic and non-electrostatic interactions depending mainly on pH conditions. However, the application of packed-bed column for naproxen adsorption on bone char is not effective for the treatment of diluted solutions due to the low degree of adsorbent utilization (below 4%) at tested operating conditions. The proposed mechanism for naproxen removal using bone char could include a complexation process via phosphate and naproxen, hydrogen bonding and the possibility of hydrophobic interactions via π–π electron. This study highlights the relevance of performing an integrated analysis of adsorbent effectiveness in batch and dynamic conditions to establish the best process configuration for the removal of emerging water pollutants such as pharmaceuticals.

Adverse drug reaction classification with deep neural networks

We study the problem of detecting sentences describing adverse drug reactions (ADRs) and frame the problem as binary classification. We investigate different neural network (NN) architectures for ADR classification. In particular, we propose two new neural network models, Convolutional Recurrent Neural Network (CRNN) by concatenating convolutional neural networks with recurrent neural networks, and Convolutional Neural Network with Attention (CNNA) by adding attention weights into convolutional neural networks. We evaluate various NN architectures on a Twitter dataset containing informal language and an Adverse Drug Effects (ADE) dataset constructed by sampling from MEDLINE case reports. Experimental results show that all the NN architectures outperform the traditional maximum entropy classifiers trained from n-grams with different weighting strategies considerably on both datasets. On the Twitter dataset, all the NN architectures perform similarly. But on the ADE dataset, CNN performs better than other more complex CNN variants. Nevertheless, CNNA allows the visualisation of attention weights of words when making classification decisions and hence is more appropriate for the extraction of word subsequences describing ADRs.

Quality of Judgment Assessment

The intersection of Artificial Intelligence and The Law stands for a multifaceted matter, and its effects set the advances on culture, organization, as well as the social matters, when the emergent information technologies are taken into consideration. From this point of view, the weight of formal and informal Conflict Resolution settings should be highlighted, and the use of defective data, information or knowledge must be emphasized. Indeed, it is hard to do it with traditional problem solving methodologies. Therefore, in this work the focus is on the development of decision support systems, in terms of its knowledge representation and reasoning procedures, under a formal framework based on Logic Programming, complemented with an approach to computing centered on Artificial Neural Networks. It is intended to evaluate the Quality-of-Judgments and the respective Degree-of-Confidence that one has on such happenings.

An Assessment to Toxicological Risk of Pesticide Exposure

On the one hand, pesticides may be absorbed into the body orally, dermally, ocularly and by inhalation and the human exposure may be dietary, recreational and/or occupational where toxicity could be acute or chronic. On the other hand, the environmental fate and toxicity of the pesticide is contingent on the physico-chemical characteristics of pesticide, the soil composition and adsorption. Human toxicity is also dependent on the exposure time and individual’s susceptibility. Therefore, this work will focus on the development of an Artificial Intelligence based diagnosis support system to assess the pesticide toxicological risk to humanoid, built under a formal framework based on Logic Programming to knowledge representation and reasoning, complemented with an approach to computing grounded on Artificial Neural Networks. The proposed solution is unique in itself, once it caters for the explicit treatment of incomplete, unknown, or even self-contradictory information, either in terms of a qualitative or quantitative setting.

Design, realization, and characterization of complex reaction networks in dynamic mechanically interlocked molecules

The experimental projects discussed in this thesis are all related to the field of artificial molecular machines, specifically to systems composed of pseudorotaxane and rotaxane architectures. The characterization of the peculiar properties of these mechano-molecules is frequently associated with the analysis and elucidation of complex reaction networks; this latter aspect represents the main focus and central thread tying my thesis work. In each chapter, a specific project is described as summarized below: the focus of the first chapter is the realization and characterization of a prototype model of a photoactivated molecular transporter based on a pseudorotaxane architecture; in the second chapter is reported the design, synthesis, and characterization of a [2]rotaxane endowed with a dibenzylammonium station and a novel photochromic unit that acts as a recognition site for a DB24C8 crown ether macrocycle; in the last chapter is described the synthesis and characterization of a [3]rotaxane in which the relative number of rings and stations can be changed on command.

Interpretable Convolutional Neural Networks for Decoding and Analyzing Neural Time Series Data

Machine learning is widely adopted to decode multi-variate neural time series, including electroencephalographic (EEG) and single-cell recordings. Recent solutions based on deep learning (DL) outperformed traditional decoders by automatically extracting relevant discriminative features from raw or minimally pre-processed signals. Convolutional Neural Networks (CNNs) have been successfully applied to EEG and are the most common DL-based EEG decoders in the state-of-the-art (SOA). However, the current research is affected by some limitations. SOA CNNs for EEG decoding usually exploit deep and heavy structures with the risk of overfitting small datasets, and architectures are often defined empirically. Furthermore, CNNs are mainly validated by designing within-subject decoders. Crucially, the automatically learned features mainly remain unexplored; conversely, interpreting these features may be of great value to use decoders also as analysis tools, highlighting neural signatures underlying the different decoded brain or behavioral states in a data-driven way. Lastly, SOA DL-based algorithms used to decode single-cell recordings rely on more complex, slower to train and less interpretable networks than CNNs, and the use of CNNs with these signals has not been investigated. This PhD research addresses the previous limitations, with reference to P300 and motor decoding from EEG, and motor decoding from single-neuron activity. CNNs were designed light, compact, and interpretable. Moreover, multiple training strategies were adopted, including transfer learning, which could reduce training times promoting the application of CNNs in practice. Furthermore, CNN-based EEG analyses were proposed to study neural features in the spatial, temporal and frequency domains, and proved to better highlight and enhance relevant neural features related to P300 and motor states than canonical EEG analyses. Remarkably, these analyses could be used, in perspective, to design novel EEG biomarkers for neurological or neurodevelopmental disorders. Lastly, CNNs were developed to decode single-neuron activity, providing a better compromise between performance and model complexity.

High-resolution stratigraphy and seismic site characterization of late Quaternary paleovalley systems (Adriatic coastal plain, Italy)

In highly urbanized coastal lowlands, effective site characterization is crucial for assessing seismic risk. It requires a comprehensive stratigraphic analysis of the shallow subsurface, coupled with the precise assessment of the geophysical properties of buried deposits. In this context, late Quaternary paleovalley systems, shallowly buried fluvial incisions formed during the Late Pleistocene sea-level fall and filled during the Holocene sea-level rise, are crucial for understanding seismic amplification due to their soft sediment infill and sharp lithologic contrasts. In this research, we conducted high-resolution stratigraphic analyses of two regions, the Pescara and Manfredonia areas along the Adriatic coastline of Italy, to delineate the geometries and facies architecture of two paleovalley systems. Furthermore, we carried out geophysical investigations to characterize the study areas and perform seismic response analyses. We tested the microtremor-based horizontal-to-vertical spectral ratio as a mapping tool to reconstruct the buried paleovalley geometries. We evaluated the relationship between geological and geophysical data and identified the stratigraphic surfaces responsible for the observed resonances. To perform seismic response analysis of the Pescara paleovalley system, we integrated the stratigraphic framework with microtremor and shear wave velocity measurements. The seismic response analysis highlights strong seismic amplifications in frequency ranges that can interact with a wide variety of building types. Additionally, we explored the applicability of artificial intelligence in performing facies analysis from borehole images. We used a robust dataset of high-resolution digital images from continuous sediment cores of Holocene age to outline a novel, deep-learning-based approach for performing automatic semantic segmentation directly on core images, leveraging the power of convolutional neural networks. We propose an automated model to rapidly characterize sediment cores, reproducing the sedimentologist's interpretation, and providing guidance for stratigraphic correlation and subsurface reconstructions.

Augmented Audiovisual Translation: the Perks and Perils of the Implementation of Artificial Intelligence in Subtitling and Dubbing

This thesis examines the state of audiovisual translation (AVT) in the aftermath of the COVID-19 emergency, highlighting new trends with regards to the implementation of AI technologies as well as their strengths, constraints, and ethical implications. It starts with an overview of the current AVT landscape, focusing on future projections about its evolution and its critical aspects such as the worsening working conditions lamented by AVT professionals – especially freelancers – in recent years and how they might be affected by the advent of AI technologies in the industry. The second chapter delves into the history and development of three AI technologies which are used in combination with neural machine translation in automatic AVT tools: automatic speech recognition, speech synthesis and deepfakes (voice cloning and visual deepfakes for lip syncing), including real examples of start-up companies that utilize them – or are planning to do so – to localize audiovisual content automatically or semi-automatically. The third chapter explores the many ethical concerns around these innovative technologies, which extend far beyond the field of translation; at the same time, it attempts to revindicate their potential to bring about immense progress in terms of accessibility and international cooperation, provided that their use is properly regulated. Lastly, the fourth chapter describes two experiments, testing the efficacy of the currently available tools for automatic subtitling and automatic dubbing respectively, in order to take a closer look at their perks and limitations compared to more traditional approaches. This analysis aims to help discerning legitimate concerns from unfounded speculations with regards to the AI technologies which are entering the field of AVT; the intention behind it is to humbly suggest a constructive and optimistic view of the technological transformations that appear to be underway, whilst also acknowledging their potential risks.

Predicting corticospinal excitability from pre-TMS EEG using convolutional neural networks

The amplitude of motor evoked potentials (MEPs) elicited by transcranial magnetic stimulation (TMS) of the primary motor cortex (M1) shows a large variability from trial to trial, although MEPs are evoked by the same repeated stimulus. A multitude of factors is believed to influence MEP amplitudes, such as cortical, spinal and motor excitability state. The goal of this work is to explore to which degree the variation in MEP amplitudes can be explained by the cortical state right before the stimulation. Specifically, we analyzed a dataset acquired on eleven healthy subjects comprising, for each subject, 840 single TMS pulses applied to the left M1 during acquisition of electroencephalography (EEG) and electromyography (EMG). An interpretable convolutional neural network, named SincEEGNet, was utilized to discriminate between low- and high-corticospinal excitability trials, defined according to the MEP amplitude, using in input the pre-TMS EEG. This data-driven approach enabled considering multiple brain locations and frequency bands without any a priori selection. Post-hoc interpretation techniques were adopted to enhance interpretation by identifying the more relevant EEG features for the classification. Results show that individualized classifiers successfully discriminated between low and high M1 excitability states in all participants. Outcomes of the interpretation methods suggest the importance of the electrodes situated over the TMS stimulation site, as well as the relevance of the temporal samples of the input EEG closer to the stimulation time. This novel decoding method allows causal investigation of the cortical excitability state, which may be relevant for personalizing and increasing the efficacy of therapeutic brain-state dependent brain stimulation (for example in patients affected by Parkinson’s disease).

Aplicação de métodos de calibração multivariada para a determinação simultânea de riboflavina (VB2), tiamina (VB1), piridoxina (VB6) e nicotinamida (VPP)

In this work, the artificial neural networks (ANN) and partial least squares (PLS) regression were applied to UV spectral data for quantitative determination of thiamin hydrochloride (VB1), riboflavin phosphate (VB2), pyridoxine hydrochloride (VB6) and nicotinamide (VPP) in pharmaceutical samples. For calibration purposes, commercial samples in 0.2 mol L-1 acetate buffer (pH 4.0) were employed as standards. The concentration ranges used in the calibration step were: 0.1 - 7.5 mg L-1 for VB1, 0.1 - 3.0 mg L-1 for VB2, 0.1 - 3.0 mg L-1 for VB6 and 0.4 - 30.0 mg L-1 for VPP. From the results it is possible to verify that both methods can be successfully applied for these determinations. The similar error values were obtained by using neural network or PLS methods. The proposed methodology is simple, rapid and can be easily used in quality control laboratories.

Engagement of multifocal neural circuits during recall of autobiographical happy events

Happy emotional states have not been extensively explored in functional magnetic resonance imaging studies using autobiographic recall paradigms. We investigated the brain circuitry engaged during induction of happiness by standardized script-driven autobiographical recall in 11 healthy subjects (6 males), aged 32.4 ± 7.2 years, without physical or psychiatric disorders, selected according to their ability to vividly recall personal experiences. Blood oxygen level-dependent (BOLD) changes were recorded during auditory presentation of personal scripts of happiness, neutral content and negative emotional content (irritability). The same uniform structure was used for the cueing narratives of both emotionally salient and neutral conditions, in order to decrease the variability of findings. In the happiness relative to the neutral condition, there was an increased BOLD signal in the left dorsal prefrontal cortex and anterior insula, thalamus bilaterally, left hypothalamus, left anterior cingulate gyrus, and midportions of the left middle temporal gyrus (P < 0.05, corrected for multiple comparisons). Relative to the irritability condition, the happiness condition showed increased activity in the left insula, thalamus and hypothalamus, and in anterior and midportions of the inferior and middle temporal gyri bilaterally (P < 0.05, corrected), varying in size between 13 and 64 voxels. Findings of happiness-related increased activity in prefrontal and subcortical regions extend the results of previous functional imaging studies of autobiographical recall. The BOLD signal changes identified reflect general aspects of emotional processing, emotional control, and the processing of sensory and bodily signals associated with internally generated feelings of happiness. These results reinforce the notion that happiness induction engages a wide network of brain regions.

Pitfalls of artificial grouping and stratification of scientific journals based on their Impact Factor: a case study in Brazilian Zoology

The present contribution explores the impact of the QUALIS metric system for academic evaluation implemented by CAPES (Coordination for the Development of Personnel in Higher Education) upon Brazilian Zoological research. The QUALIS system is based on the grouping and ranking of scientific journals according to their Impact Factor (IF). We examined two main points implied by this system, namely: 1) its reliability as a guideline for authors; 2) if Zoology possesses the same publication profile as Botany and Oceanography, three fields of knowledge grouped by CAPES under the subarea "BOZ" for purposes of evaluation. Additionally, we tested CAPES' recent suggestion that the area of Ecology would represent a fourth field of research compatible with the former three. Our results indicate that this system of classification is inappropriate as a guideline for publication improvement, with approximately one third of the journals changing their strata between years. We also demonstrate that the citation profile of Zoology is distinct from those of Botany and Oceanography. Finally, we show that Ecology shows an IF that is significantly different from those of Botany, Oceanography, and Zoology, and that grouping these fields together would be particularly detrimental to Zoology. We conclude that the use of only one parameter of analysis for the stratification of journals, i.e., the Impact Factor calculated for a comparatively small number of journals, fails to evaluate with accuracy the pattern of publication present in Zoology, Botany, and Oceanography. While such simplified procedure might appeals to our sense of objectivity, it dismisses any real attempt to evaluate with clarity the merit embedded in at least three very distinct aspects of scientific practice, namely: productivity, quality, and specificity.

The Missing Part of Seed Dispersal Networks: Structure and Robustness of Bat-Fruit Interactions

Mutualistic networks are crucial to the maintenance of ecosystem services. Unfortunately, what we know about seed dispersal networks is based only on bird-fruit interactions. Therefore, we aimed at filling part of this gap by investigating bat-fruit networks. It is known from population studies that: (i) some bat species depend more on fruits than others, and (ii) that some specialized frugivorous bats prefer particular plant genera. We tested whether those preferences affected the structure and robustness of the whole network and the functional roles of species. Nine bat-fruit datasets from the literature were analyzed and all networks showed lower complementary specialization (H(2)' = 0.3760.10, mean 6 SD) and similar nestedness (NODF = 0.5660.12) than pollination networks. All networks were modular (M=0.32 +/- 0.07), and had on average four cohesive subgroups (modules) of tightly connected bats and plants. The composition of those modules followed the genus-genus associations observed at population level (Artibeus-Ficus, Carollia-Piper, and Sturnira-Solanum), although a few of those plant genera were dispersed also by other bats. Bat-fruit networks showed high robustness to simulated cumulative removals of both bats (R = 0.55 +/- 0.10) and plants (R = 0.68 +/- 0.09). Primary frugivores interacted with a larger proportion of the plants available and also occupied more central positions; furthermore, their extinction caused larger changes in network structure. We conclude that bat-fruit networks are highly cohesive and robust mutualistic systems, in which redundancy is high within modules, although modules are complementary to each other. Dietary specialization seems to be an important structuring factor that affects the topology, the guild structure and functional roles in bat-fruit networks.