Letteratura scientifica selezionata sul tema "Frugal AI"

Recent evidence suggests that the take-the-best heuristic—flagship of “fast and frugal heuristics” research program—might in fact not be as frugal as tallying, which is considered to be a more complex strategy. Characterizing a simple decision strategy has always seemed straightforward, and the debate around the simplicity of the take-the-best heuristic is mostly concerned with a proper specification of the heuristic. I argue that the predominate conceptions of “simplicity” and “frugality” need to be revised. To this end, a number of recent behavioral and neuroscientific results are discussed. The example of take-the-best heuristic serves as an entry point to a foundational debate on bounded agency. I argue that the fast and frugal heuristics needs to question some of its legacy from the classical AI research. For example, the assumption that the bottleneck of decision-making process is information processing due to its serial nature. These commitments are hard to reconcile with the modern neuroscientific view of a human decision-maker. In addition, I discuss an overlooked source of uncertainty, namely neural noise, and compare a generic heuristic model to a similar neural algorithm.

O cálcio é considerado um nutriente indispensável para o crescimento e desenvolvimento ósseo, além de participar das funções biológicas do organismo. Estudos têm mostrado a relação entre o consumo adequado de cálcio na infância com a prevenção de doenças como osteoporose, hipertensão arterial e obesidade na vida adulta, justificando-se a importância da avaliação da ingestão desse nutriente em crianças. Investigar a relação entre fatores sociodemográficos e o consumo de cálcio em crianças a partir da introdução da alimentação complementar. Estudo de coorte, com 226 crianças, realizado no período de 2011 a 2013 no município de Viçosa-MG. Foi aplicado ao responsável um questionário semiestruturado para obtenção de informações referentes à situação sociodemográfica da criança. O consumo alimentar foi avaliado por meio de um recordatório de 24 horas com o responsável pela criança no sexto, nono ou décimo segundo mês de vida. Os dados do consumo alimentar foram avaliados segundo as Dietary Reference Intakes (DRI). Utilizou-se AI (Adequate Intake) como referência para o cálcio. Considerou-se para a classificação econômica o modelo proposto pela Associação Brasileira de Estudos Populacionais (ABEP, 2010), o qual leva em consideração itens de posse do individuo e grau de instrução do chefe da família. Os indivíduos são divididos nas classes: A, B, C, D e E, sendo que a classe econômica A é a que possui uma maior renda familiar. As análises foram realizadas no software Stata versão 10.0. Na análise bivariada dos fatores associados ao consumo de cálcio abaixo da AI, estimou-se a razão de prevalência e intervalo de confiança pela regressão de Poisson com variância robusta, sendo as variáveis com valor de p<0,20 incluídas no modelo múltiplo. A prevalência de consumo de cálcio abaixo da recomendação foi de 27,0%. Observou-se que crianças classificadas na categoria C, D e E (RP: 0,77; IC95%: 0,64 – 0,91) e do sexo feminino (RP: 0,79; IC95%: 0,65 – 0,95) apresentaram menor risco de consumo de cálcio abaixo da recomendação. Crianças pertencentes a famílias com mais de 4 pessoas no domicílio (RP: 1,21; IC95%: 1,001 – 1,46) tiveram uma prevalência 21% vezes maior do consumo de cálcio abaixo da AI. Encontrou-se uma maior adequação do consumo de cálcio em crianças do sexo feminino e cuja família se encontrava na categoria C, D e E. Uma possível explicação se deva ao fato do maior consumo de mingaus e preparações à base de leite nos níveis socioeconômicos inferiores, principalmente nas pequenas refeições, em substituição ao consumo de frutas. Entretanto, as crianças residentes em domicílios com mais que 4 pessoas apresentaram inadequado consumo de cálcio.

RESUMO Objetivo: Avaliar a composição e o valor nutricional dos lanches intermediários das crianças escolares brasileiras. Metodologia: Análise secundária de questionário estruturado, sobre o tipo e quantidade dos alimentos consumidos nos lanches por um período de 3 dias, respondido pelas mães de 2.365 crianças de 7 a 11 anos de todas as regiões do Brasil. Resultados: 97,46% das crianças realizaram os lanches intermediários que, em média, foram compostos por 2,9 grupos de alimentos. Os lanches da manhã (LM) foram caracterizados por biscoitos, frutas e iogurtes. Nos lanches da tarde (LT), que geralmente se mostraram mais calóricos, o grupo dos iogurtes foi substituído pelos pães em geral. Em relação aos nutrientes, observou-se um consumo expressivo de açúcar de adição, com destaque para as regiões Centro-Oeste e Norte + Nordeste, cujo consumo total (LM + LT) de 28,3 gramas ultrapassou o limite diário preconizado pela OMS. Outro importante achado foi o consumo sódio, cujo consumo nas crianças da amostra Brasil, nos dois lanches, atingiu 65% do valor de AI em todos os lanches, sendo maior no LT. Conclusão: As crianças escolares brasileiras têm o hábito de realizar os lanches intermediários, o que favorece o preenchimento dos requerimentos nutricionais, principalmente de vitaminas e fibras. Porém, a composição dos lanches revelou um consumo expressivo de açúcar de adição e sódio, o que pode impactar negativamente na saúde das crianças. Isso reforça a necessidade de campanhas educativas tanto às crianças quanto aos responsáveis por sua alimentação.

RIASSUNTO Una "lingua madre" parlata principalmente dai padri L'articolo passa in rassegna quello che si conosce sull'esperanto come lingua di uso domestico e come prima lingua. Casi documentati di famiglie che usavano l'esperanto risalgono al 1919. Dagli anni '70 la loro visibilità nel movimento esperantista è aumentata; un ente di coordinamento, Rondo Familia, è stato formato nel 1995. In quasi tutte le circa 350 famiglie attualmente documentate, l'esperanto è parlato ai bambini dal padre. Coppie con una origine etnicolinguistica mista sono solo circa un terzo di tutte le coppie. Le informazioni a disposizione suggeriscono che il "bilinguismo artificiale" praticato dai due terzi rimanenti delle coppie puö avere altrettanto successo del "bilinguismo naturale". Il parallelo linguistico più vicino sembra essere quello delle prime fasi di rinascita dell'ebraico. RESUMO "Patrina lingvo " parolata precipe de patroj La artikolo prezentas superrigardon pri Esperanto kiel familia kaj denaska lingvo. Kazoj de Esperanto-familioj estas dokumentitaj ekde la jaro 1919. Ekde la 70-aj jaroj ili akiras pli da atento en la Esperanto-movado; kunordiga instanco, Rondo Familia, fondigis en 1995. En preskaŭ ĉiu familio el la proksimume 350 aktuale registritaj, temas pri la patro kiu parolas Esperanton al la infanoj. Miksetnaj paroj konsistigas nur trionon de la konataj kazoj. La haveblaj informoj indikas ke la "artefarita dulingveco" pratikata de la ceteraj du trionoj estas same sukcespova kiel la "natura dulingveco". La plej trafa lingvistika komparo sajnas esti la fruaj stadioj de la revivigo de la hebrea lingvo.

A alimentação da população brasileira apresenta o baixo consumo de frutas e hortaliças, pouca variedade de espécies alimentícias, além do elevado consumo de alimentos ultraprocessados, o que contribui para o desenvolvimento de doenças crônicas não transmissíveis. O desenvolvimento das habilidades culinárias resgata a cultura alimentar e promove a alimentação saudável e sustentável. Este trabalho relata a experiência de uma oficina culinária com a hortaliça chaya (Cnidoscolus aconitifolius) como principal ingrediente, realizada na V Jornada Universitária em Defesa da Reforma Agrária na UFRJ (2018) em parceria com o Movimento dos Trabalhadores Rurais Sem Terra (MST). As etapas da ação foram planejamento da oficina, pesquisa e elaboração dos materiais de apoio e didático, execução da oficina e degustação das preparações, por meio de um teste de aceitabilidade. A oficina iniciou com uma roda de conversa para apresentar orientações sobre a ação e possibilitar diálogos, seguida pela execução das preparações culinárias e degustação. Participaram da atividade 12 pessoas, com idade entre 19 e 70 anos, e 67% eram mulheres. Na roda de conversa, observou-se grande interesse e troca de saberes entre os participantes, o que demonstra que as metodologias utilizadas se adequaram ao público participante. As preparações elaboradas na oficina culinária foram: massa fresca de chaya com molho de tomate, torta de ricota com chaya, arroz de cuxá, grissini de chaya e bolo com chaya. As preparações tiveram Índice de Aceitabilidade (IA) maior que 85% para todos os atributos sensoriais avaliados, verificando-se interesse em reproduzi-las casa. Concluiu-se que as oficinas culinárias podem articular os movimentos sociais e a comunidade acadêmica, fomentando a alimentação saudável. Palavras-chave: Extensão Universitária; Agrobiodiversidade; Cnidoscolus aconitifolius; Habilidades Culinárias Culinary workshop as an articulation strategy between social movements and the academic community to promote healthy and sustainable diets: an experience report Abstract: The eating habits of the Brazilian population are characterized by low consumption of fruits and vegetables, little variety of edible species, besides the high consumption of ultra-processed food, which contributes to non-transmissible chronic diseases. The development of cooking skills promotes rediscovering food culture and a healthy and sustainable diet. The objective of this paper was to report the experience of a culinary workshop using chaya (Cnidoscolus aconitifolius) as the main ingredient held during the 5th University Journey for the Support of Agrarian Reform at UFRJ (2018) in partnership with the Landless Rural Workers Movement (MST). The action had the stages of planning the culinary workshop, researching and preparing the support and didactic material, holding the workshop, and sensory evaluation of the preparations, using an acceptability test. The workshop began with a round-table discussion initially held to provide guidelines about the action and allow dialogues, followed by the culinary preparations and tasting. Twelve people between the ages of 19 and 70 participated in the activity, 67% of whom were women. During the round-table discussion, great interest and knowledge exchange among all participants were observed, showing that the methodologies were adequate to the participants. The preparations developed during the cooking workshop were: fresh chaya pasta with tomato sauce, ricotta pie with chaya, cuxá rice, chaya grissini, and (E) cake with chaya. The preparations made during the workshop presented an Acceptability Index (AI) higher than 85% for all evaluated sensory attributes. The participants also showed an interest in reproducing the workshop preparations at home. It was concluded that culinary workshops could be essential to articulate social movements and the academic community and promote a healthy diet. Keywords: University Extension; Agrobiodiversity; Cnidoscolus aconitifolius; Culinary Skills

None of the global targets for protecting nature are currently met, although humanity is critically dependent on biodiversity. A significant issue is the lack of data for most biodiverse regions of the planet where the use of frugal methods for biomonitoring would be particularly important because the available funding for monitoring is insufficient, especially in low-income countries. We here discuss how three approaches to insect biomonitoring (computer vision, lidar, DNA sequences) could be made more frugal and urge that all biomonitoring techniques should be evaluated for global suitability before becoming the default in high-income countries. This requires that techniques popular in high-income countries should undergo a phase of ‘innovation through simplification’ before they are implemented more broadly. We predict that techniques that acquire raw data at low cost and are suitable for analysis with AI (e.g. images, lidar-signals) will be particularly suitable for global biomonitoring, while techniques that rely heavily on patented technologies may be less promising (e.g. DNA sequences). We conclude the opinion piece by pointing out that the widespread use of AI for data analysis will require a global strategy for providing the necessary computational resources and training. This article is part of the theme issue ‘Towards a toolkit for global insect biodiversity monitoring’.

Background: Quality determination of frozen semen before using for Artificial Insemination (AI) in cattle is of paramount importance. The success of cattle AI programs is largely depends on the use of good quality semen and thus assessment of semen quality parameters before AI will improve the conception rate and reduces the loss associated with repeated breeding. Among the various semen quality indicators, sperm motility is one of the important parameter and it should be evaluated immediately after semen collection (before freezing) and 24 hrs after freezing as per minimum standard protocol in frozen semen production. Though the motility is a major determinant of spermatozoa ability to reach the site of fertilization and it is affected by cryopreservation and subsequent poor cold storage facilities, motility is not frequently evaluated under field conditions. In the current scenario, many of our government AI sub-stations in remote villages are not equipped with microscope facility for semen quality evaluation before AI. Methods: In this study, we used foldscope, an origami-based paper microscope which give optical quality similar to conventional microscopes (magnification of 140X and 2-micron resolution), to evaluate the frozen semen motility at field level. In this study, the motility of spermatozoa was observed under foldscope and concluded that foldscope is very useful to asses the motility befor AI and improve the fertility in dairy animals. Result: We revealed that, utility of foldscope to assess the motility of spermatozoa was evaluated under field conditions and observed that 46% of sperms were motile. Further, evalaution of sperm motility before AI is not very common practice due to lack of microscope in many AI centres. Therefore, foldscope will be very useful at field conditions to avoid AI using poor quality semen and subsequent loss associated with repeat breeding problems.

Today, diagnostic reasoning combines common and specialized knowledge, elements of numeracy, some facility with the basics of probability theory and, last but not least, ease in interactions with AI tools. We present procedures and tools for constructing trees that lead to understandable, transparent, simple, and robust classifications and decisions. These tools are more heuristic than optimal models, inspired by the perspective of Bounded Rationality. We describe how the tenets of Bounded Rationality provide a framework for the human-machine interaction this paper is devoted to. We claim that, because of this rationality, our proposed tools facilitate machine-aided decision making that is smooth, transparent and successful.

Tiny Machine Learning (TinyML) is an upsurging research field that proposes to democratize the use of Machine Learning and Deep Learning on highly energy-efficient frugal Microcontroller Units. Considering the general assumption that TinyML can only run inference, growing interest in the domain has led to work that makes them reformable, i.e., solutions that permit models to improve once deployed. This work presents a survey on reformable TinyML solutions with the proposal of a novel taxonomy. Here, the suitability of each hierarchical layer for reformability is discussed. Furthermore, we explore the workflow of TinyML and analyze the identified deployment schemes, available tools and the scarcely available benchmarking tools. Finally, we discuss how reformable TinyML can impact a few selected industrial areas and discuss the challenges, and future directions, and it’s fusion with next-generation AI.

ABSTRACT Communication between bacterial cells is crucial for the coordination of diverse cellular processes that facilitate environmental adaptation and, in the case of pathogenic species, virulence. This is achieved by the secretion and detection of small signaling molecules called autoinducers, a process termed quorum sensing. To date, the only signaling molecule recognized by both Gram-positive and Gram-negative bacteria is autoinducer 2 (AI-2), synthesized by the metabolic enzyme LuxS ( S -ribosylhomocysteine lyase) as a by-product of the activated methyl cycle. Homologues of LuxS are ubiquitous in bacteria, suggesting a key role in interspecies, as well as intraspecies, communication. Gram-negative bacteria sense and respond to AI-2 via the Lsr ABC transporter system or by the LuxP/LuxQ phosphorelay system. However, homologues of these systems are absent from Gram-positive bacteria and the AI-2 receptor is unknown. Here we show that in the major human pathogen Streptococcus pneumoniae , sensing of exogenous AI-2 is dependent on FruA, a fructose-specific phosphoenolpyruvate-phosphotransferase system that is highly conserved in Gram-positive pathogens. Importantly, AI-2 signaling via FruA enables the bacterium to utilize galactose as a carbon source and upregulates the Leloir pathway, thereby leading to increased production of capsular polysaccharide and a hypervirulent phenotype. IMPORTANCE S. pneumoniae is a Gram-positive bacterium frequently carried asymptomatically in the human nasopharynx. However, in a proportion of cases, it can spread to other sites of the body, causing life-threatening diseases that translate into massive global morbidity and mortality. Our data show that AI-2 signaling via FruA promotes the transition of the pneumococcus from colonization to invasion by facilitating the utilization of galactose, the principal sugar available in the upper respiratory tract. AI-2-mediated upregulation of Leloir pathway enzymes results in increased production of capsular polysaccharide and hypervirulence in a murine intranasal challenge model. This identifies the highly conserved FruA phosphotransferase system as a target for new antimicrobials based on the disruption of this generic quorum-sensing system.

Dans le cadre de la maintenance prédictive du constructeur automobile Renault, cette thèse vise à fournir des solutions à faible coût énergétique pour la détection non supervisée d'anomalies sur des séries temporelles. Avec l'évolution récente de l'automobile, de plus en plus de données sont produites et doivent être traitées par des algorithmes d'apprentissage automatique. Ce traitement peut être effectué dans le cloud ou directement à bord de la voiture. Dans un tel cas, la bande passante du réseau, les coûts des services cloud, la gestion de la confidentialité des données et la perte de données peuvent être économisés. L'intégration d'un modèle d'apprentissage automatique dans une voiture est un défi car elle nécessite des modèles frugaux en raison des contraintes de mémoire et de calcul. Dans ce but, nous étudions l'utilisation de réseaux de neurones impulsionnels (SNN) pour la detection d'anomalies, la prédiction et la classification sur des séries temporelles. Les performances et les coûts énergétiques des modèles d'apprentissage automatique sont évalués dans un scénario Edge à l'aide de modèles matériels génériques qui prennent en compte tous les coûts de calcul et de mémoire. Pour exploiter autant que possible l'activité neuronale parcimonieuse des SNN, nous proposons un modèle avec des connexions peu denses et entraînables qui consomme la moitié de l'énergie de sa version dense. Ce modèle est évalué sur des benchmarks publics de détection d'anomalies, un cas d'utilisation réel de détection d'anomalies sur les voitures de Renault Alpine, les prévisions météorologiques et le dataset Google Speech Command. Nous comparons également ses performances avec d'autres modèles d'apprentissage automatique existants. Nous concluons que, pour certains cas d'utilisation, les modèles SNN peuvent atteindre les performances de l'état de l'art tout en consommant 2 à 8 fois moins d'énergie. Pourtant, d'autres études devraient être entreprises pour évaluer ces modèles une fois embarqués dans une voiture. Inspirés par les neurosciences, nous soutenons que d'autres propriétés bio-inspirées telles que l'attention, l'activité parcimonieuse, la hiérarchie ou la dynamique des assemblées de neurons pourraient être exploités pour obtenir une meilleure efficacité énergétique et de meilleures performances avec des modèles SNN. Enfin, nous terminons cette thèse par un essai à la croisée des neurosciences cognitives, de la philosophie et de l'intelligence artificielle. En plongeant dans les difficultés conceptuelles liées à la conscience et en considérant les mécanismes déterministes de la mémoire, nous soutenons que la conscience et le soi pourraient être constitutivement indépendants de la mémoire. L'objectif de cet essai est de questionner la nature de l'humain par opposition à celle des machines et de l'IA
In the context of the predictive maintenance of the car manufacturer Renault, this thesis aims at providing low-power solutions for unsupervised anomaly detection on time-series. With the recent evolution of cars, more and more data are produced and need to be processed by machine learning algorithms. This processing can be performed in the cloud or directly at the edge inside the car. In such a case, network bandwidth, cloud services costs, data privacy management and data loss can be saved. Embedding a machine learning model inside a car is challenging as it requires frugal models due to memory and processing constraints. To this aim, we study the usage of spiking neural networks (SNNs) for anomaly detection, prediction and classification on time-series. SNNs models' performance and energy costs are evaluated in an edge scenario using generic hardware models that consider all calculation and memory costs. To leverage as much as possible the sparsity of SNNs, we propose a model with trainable sparse connections that consumes half the energy compared to its non-sparse version. This model is evaluated on anomaly detection public benchmarks, a real use-case of anomaly detection from Renault Alpine cars, weather forecasts and the google speech command dataset. We also compare its performance with other existing SNN and non-spiking models. We conclude that, for some use-cases, spiking models can provide state-of-the-art performance while consuming 2 to 8 times less energy. Yet, further studies should be undertaken to evaluate these models once embedded in a car. Inspired by neuroscience, we argue that other bio-inspired properties such as attention, sparsity, hierarchy or neural assemblies dynamics could be exploited to even get better energy efficiency and performance with spiking models. Finally, we end this thesis with an essay dealing with cognitive neuroscience, philosophy and artificial intelligence. Diving into conceptual difficulties linked to consciousness and considering the deterministic mechanisms of memory, we argue that consciousness and the self could be constitutively independent from memory. The aim of this essay is to question the nature of humans by contrast with the ones of machines and AI

Artificial Intelligence (AI) has produced extremely efficient and effective classification and decision “machines” that learn from given data sets and generalize well to unknown data. These are mostly celebrated tools produced by methodologies of machine learning. There is a drawback, though, namely their lack of transparency in construction. Agents often ignore the construction steps and use them as black-box algorithms. We exhibit simple and transparent steps for creating robust and yet simple heuristics for classification based on the AI tool ARBOR. We also claim that these transparent classifiers compete well against powerful machines, especially when training sets are small.