Deafness has been associated with poor abilities to deal with digits in the context of arithmetic and memory, and language modality-specific differences in the phonological similarity of digits have been shown to influence short-term memory (STM). Therefore, the overall aim of the present thesis was to find out whether language modality-specific differences in phonological processing between sign and speech can explain why deaf signers perform at lower levels than hearing peers when dealing with digits. To explore this aim, the role of phonological processing in digit-based arithmetic and memory tasks was investigated, using both behavioural and neuroimaging methods, in adult deaf signers and hearing non-signers, carefully matched on age, sex, education and non-verbal intelligence. To make task demands as equal as possible for both groups, and to control for material effects, arithmetic, phonological processing, STM and working memory (WM) were all assessed using the same presentation and response mode for both groups. The results suggested that in digit-based STM, phonological similarity of manual numerals causes deaf signers to perform more poorly than hearing non-signers. However, for  digit-based WM there was no difference between the groups, possibly due to differences in allocation of resources during WM. This indicates that similar WM for the two groups can be generalized from lexical items to digits. Further, we found that in the present work deaf signers performed better than expected and on a par with hearing peers on all arithmetic tasks, except for multiplication, possibly because the groups studied here were very carefully matched. However, the neural networks recruited for arithmetic and phonology differed between groups. During multiplication tasks, deaf signers showed an increased  reliance on cortex of the right parietal lobe complemented by the left inferior frontal gyrus. In contrast, hearing non-signers relied on cortex of the left frontal and parietal lobes during multiplication. This suggests that while hearing non-signers recruit phonology-dependent arithmetic fact retrieval processes for multiplication, deaf signers recruit non-verbal magnitude manipulation processes. For phonology, the hearing non-signers engaged left lateralized frontal and parietal areas within the classical perisylvian language network. In deaf signers, however, phonological processing was limited to cortex of the left occipital lobe, suggesting that sign-based phonological processing does not necessarily activate the classical language network. In conclusion, the findings of the present thesis suggest that language modality-specific differences between sign and speech in different ways can explain why deaf signers perform at lower levels than hearing non-signers on tasks that include dealing with digits. / Dövhet har kopplats till bristande förmåga att hantera siffror inom områdena aritmetik och minne. Särskilt har språkmodalitetsspecifika skillnader i fonologisk likhet för siffror visat sig påverka korttidsminnet. Det övergripande syftet med den här avhandlingen var därför att undersöka om språkmodalitetsspecifika skillnader i fonologisk bearbetning mellan teckenoch talspråk kan förklara varför döva presterar sämre än hörande på sifferuppgifter. För att utforska det området undersöktes fonologisk bearbetning i sifferbaserade minnesuppgifter och aritmetik med hjälp av både beteendevetenskapliga metoder och hjärnavbildning hos grupper av teckenspråkiga döva och talspråkiga hörande som matchats noggrant på ålder, kön, utbildning och icke-verbal intelligens. För att testförhållandena skulle bli så likartade som möjligt för de båda grupperna, och för att förebygga materialeffekter, användes samma presentations- och svarssätt för båda grupperna. Resultaten visade att vid sifferbaserat korttidsminne påverkas de dövas prestation av de tecknade siffrornas fonologiska likhet. Däremot fanns det ingen skillnad mellan grupperna gällande sifferbaserat arbetsminne, vilket kan bero på att de båda grupperna fördelar sina kognitiva resurser på olika sätt. Dessutom fann vi att den grupp teckenspråkiga döva som deltog i studien presterade bättre på aritmetik än vad tidigare forskning visat och de skiljde sig bara från hörande på multiplikationsuppgifter, vilket kan bero på att grupperna var så noggrant matchade. Däremot fanns det skillnader mellan grupperna i vilka neurobiologiska nätverk som aktiverades vid aritmetik och fonologi. Vid multiplikationsuppgifter aktiverades cortex i höger parietallob och vänster frontallob för de teckenspråkiga döva, medan cortex i vänster frontal- och parietallob aktiverades för de talspråkiga hörande. Detta indikerar att de talspråkiga hörande förlitar sig på fonologiberoende minnesstrategier medan de teckenspråkiga döva förlitar sig på ickeverbal magnitudmanipulering och artikulatoriska processer. Under den fonologiska uppgiften aktiverade de talspråkiga hörande vänsterlateraliserade frontala och parietala områden inom det klassiska språknätverket. För de teckenspråkiga döva var fonologibearbetningen begränsad till cortex i vänster occipitallob, vilket tyder på att teckenspråksbaserad fonologi inte behöver aktivera det klassiska språknätverket. Sammanfattningsvis visar fynden i den här avhandlingen att språkmodalitetsspecifika skillnader mellan tecken- och talspråk på olika sätt kan förklara varför döva presterar sämre än hörande på vissa sifferbaserade uppgifter.

Orientador: Elenice Aparecida de Moraes Ferrari / Dissertação (mestrado) - Universidade Estadual de Campinas, Instituto de Biologia / Made available in DSpace on 2018-08-24T09:45:43Z (GMT). No. of bitstreams: 1 Mendes_RubiaAparecidaPereiradeCarvalho_M.pdf: 3863370 bytes, checksum: 488ca978a38a7cb5f131a008fc5ceac4 (MD5) Previous issue date: 2013 / Resumo: Este estudo investigou a distribuição do cronotipo (matutino, intermediário e vespertino), a sonolência diurna, o desempenho em testes de memória e de compreensão de leitura em estudantes do turno da manhã e da tarde (12 a 17 anos). Na Fase 1, foram utilizados o Questionário de Cronotipo (HO) e o formulário para identificação de queixas de sono (n = 435). Na Fase 2 (n = 81) utilizou-se a Escala de Sonolência de Karolinska (KSS) aplicada imediatamente antes da sessão de testes, que ocorreram em dois horários (7h00 e 16h45), as Escalas de Leitura (EL) e de Desempenho Escolar em Língua Portuguesa (ED), o Teste de Extensão de Dígitos direto e inverso (SPANdir; SPANinv), o Teste dos Blocos de Corsi direto e inverso (CORSIdir; CORSIinv), o Teste de Memória Episódica imediato e tardio (MEI; MET) e o Teste de compreensão da leitura (CLOZE). Os resultados mostraram uma distribuição normal para os diferentes cronotipos, com aumento na vespertinidade em adolescentes mais velhos. As queixas mais freqüentes (Teste qui-quadrado) foram insônia e muita necessidade de sono predominantes em vespertinos, independentemente do turno de estudo. Os adolescentes de cronotipo matutinos estavam menos sonolentos em relação aos vespertinos, em ambos os horários testados, sendo que os testados no horário da tarde estavam menos sonolentos que os da manhã (ANOVA; p < 0,05). A pontuação média na EL e ED classificou os adolescentes em alto nível acadêmico (ANA), média = ou > que 5 (n = 50) e baixo nível acadêmico (BNA), média < que 5 (n = 31). A análise do conjunto de dados de todos os 81 sujeitos (ANA e BNA) mostrou melhor desempenho no teste CORSIinv realizado no horário da tarde (R = -0,5) em adolescentes vespertinos e no teste MET no horário da manhã (R = -0,4) em adolescentes menos sonolentos (Teste de Spearman; p < 0,05). Quando se analisou apenas os sujeitos ANA observou-se melhor desempenho dos adolescentes matutinos nos testes SPANinv realizados no horário da manhã (R = 0,3) e no teste MEI no horário da tarde (R = 0,4). Também foi verificado melhor desempenho em adolescentes menos sonolentos nos testes CORSIinv (R = -0,6) e CLOZE (R = -0,5) realizados no horário da manhã (Teste de Spearman; p < 0,05). Em conjunto, essas análises demonstraram que o cronotipo, a sonolência diurna, o nível de leitura e o desempenho escolar exercem influência no desempenho em testes de memória. Mais ainda, mostram a importância de se avaliar o nível acadêmico do indivíduo para precisar a relação que cronotipo e a sonolência exercem sobre a memória / Abstract: This study investigated the distribution of different chronotypes (morning, indifferent and evening), the diurnal sleepiness, the performance in tests of memory and reading comprehension in students (12-17 years old) of both morning and afternoon school periods. In the Phase 1, we used the Chronotype Questionnaire (HO) and Sleep Complaints form (n = 435). In Phase 2 (n = 81) there has been used the Karolinska Sleepiness Scale (KSS) immediately before the morning (7am) and afternoon (4:45pm) testing sessions, the Reading (SR) and School Performance in Portuguese Language (SP) Scales, the Digit Span (direct, SPANdir and inverse, SPANinv), Corsi Block (direct, CORSIdir and inverse, CORSIinv), Episodic Memory (immediate, IEM and late, LEM) and the reading comprehension test (Cloze). The results showed a normal distribution of the different chronotypes, with increased eveningness in older adolescents. The most frequent sleep complaints ( x -square test) were the need for longer sleep duration and the insomnia. Those complaints were more frequent in the evening type adolescents, regardless of the school period. The morning type adolescents were less sleepier than the evening type adolescents in both testing times, but during afternoon sessions they were less sleepier than during the morning sessions (ANOVA; p < 0,05). The adolescents were classified in high (HAL, score > 5, n = 50), or low academic level (LAL, score < 5, n = 31), accordingly to their SR and SP average score. The analysis considering all the 81 adolescents (ANA e BNA) indicated that evening type had better performance than the morning type adolescents in the CORSIinv test conducted at the afternoon time (R = -0.5), whereas less sleepy adolescents showed better performance than more sleepy adolescents (R = -0.4) in the MET test during the morning session (Spearman test; p < 0.05). Data analysis of the HAL students only showed that morning type adolescents performed better than evening type adolescents in the SPANinv test conducted in the morning (R = 0.3) and MEI test conducted in the afternoon (R= 0.4). Also the performance in CORSIinv (R = -0.6) and CLOZE tests (R = -0.5) was better in less sleepy than in more sleepy adolescents during the morning sessions (Spearman test; p < 0.05). Together, these results demonstrate that the performance of adolescents in memory tests can be influencied by their chronotype, sleepiness and level of academic performance. Furthermore, this study demonstrates that assessment of the academic level helps to clarify the role that chronotype and sleepiness have on memory / Mestrado / Fisiologia / Mestra em Biologia Funcional e Molecular

Multi-Task Learning is today an interesting and promising field which many mention as a must for achieving the next level advancement within machine learning. However, in reality, Multi-Task Learning is much more rarely used in real-world implementations than its more popular cousin Transfer Learning. The questionis why that is and if Multi-Task Learning outperforms its Single-Task counterparts. In this thesis different Multi-Task Learning architectures were utilized in order to build a model that can handle labeling real technical issues within two categories. The model faces a challenging imbalanced data set with many labels to choose from and short texts to base its predictions on. Can task-sharing be the answer to these problems? This thesis investigated three Multi-Task Learning architectures and compared their performance to a Single-Task model. An authentic data set and two labeling tasks was used in training the models with the method of supervised learning. The four model architectures; Single-Task, Multi-Task, Cross-Stitched and the Shared-Private, first went through a hyper parameter tuning process using one of the two layer options LSTM and GRU. They were then boosted by auxiliary tasks and finally evaluated against each other.

Bakgrund: Obstruktiv sömnapné (OSA) är en sjukdom med återkommande andningsuppehåll i de övre luftvägarna till följd av delvis eller total tilltäppning av luftflödet vid sömn, vilket leder till en lägre syremättnad i blodet. Det är känt att OSA försämrar kognitiva funktioner såsom minne och inlärning. Det saknas evidens om samband mellan kognitiv funktion, grad av symptom och fysisk aktivitetsnivå.   Syfte: Att beskriva kognitiv funktion hos en grupp patienter med övervikt och OSA. Undersöka samband mellan kognitiv funktion (korttidsminne) och fysisk aktivitet på moderat till hög intensitet (MVPA) samt undersöka samband mellan kognitiv funktion och svårighetsgrad av OSA.   Metod: Studien är en tvärsnittsstudie med korrelerande design. Grad av OSA mättes med apné-hypopné-index (AHI), kognitiv funktion med sifferrepetitionstest och fysisk aktivitet med accelerometer. I studien inkluderades 86 patienter diagnostiserade med OSA (AHI &gt;15), Body Mass Index (BMI) &gt;25 och ha en självskattad fysisk aktivitetsnivå på mindre än 150 minuter per vecka.   Resultat: Resultatet av sifferrepetitionstestet i studien visade ett medelvärde på 15,7. Det fanns inget signifikant samband (r=-0,10, p=0,38) mellan mängden MVPA och kognitiv funktion hos patienter med OSA. Det fanns heller inget signifikant samband (r=0,09, p=0,40) mellan AHI och prestation på sifferrepetitionstestet.   Slutsats: Mängden fysisk aktivitet verkar inte ha någon effekt på kognitiv funktion. Det verkar inte heller finnas något samband mellan AHI och en patients kognitiva funktion. Ytterligare forskning på ämnet krävs för att kunna bekräfta resultatet från denna studie. / Background: Obstructive sleep apnea (OSA) is a disease with repeated episodes of partially or completely blocked airways during sleep, which leads to a lower blood saturation. OSA is known to decrease cognitive abilities like memory and learning. There is a lack of studies which investigate if there is an association between cognitive function, symptoms and physical activity in patients with OSA.   Purpose: To describe the cognitive function in a group of patients who are overweight and has OSA. Examine the association between cognitive function (short-term memory) and moderate to vigorous physical activity (MVPA) and examine the association between cognitive function and severity of OSA.   Method: This paper is a cross-sectional study with correlational design. The severity of the disease was measured with apnea-hypopnea index (AHI). The cognitive function was measured by digit span test and physical activity with an accelerometer. The study includes 86 patients diagnosed with OSA (AHI &gt;15), with a BMI &gt;25 and a self-assessed moderate intensity physical activity less than 150 minutes/week.   Results :The digit span test in this study showed a mean value of 15,7. There was no significant correlation (r=-0.10, p=0.38) between amount of MVPA and short-term memory in patients with OSA (p=0.38). There was neither any significant correlation (r=0.09, p=0.40) between patients AHI and the cognitive functions.   Conclusion: The amount of physical activity didn’t seem to have any effect on cognitive function. There doesn't seem to be any association between AHI and patients cognitive function. Further research is needed to confirm the results from this study.

Many sensor nodes are scattered in the sensor network,which are used in all aspects of life due to their small size, low power consumption, and multiple functions. With the advent of the Internet of Things, more small sensor devices will appear in our lives. The research of deep learning neural networks is generally based on large and medium-sized devices such as servers and computers, and it is rarely heard about the research of neural networks based on small Internet of Things devices. In this study, the Internet of Things devices are divided into three types: large, medium, and small in terms of device size, running speed, and computing power. More vividly, I classify the laptop as a medium- sized device, the device with more computing power than the laptop, like server, as a large-size IoT(Internet of Things) device, and the IoT mobile device that is smaller than it as a small IoT device. The purpose of this paper is to explore the feasibility, usefulness, and effectiveness of long-short-term memory neural network model value prediction research based on small IoT devices. In the control experiment of small and medium-sized Internet of Things devices, the following results are obtained: the error curves of the training set and verification set of small and medium-sized devices have the same downward trend, and similar accuracy and errors. But in terms of time consumption, small equipment is about 12 times that of medium-sized equipment. Therefore, it can be concluded that the LSTM(long-and-short-term memory neural networks) model value prediction research based on small IoT devices is feasible, and the results are useful and effective. One of the main problems encountered when the LSTM model is extended to small devices is time-consuming.

In this thesis, two different types of bus passengers are identified. These two types, namely current passengers and passengers-to-be have different needs in terms of arrival time predictions. A set of machine learning models based on recurrent neural networks and long short-term memory units were developed to meet these needs. Furthermore, bus data from the public transport in Östergötland county, Sweden, were collected and used for training new machine learning models. These new models are compared with the current prediction system that is used today to provide passengers with arrival time information. The models proposed in this thesis uses a sequence of time steps as input and the observed arrival time as output. Each input time step contains information about the current state such as the time of arrival, the departure time from thevery first stop and the current position in Cartesian coordinates. The targeted value for each input is the arrival time at the next time step. To predict the rest of the trip, the prediction for the next step is simply used as input in the next time step. The result shows that the proposed models can improve the mean absolute error per stop between 7.2% to 40.9% compared to the system used today on all eight routes tested. Furthermore, the choice of loss function introduces models thatcan meet the identified passengers need by trading average prediction accuracy for a certainty that predictions do not overestimate or underestimate the target time in approximately 95% of the cases.

<p>Due to strong interactions among various phases and among the phases and fluid motions, multiphase flows (MPFs) are so complex that lots of efforts have to be paid to predict its sequential patterns of phases and motions. The present paper applies the physical constraints inherent in MPFs and enforces them to a physics-informed neural network (PINN) model either explicitly or implicitly, depending on the type of constraints. To predict the unobserved order parameters (OPs) (which locate the phases) in the future steps, the conditional neural processes (CNPs) with long short-term memory (LSTM, combined as CNPLSTM) are applied to quickly infer the dynamics of the phases after encoding only a few observations. After that, the multiphase consistent and conservative boundedness mapping (MCBOM) algorithm is implemented the correction the predicted OPs from CNP-LSTM so that the mass conservation, the summation of the volume fractions of the phases being unity, the consistency of reduction, and the boundedness of the OPs are strictly satisfied. Next, the density of the fluid mixture is computed from the corrected OPs. The observed velocity and density of the fluid mixture then encode in a physics-informed conditional neural processes and long short-term memory (PICNP-LSTM) where the constraint of momentum conservation is included in the loss function. Finally, the unobserved velocity in future steps is predicted from PICNP-LSTM. The proposed physics-informed neural processes (PINPs) model (CNP-LSTM-MCBOM-PICNP-LSTM) for MPFs avoids unphysical behaviors of the OPs, accelerates the convergence, and requires fewer data. The proposed model successfully predicts several canonical MPF problems, i.e., the horizontal shear layer (HSL) and dam break (DB) problems, and its performances are validated.</p>

A very general framework for modeling uncertainty in learning environments is given by Partially Observable Markov Decision Processes (POMDPs). In a POMDP setting, the learning agent infers a policy for acting optimally in all possible states of the environment, while receiving only observations of these states. The basic idea for coping with partial observability is to include memory into the representation of the policy. Perfect memory is provided by the belief space, i.e. the space of probability distributions over environmental states. However, computing policies defined on the belief space requires a considerable amount of prior knowledge about the learning problem and is expensive in terms of computation time. In this thesis, we present a reinforcement learning algorithm for solving deterministic POMDPs based on short-term memory. Short-term memory is implemented by sequences of past observations and actions which are called history lists. In contrast to belief states, history lists are not capable of representing optimal policies, but are far more practical and require no prior knowledge about the learning problem. The algorithm presented learns policies consisting of two separate phases. During the first phase, the learning agent collects information by actively establishing a history list identifying the current state. This phase is called the efficient identification strategy. After the current state has been determined, the Q-Learning algorithm is used to learn a near optimal policy. We show that such a procedure can be also used to solve large Markov Decision Processes (MDPs). Solving MDPs with continuous, multi-dimensional state spaces requires some form of abstraction over states. One particular way of establishing such abstraction is to ignore the original state information, only considering features of states. This form of state abstraction is closely related to POMDPs, since features of states can be interpreted as observations of states.

Researchers have linked dementia to common psychiatric symptoms such as agitation and aggression, known as behavioral and psychological symptoms of dementia (BPSD). To treat residents manifesting BPSD, nursing homes (NHs) use psychoactive medications. However, research is limited and inconsistent regarding the impact of Depakote treatment on agitation and short-term memory (STM) in NH residents who have dementia. The purpose of this nonexperimental quantitative study was to evaluate for 1 year the impact of Depakote treatment on agitation and STM in NH residents as measured by each resident's Minimum Data Set (MDS). Moncrieff and Cohen's drug-centered theory served as the theoretical foundation for the study. Archival data from the consulting pharmacist and NH MDS included 16 NH dementia residents. Data were analyzed using a repeated-measures within-subject ANOVA. Results indicated no significant impact of Depakote treatment on agitation and STM scores over a 1-year period. Results may be used to assess the impact and efficacy of a common yet largely unexamined invasive treatment on an underserved, vulnerable population.

The purpose of this study was to collect normative data on the auditory memory performance of three- and four-year old children as measured by the Auditory Memory Test Package (AMTP). Specifically, this investigation sought to answer one question: is the AMTP sensitive to age differences when administered to young children ages 3.0-4.11?