We consider a mechanistic non-linear machine learning approach to learning signals in financial time series data. A modularised and decoupled algorithm framework is established and is proven on daily sampled closing time-series data for JSE equity markets. The input patterns are based on input data vectors of data windows preprocessed into a sequence of daily, weekly and monthly or quarterly sampled feature measurement changes (log feature fluctuations). The data processing is split into a batch processed step where features are learnt using a Stacked AutoEncoder (SAE) via unsupervised learning, and then both batch and online supervised learning are carried out on Feedforward Neural Networks (FNNs) using these features. The FNN output is a point prediction of measured time-series feature fluctuations (log differenced data) in the future (ex-post). Weight initializations for these networks are implemented with restricted Boltzmann machine pretraining, and variance based initializations. The validity of the FNN backtest results are shown under a rigorous assessment of backtest overfitting using both Combinatorially Symmetrical Cross Validation and Probabilistic and Deflated Sharpe Ratios. Results are further used to develop a view on the phenomenology of financial markets and the value of complex historical data under unstable dynamics.

Alzheimer’s disease (AD) is a chronic and progressive, irreversible syndrome that deteriorates the cognitive functions. Official death certificates of 2013 reported 84,767 deaths from Alzheimer’s disease, making it the 6th leading cause of death in the United States. The rate of AD is estimated to double by 2050. The neurodegeneration of AD occurs decades before symptoms of dementia are evident. Therefore, having an efficient methodology for the early and proper diagnosis can lead to more effective treatments. Neuroimaging techniques such as magnetic resonance imaging (MRI) can detect changes in the brain of living subjects. Moreover, medical imaging techniques are the best diagnostic tools to determine brain atrophies; however, a significant limitation is the level of training, methodology, and experience of the diagnostician. Thus, Computer aided diagnosis (CAD) systems are part of a promising tool to help improve the diagnostic outcomes. No publications addressing the use of Feedforward Artificial Neural Networks (ANN), and MRI image attributes for the classification of AD were found. Consequently, the focus of this study is to investigate if the use of MRI images, specifically texture and frequency attributes along with a feedforward ANN model, can lead to the classification of individuals with AD. Moreover, this study compared the use of a single view versus a multi-view of MRI images and their performance. The frequency, texture, and MRI views in combination with the feedforward artificial neural network were tested to determine if they were comparable to the clinician’s performance. The clinician’s performances used were 78 percent accuracy, 87 percent sensitivity, 71 percent specificity, and 78 percent precision from a study with 1,073 individuals. The study found that the use of the Discrete Wavelet Transform (DWT) and Fourier Transform (FT) low frequency give comparable results to the clinicians; however, the FT outperformed the clinicians with an accuracy of 85 percent, precision of 87 percent, sensitivity of 90 percent and specificity of 75 percent. In the case of texture, a single texture feature, and the combination of two or more features gave results comparable to the clinicians. However, the Gray level co-occurrence matrix (GLCOM), which is the combination of texture features, was the highest performing texture method with 82 percent accuracy, 86 percent sensitivity, 76 percent specificity, and 86 percent precision. Combination CII (energy and entropy) outperformed all other combinations with 78 percent accuracy, 88 percent sensitivity, 72 percent specificity, and 78 percent precision. Additionally, a combination of views can increase performance for certain texture attributes; however, the axial view outperformed the sagittal and coronal views in the case of frequency attributes. In conclusion, this study found that both texture and frequency characteristics in combinations with a feedforward backpropagation neural network can perform at the level of the clinician and even higher depending on the attribute and the view or combination of views used.

No description available.

科学研究費補助金 研究種目:基盤研究(A)(2) 課題番号:08555101 研究代表者:早川 義一 研究期間:1996-1997年度

This thesis aims to solve the mathematical inverse problem of characterizing sea waves based on the responses obtained from a marine vessel sailing under certain sea conditions. By researching this problem the thesis contributes to the marine industry by improving products that are using ocean behavior for controlling ship's dynamics. Knowledge about the current state of the sea, such as the wave frequency and height, is important for navigation, control, and for the safety of a vessel. This information can be retrieved from specialized weather reports. However, such information is not at all time possible to obtain during a voyage, and if so usually comes with a certain delay. Therefore this thesis seeks solutions that can estimate on-line the waves' state using methods in the field of Artificial Intelligence. The specific investigation methods are Transfer Functions augmented with Genetic Algorithm, Artificial Neural Networks and Case-Based Reasoning. These methods have been configured and validated using the n-fold cross validation method. All the methods have been tested with an actual implementation. The algorithms have been trained with data acquired from a marine simulation program developed in Simulink. The methods have also been trained and tested using monitored data acquired from an actual ship sailing on the Baltic Sea as well as wave data obtained from a buoy located nearby the vessel's route. The proposed methods have been compared with state-of-the art reports in order evaluate the novelty of the research and its potential applications in industry. The results in this thesis show that the proposed methods can in fact be used for solving the inverse problem. It was also found that among the investigated methods it is the Transfer Function augmented with Genetic Algorithm which yields best results. This Master Thesis is conducted under the Master of Engineering Program in Robotics at Mälardalens högskola in Västerås, Sweden. The thesis was proposed by Q-TAGG R&D AB in Västerås, Sweden, a company which specializes in marine vessel dynamics research.

La décharge des cellules de Purkinje (CP), neurone de sortie du cortex cérébelleux, joue un rôle majeur dans le contrôle moteur. Les CP reçoivent des entrées excitatrices provenant des cellules des grains (CG), lesquelles génèrent également une inhibition antérograde sur les CP via l’activation d’interneurones de la couche moléculaire (IN). Lors de ma thèse, j’ai étudié l’influence simultanée de la balance excitation-inhibition (E/I) et des plasticités à court terme aux synapses CG-IN-CP sur la décharge des CP, par des techniques d’électrophysiologie, d’optogénétique et de simulation. Ces travaux démontrent l’existence d’une hétérogénéité d’E/I dans le cortex cérébelleux ainsi qu’une grande diversité de modulation des CP en réponse à la stimulation de CG. Le nombre de stimulation des CG influence fortement la direction et l’intensité de la modulation observée. Enfin, la combinaison de plasticités à court terme et d’E/I génère dans la décharge des CP des motifs de réponses complexes mais reproductibles, ayant sans doute un rôle essentiel dans l’encodage sensoriel. / Rate and temporal coding in Purkinje cells (PC), the sole output of the cerebellar cortex, play a major role in motor control. PC receives excitatory inputs from granule cells (GC) which also provide feedforward inhibition on PC through the activation of molecular layer interneurons (MLI). In this thesis, I studied the influence of the combined action of excitation/inhibition (E/I) balance and short-term plasticity of GC-MLI-PC synapses on PC discharge, by using electrophysiological recordings, optogenetic stimulation and modelling. This work demonstrates that E/I balances are not equalized in the cerebellar cortex and showed a wide distribution of PC discharge modulation in response to GC inputs, from an increase to a shut down of the discharge. The number of stims in GC bursts strongly controls the strength and sign of PC modulation. Lastly, the interplay between short-term plasticity and E/I balance implements complex but reproducible output patterns of PC responses to GC inputs that should play a key role in stimulus encoding by the cerebellar cortex.

The growing urgency of climate change has led to growth in the electrification technology field, where batteries have emerged as an essential role in the renewable energy transition, supporting the implementation of environmentally friendly technologies such as smart grids, energy storage systems, and electric vehicles. Battery cell degradation is a common occurrence indicating battery usage. Optimizing lithium-ion battery degradation during operation benefits the prediction of future degradation, minimizing the degradation mechanisms that result in power fade and capacity fade. This degree project aims to investigate battery degradation prediction based on capacity using deep learning methods. Through analysis of battery degradation and health prediction for lithium-ion cells using non-destructive techniques. Such as electrochemical impedance spectroscopy obtaining ECM and three different deep learning models using multi-channel data. Additionally, the AI models were designed and developed using multi-channel data and evaluated performance within MATLAB. The results reveal an increased resistance from EIS measurements as an indicator of ongoing battery aging processes such as loss o active materials, solid-electrolyte interphase thickening, and lithium plating. The AI models demonstrate accurate capacity estimation, with the LSTM model revealing exceptional performance based on the model evaluation with RMSE. These findings highlight the importance of carefully managing battery charging processes and considering factors contributing to degradation. Understanding degradation mechanisms enables the development of strategies to mitigate aging processes and extend battery lifespan, ultimately leading to improved performance.

The issue of learner engagement is an important question for education and for instructional design. It is acknowledged that computer games in general are engaging. Thus, one possible solution to learner engagement is to integrate computer games into education; however, the literature indicates that pedagogical, logistical and political barriers remain. Another possible solution is to derive principles for the design of engaging experiences from a critical examination of computer game design. One possible application of the derived design principles is that instruction may be designed to be inherently more engaging. The purpose of this dissertation was to look for operational principles underlying the design of computer games in order to better understand the design of engaging experiences. Core design components and associated operational principles for the design of engaging experiences were identified. Selected computer games were analyzed to demonstrate that these components and principles were present in the design of successful computer games. Selected instructional units were analyzed to show evidence that these operational principles could be applied to the design of instruction. An instructional design theory—called Challenge-driven Instructional Design—and design considerations for the theory were proposed. Finally, suggestions were made for continued development and research of the instructional design theory.

The Permanent Magnet Synchronous Motor (PMSM, also known as the servo motor) is a crucial component within robotic servo systems. To optimally respond to the torque demands sent from the high-level motion controller, the PMSM current controller must track the reference with speed and precision. Nevertheless, the operation of servo motors could be compromised due to the nonlinearity of flux linkage and inaccuracies in parameters induced by unpredictable fluctuations in temperature. This Master’s thesis proposes a novel Robust Nonlinear Deadbeat Predictive Current Control (RN-DPCC) scheme to counter these challenges effectively. The nonlinear mappings between flux linkage and current on the dq-axis are established using polynomial fitting based on experimental data. Furthermore, the Nonlinear Deadbeat Predictive Current Control (N-DPCC) is derived using nonlinear feedforward. Meanwhile, Delayed Integral Action (DIA) is introduced as a robustness-enhancing measure for N-DPCC, thus evolving it into the Robust N-DPCC (RN-DPCC). Compared to conventional Integral Action (IA), DIA effectively curtails overshoot triggered by integral error and accelerates the current transient without incorporating additional tunable parameters. Numerical simulations that leverage the mathematical modeling of the converter and nonlinear PMSM are implemented using fundamental blocks in Simulink, which replicates the actual experimental setup employed within the Motor Control Lab at ABB Corporate Research. The effectiveness of employing nonlinear feedforward compensation is confirmed through a comparative analysis of the simulation results from N-DPCC and conventional Deadbeat Predictive Current Control (DPCC). The enhancements in transient response brought about by DIA are demonstrated through a comparison of RNDPCC and N-DPCC with IA. The robustness of RN-DPCC is demonstrated by comparing it with N-DPCC under conditions where parameter inaccuracies are present. / Den permanenta magnet-synkronmotorn (PMSM, även känd som servomotorn) är en avgörande komponent inom robotiserade servosystem. För att optimalt kunna reagera på momentkraven som skickas från högnivårörelsekontrollern måste PMSM-strömregulatorn följa referensen med hastighet och precision. Trots detta kan driften av servomotorer påverkas av ickelinjäriteter i flödeslänkningen och felaktigheter i parametrar som orsakas av oförutsägbara temperaturfluktuationer. Denna magisteravhandling föreslår en ny robust icke-linjär deadbeat-prediktiv strömreglering (RN-DPCC) för att effektivt hantera dessa utmaningar. De icke-linjära avbildningarna mellan flödeslänkning och ström på dq-axeln etableras med hjälp av polynomisk anpassning baserat på experimentella data. Dessutom härleds den ickelinjära deadbeat-prediktiva strömregleringen (N-DPCC) med hjälp av Ickelinjär feedforward. Samtidigt introduceras fördröjd integralåtgärd (DIA) som en robusthetsförbättrande åtgärd för N-DPCC, vilket förvandlar den till Robust N-DPCC (RN-DPCC). Jämfört med konventionell integralåtgärd (IA) minskar DIA effektivt överhäng som utlöses av integralfel och accelererar strömövergången utan att införa ytterligare justerbara parametrar. Numeriska simuleringar som utnyttjar den matematiska modelleringen av omvandlaren och den icke-linjära PMSM implementeras med hjälp av grundläggande block i Simulink, vilket återskapar den faktiska experimentella uppställningen som används i Motor Control Lab vid ABB Corporate Research. Effektiviteten i att använda icke-linjär framåtmatningskompensation bekräftas genom en jämförande analys av simulationsresultaten från N-DPCC och konventionell deadbeat-prediktiv strömreglering (DPCC). Förbättringarna i transientrespons som DIA medför demonstreras genom en jämförelse av RN-DPCC och NDPCC med IA. Robustheten hos RN-DPCC demonstreras genom att jämföra den med N-DPCC under förhållanden där parameterfel förekommer.

Published works cite that 70-80% of the total cost of a product is established during conceptual design, and that improvements in time-to-market, quality, affordability, and global competitiveness require the development of better approaches to assist decision-making during the early stages of product design, as well as facilitate enterprise knowledge management and reuse. For many years, concurrent engineering and teaming have been viewed as “the answer” to product development woes, but studies reveal teaming is not sufficient to handle the task complexities of product development and the long-term goal of enterprise learning. The work of Roberto Verganti provides new insights with regard to reciprocal interdependencies (RIs), feedforward planning, selective anticipation in the context of improving teaming and concurrent engineering, as well as enterprise learning, knowledge management, reuse. In this research, reciprocal interdependencies management (RIM) is offered as a means of addressing product development and concurrent engineering issues occurring in the early stages of design. RIM is combination of Verganti’s concepts, a conceptual RIs structure, new RIM-application strategies, RIM-diagramming, and a conceptual RIM-based decisions support system, which come together to form a vision of a RIM-based enterprise knowledge management system. The conceptual RIM-based DSS is presented using the specific case of supporting a working-level integrated product team (IPT) engaged in the design of an aircraft bulkhead. A qualitative assessment tool is used to compare RIM to other approaches in the literature, and initial results are very favorable.