Magistriniame darbe tiriamos rašto ženklų atpažinimo problemos, nagrinėjami neuroniniai tinklai skirti rašto ženklams atpažinti. Apžvelgiamos problemos, kylančios sprendžiant rašto atpažinimo uždavinius, įvairūs problemų sprendimų būdai. Pasiūlytas dirbtinio neuroninio tinklo mokymo strategijos, pagrįstos klaidos skleidimo atgal algoritmu, patobulinimas. Patobulinimo esmė yra ta, kad mokymo aibė į tinklą paduodama ne visa iškarto, o dalimis. Kai neuroninis tinklas išmoksta atpažinti tą dalį, mokymo aibė papildoma naujais duomenimis, bet pradinių svorių vektoriai nesikeičia ir tinklas mokomas toliau. Šis algoritmo patobulinimas leidžia ženkliai sumažinti apmokymo laiką neprarandant tikslumo. Kai kuriais atvejais neuroninis tinklas, mokomas pagal įprastą klaidos skleidimo atgal algoritmą, nerodė jokių mokymo perspektyvų, po ~ 9 val. paklaida nesikeisdavo, neuroninis tinklas negalėdavo teisingai atpažinti nei vienos raidės. Panaudojus patobulintą strategiją, mus tenkinanti paklaida pasiekiama po ~ 3 val. 30 min. Taip pat darbe tiriama atpažinimo tikslumo priklausomybė nuo svorių pradinių reikšmių ir neuronų skaičiaus paslėptuose sluoksniuose. Nustatyti intervalai, kuriuose turi būti generuojamos svorių pradinės reikšmės, siekiant gauti tikslius atpažinimo rezultatus. Neuronų skaičius paslėptuose sluoksniuose turi būti daugiau nei penki. / The master thesis presents investigations of the problems of the optical character recognition. It also deals with the artificial neural networks that are designed for the optical character recognition. The work surveys the problems that emerge during the process of the optical character recognition. Various solutions are investigated. The improvement of a strategy for teaching the neural network that is based on the error back propagation is suggested. The essence of the improvement is that the training data set is divided into some parts and these parts are presented to the network one by one. When the neural network learns to recognize a part, the next part is presented to the network without any changes of the initial weight vectors and the network is trained further. This improvement allows us to reduce the training time significantly without losing the recognition accuracy. In some cases, the neural network that is trained according the ordinary error back propagation algorithm does not show any prospects. After ~ 9 hours, the error remains the same, the neural networks cannot recognize any letters. Using the improved strategy, the error satisfied is reached after ~3 hours 30 minutes. The dependence of the recognition accuracy on the values of the initial weight vectors and on the number of neurons in hidden layers is also investigated. The intervals in which the values of the initial weight vectors must be generated are identified, in order to get the correct results... [to full text]

The ability to process and respond to environmental cues requires the transformation of a sensory stimulus into an appropriate motor response, a process known as a sensorimotor transformation. The anti-saccade task can be used to investigate the ability of a subject to suppress a reflexive saccade towards a visual stimulus (pro-saccade) and generate a voluntary saccade 180° away from it. Additional steps are involved in the anti-saccade sensorimotor transformation that do not occur in the pro-saccade, which may produce performance differences between pro- and anti-saccade metrics. We were interested in exploring these differences to gain insight on the mechanism of the sensorimotor transformation of the anti-saccade and to uncover any directional biases in saccadic performance. Two experiments were performed, one in which stimuli were presented at 20 angular positions with a constant eccentricity of 12°, and another using 18 possible eccentricities along the horizontal. Pro-saccades had faster SRTs and velocities, larger amplitudes, higher accuracy and less variation in their trajectories than anti-saccades. Pro- and anti-saccade performance was shown to exhibit a similar dependence on both saccade goal direction and eccentricity. Differences manifested as a generalized reduction in anti-saccade performance that can be described as a scalar multiple of pro-saccade performance at all locations. Possible causes of this reduced performance were speculated to be i) the involvement of higher cortical structures, ii) errors in the internal representation of the stimulus, iii) sensorimotor coordinate transformation inaccuracy, and iv) online updating of the motor plan and the speed accuracy trade off inherent to saccades. The results of this study are comparable to previous monkey and human studies however certain differences were found that require further investigation. Further investigation is also required to determine the validity of the possible causes of performance reduction in the anti-saccade task and their specific contributions. / Thesis (Master, Neuroscience Studies) -- Queen's University, 2011-09-08 16:31:18.398

No description available.

We study the hadronic phase of Quantum Chromodynamics within a formalism that uses strongly interacting resonances in the Regge theory approach. Our aim is to investigate the properties of the medium created in a relativistic nucleus-nucleus collision, using the electromagnetic decays of vector mesons as probe. To do so, we first layout the foundations of Regge theory, which is then used to calculate the in-medium properties of vector mesons. This methodology will be used to calculate for the first time the properties of the &phis; meson in a hadronic medium. The properties are contained within the imaginary part of the propagator (IMPP). Since it is related to electromagnetic decay via vector meson dominance, the IMPP is a probe of strongly interacting hadron gas created in the collision.

No description available.

No description available.

The structure of connected graphs of given size and order that have minimal algebraic connectivity is investigated. It is shown that they must consist of a chain of cliques. Moreover, an upper bound for the number of maximal cliques of size 2 or larger is derived. (author's abstract) / Series: Research Report Series / Department of Statistics and Mathematics

Session ID: TH-F3-4 / SCIS & ISIS 2010, Joint 5th International Conference on Soft Computing and Intelligent Systems and 11th International Symposium on Advanced Intelligent Systems. December 8-12, 2010, Okayama Convention Center, Okayama, Japan

No description available.

No description available.