A Novelty Detection Approach to Seizure Analysis from Intracranial EEG Andrew B. Gardner 146 pages Directed by Dr. George Vachtsevanos and Dr. Brian Litt A framework for support vector machine classification of time series events is proposed and applied to analyze physiological signals recorded from epileptic patients. In contrast to previous works, this research formulates seizure analysis as a novelty detection problem which allows seizure detection and prediction to be treated uniformly, in a way that is capable of accommodating multichannel and/or multimodal measurements. Theoretical properties of the support vector machine algorithm employed provide a straightforward means for controlling the false alarm rate of the detector. The resulting novelty detection system was evaluated both offline and online on a corpus of 1077 hours of intracranial electroencephalogram (IEEG) recordings from 12 patients diagnosed with medically resistant temporal lobe epilepsy during evaluation for epilepsy surgery. These patients collectively had 118 seizures during the recording period. The performance of the novelty detection framework was assessed with an emphasis on four key metrics: (1) sensitivity (probability of correct detection), (2) mean detection latency, (3) early-detection fraction (prediction or detection of seizure prior to electrographic onset), and (4) false positive rate. Both the offline and online novelty detectors achieved state-of-the-art seizure detection performance. In particular, the online detector achieved 97.85% sensitivity, -13.3 second latency, and 40% early-detection fraction at an average of 1.74 false positive predictions per hour (Fph). These results demonstrate that a novelty detection approach is not only feasible for seizure analysis, but it improves upon the state-of-the-art as an effective, robust technique. Additionally, an extension of the basic novelty detection framework demonstrated its use as a simple, effective tool for examining the spread of seizure onsets. This may be useful for automatically identifying seizure focus channels in patients with focal epilepsies. It is anticipated that this research will aid in localizing seizure onsets, and provide more efficient algorithms for use in a real device.

Novelty and distinctive character are the conditions of protection for a registered design in Turkey. This study investigates the ways in which novelty and distinctive character are interpreted and assessed by the parties actively involved in the assessment process &ndash / judges, court experts, the Turkish Patent Institute experts, attorneys and design agents. The face-to-face interviews conducted with 51 participants from the parties involved indicate that the assessment of distinctive character presents more challenges than the assessment of novelty. Not being exactly the same with a previous design is the main consideration in the assessment of novelty. The assessment of distinctive character is primarily identified with the comparison of designs&rsquo / overall impressions whereas designer&rsquo / s degree of freedom remains as the least mentioned consideration. The study suggests that being subject to protection of designs not fulfilling the conditions of protection coupled with the uncertainties in the assessment of novelty and distinctive character, which may have a negative impact on the public trust in the design registration system. The study concludes with a set of suggestions for developing an assessment guide for the design registration system in Turkey.

Events that violate expectations are biologically significant and accordingly elicit various physiological responses. We investigated the functional relationship between three of these responses: the P300, the Novelty P3 and the pupil dilation response (PDR), with a particular focus on their co-variance with reaction time and measures of subsequent memory. In a modified Novelty P3 oddball paradigm, participants semantically categorized a sequence of stimuli including (1) words of a frequent category, (2) words of an infrequent category (14% of the trials) and (3) pictures of the frequent category (14% of the trials). The Novelty P3 oddball task was followed by a recall- and a recognition test. Larger amplitudes of the P300, identified by a spatial principal component analysis (PCA), were associated with enhanced subsequent recall as well as faster reaction times during the recognition test, suggesting a close relationship between the cognitive process indexed by the P300 and memory encoding. The PDR was larger for infrequents (which required a response switch) than both frequents and pictures (which did not require a switch). Furthermore, its latency was correlated with reaction time on the same trial and with reaction time on the immediately following trial. There was only weak evidence for a correlation with subsequent memory, suggesting that the cognitive process associated with the PDR might be a direct link in the stimulus-response stream. Larger Novelty P3 amplitudes were associated with both faster reaction times on the same trial and stronger memory traces, suggesting that its amplitude might index resource allocation. These findings suggest that each of the physiological responses carries a distinct functional significance in detecting, processing, or responding to novel events, and we discuss the findings in the light of the prevalent theories of the functional significance of each response.

One unresolved issue in early word learning research is the relationship between word learning, categorization, and attention. Two distinct cognitive processes, attentional preferences related to categorical processing and inter-modal matching are involved in this relationship. Keeping the effects of these processes separate and controlled can be a difficult task. Not doing so can potentially confound the interpretation of research in this area. In a series of four preferential looking studies, the effects of referential assignment and novelty seeking in infancy were teased apart. In Study 1, 13-month olds preferred to look toward a monitor on which the stimuli changed category on every trial, and away from a monitor on which the stimuli were drawn from a single category. This preference developed in conditions in which infants listened to labels, non-language sound, or participated in silence. In Study 2, 18-month-olds developed the same preference when listening to non-language sounds or when participating in silence, but developed no preference when listening to labels. Results of studies 3 and 4 suggest that the lack of preference by 18-month-olds in the label condition result from competing behaviors of novelty seeking and referential assignment. / text

Adolescence is a period of development that is associated with increased risk taking behaviors and experimenting with drugs of abuse, including alcohol and nicotine. Early onset of use of these agents may be associated with long-term changes in behavior and enhanced sensitivity to the subsequent effects of alcohol in adulthood. The present experiment was designed to assess the long-term behavioral alterations that occur due to adolescent exposure to ethanol and nicotine, either alone or in combination, on adulthood responsivity to the rewarding properties of environmental cues paired with ethanol. It was hypothesized that adolescent rats exposed to the combination of ethanol and nicotine would exhibit enhanced novelty seeking behaviors in adulthood. When assessing the rewarding properties of environmental cues paired with ethanol in adulthood using the CPP paradigm, it was hypothesized that adolescent rats exposed to the combination of a moderate dose of alcohol (0.75 g/kg) and nicotine (0.4 mg/kg) would more readily acquire a CPP in adulthood as compared to animals exposed to either drug alone. However, no changes in novelty seeking behaviors or conditioned place preference in adulthood were observed due to exposure to ethanol and/or nicotine during adolescence .Methodological considerations are discussed. Currently, other experiments are being conducted to assess the effects of nicotine on voluntary ethanol treatment in adolescent and adult male rats.

During adolescent brain maturation, there are likely sensitive periods where environmental conditions, including drug exposure, may influence development by modifying neuronal connections. Altering neuronal function may produce different phenotypes than expected under normal conditions that may influence subsequent responding to drugs of abuse after the brain is fully mature. Experiment one investigated the relationship between novelty preference and cocaine place preference in adolescent and adult rats. High responding adolescent rats displaying greater free choice novelty exploration (but not forced novelty locomotion) expressed decreased cocaine place conditioning compared to low responding rats. No relationship was found in adult rats. Experiment two evaluated novelty-induced behaviors in adulthood after adolescent cocaine exposure. Repeated cocaine administration produced greater stress and anxiogenic behavioral responses to novelty in adult rats. Repeated alcohol administration produced less-inhibited novelty-induced behaviors in adulthood. Experiment three and four evaluated the consequence of developmental cocaine exposure on the rewarding efficacy of cocaine in adolescence and adulthood. Additionally, the interaction of D2 receptors and the rewarding efficacy of cocaine were investigated. After developmental cocaine exposure, adolescent and adult rats demonstrate decreased rewarding efficacy to cocaine. Importantly, blockade of the D2 receptor prevents cocaine-induced neurochemical changes, potentially regulating the behavioral and neurochemical alterations that occur after repeated drug use that increases the likelihood of dependence. Together, these data implicate both short and long-term behavioral adaptations that occur after developmental cocaine exposure that may result in a predisposition to develop adulthood drug dependence.

This study sought to determine the roles of D1 and D2 receptors in mediating gambling reinforcement in pathological gamblers and controls (n=24/group), and the influence of reward novelty on these effects. Subjects received D2 antagonist, haloperidol (3mg), or D1-D2 antagonist, fluphenazine (3mg) in a placebo-controlled, counterbalanced, two-session design. Incentive motivation and hedonic impact were assessed before and after a 15-min slot machine game. Haloperidol tended to increase pre-game motivation but reduce the priming effect of the slot machine, while fluphenazine increased positive mood ratings but reduced motivation to gamble. Haloperidol effects were stronger when it was received on the first session, while fluphenazine had stronger effects after prior drug-free exposure. Results suggest D1 signaling is central to reward expectancy and motivation to gamble, and that moderate stimulation increases positive affect while reducing motivation to gamble. D1 blockade may also enhance reinforcement of a familiar task by interfering with reward expectancy.

This study sought to determine the roles of D1 and D2 receptors in mediating gambling reinforcement in pathological gamblers and controls (n=24/group), and the influence of reward novelty on these effects. Subjects received D2 antagonist, haloperidol (3mg), or D1-D2 antagonist, fluphenazine (3mg) in a placebo-controlled, counterbalanced, two-session design. Incentive motivation and hedonic impact were assessed before and after a 15-min slot machine game. Haloperidol tended to increase pre-game motivation but reduce the priming effect of the slot machine, while fluphenazine increased positive mood ratings but reduced motivation to gamble. Haloperidol effects were stronger when it was received on the first session, while fluphenazine had stronger effects after prior drug-free exposure. Results suggest D1 signaling is central to reward expectancy and motivation to gamble, and that moderate stimulation increases positive affect while reducing motivation to gamble. D1 blockade may also enhance reinforcement of a familiar task by interfering with reward expectancy.

The explosive growth of the World-Wide-Web and the emergence of ecommerce are the major two factors that have led to the development of recommender systems (Resnick and Varian, 1997). The main task of recommender systems is to learn from users and recommend items (e.g. information, products or books) that match the users’ personal preferences. Recommender systems have been an active research area for more than a decade. Many different techniques and systems with distinct strengths have been developed to generate better quality recommendations. One of the main factors that affect recommenders’ recommendation quality is the amount of information resources that are available to the recommenders. The main feature of the recommender systems is their ability to make personalised recommendations for different individuals. However, for many ecommerce sites, it is difficult for them to obtain sufficient knowledge about their users. Hence, the recommendations they provided to their users are often poor and not personalised. This information insufficiency problem is commonly referred to as the cold-start problem. Most existing research on recommender systems focus on developing techniques to better utilise the available information resources to achieve better recommendation quality. However, while the amount of available data and information remains insufficient, these techniques can only provide limited improvements to the overall recommendation quality. In this thesis, a novel and intuitive approach towards improving recommendation quality and alleviating the cold-start problem is attempted. This approach is enriching the information resources. It can be easily observed that when there is sufficient information and knowledge base to support recommendation making, even the simplest recommender systems can outperform the sophisticated ones with limited information resources. Two possible strategies are suggested in this thesis to achieve the proposed information enrichment for recommenders: • The first strategy suggests that information resources can be enriched by considering other information or data facets. Specifically, a taxonomy-based recommender, Hybrid Taxonomy Recommender (HTR), is presented in this thesis. HTR exploits the relationship between users’ taxonomic preferences and item preferences from the combination of the widely available product taxonomic information and the existing user rating data, and it then utilises this taxonomic preference to item preference relation to generate high quality recommendations. • The second strategy suggests that information resources can be enriched simply by obtaining information resources from other parties. In this thesis, a distributed recommender framework, Ecommerce-oriented Distributed Recommender System (EDRS), is proposed. The proposed EDRS allows multiple recommenders from different parties (i.e. organisations or ecommerce sites) to share recommendations and information resources with each other in order to improve their recommendation quality. Based on the results obtained from the experiments conducted in this thesis, the proposed systems and techniques have achieved great improvement in both making quality recommendations and alleviating the cold-start problem.

Doctor of Philosophy(PhD) / The main aim of this thesis was to examine whether learning processes occur in fluid intelligence (Gf) tasks, whether it is essential for them to occur for induction to take place and whether they contribute to individual differences in performance. In mainstream differential research, Gf is conceptualised as a factor important in induction tasks that are considered novel and context-free (Cattell, 1963, 1987). Thus, performance has typically been assumed to be uninfluenced by previous acquisitions of knowledge structures. Sources of individual differences in Gf task performance have been attributed to working memory capacity (WMC), particularly individual differences in the ability to combat proactive interference. In contrast, the cognitive reasoning literature associates induction with the use of prior conceptual knowledge. A middle-ground position is that Gf tasks may require learning to occur across the task, which would draw upon WMC. That is, individual differences in Gf task performance may be due to knowledge learnt across the task, rather than knowledge brought to the task. Gf items have traditionally been presented in easy-to-hard order but easier items may unintentionally provide learning opportunity for harder items. This would contradict both classic and modern test theories which make the assumption that items within a task are independent of each other. The learning hypothesis was explored in the current work along with the issue of whether it is possible to reliably solve complex Gf items without some relevant, prior knowledge. Also, the distinction between within-item induction and across-item learning was investigated, along with the relationship between across-item learning and proactive interference. An experimental-differential approach was used to manipulate learning opportunity within Gf tasks in four experiments. The first experiment examined whether learning takes place in Raven’s Advanced Progressive Matrices (Raven, 1962) and if so, to what extent this learning is a source of individual differences. Specifically, whether rule learning within the task is necessary for abstraction to take place and whether those of higher Gf ability learn faster than those of lower Gf ability. The next three experiments examined the distinction between knowledge that may be brought to the task, learning that occurs across multiple items in the task and induction within a single item that may be independent of any prior knowledge including knowledge learnt across the task. The effect of proactive interference as a consequence of learning and knowledge was also investigated. The experiments examined which of these are relevant to general performance (i.e., common to everyone) and which contribute to individual differences. Learning-opportunity was manipulated in a task from the cognitive reasoning literature – the Modified Sweller and Gee (MSG) Task. Traditional Series Completion tasks were used as Gf markers and data analyses employed Hierarchical Linear Modelling (HLM). The advantage of the MSG Task is that it has qualities typical of Gf tasks but unlike conventional Gf tasks, it is able to assess within-item induction in isolation from any potential influences from across-item learning. This is because it involves multiple attempts within each item with feedback, allowing single items to be administered reliably. When across-item learning opportunity is absent, the MSG Task is able to provide an estimate of participants’ within-item induction success through the number of attempts they need within a single item. The amount participants learn across items can be approximated by comparing performance on items preceded by learning opportunity (i.e., easier items with similar rule-types), with items not preceded by learning opportunity. Lastly, the effects of proactive interference can be evaluated by comparing performance on items preceded by interference (i.e., items with different rule-types) with those that are not preceded by interference. Overall, it was found that with no learning opportunity leading up to novel items (to provide relevant prior knowledge), solution was nearly impossible for all participants. When learning opportunity was provided, all participants were able to greatly improve their performance but those of higher Gf improved more. It was concluded that while Gf tasks appear visually novel, they must contain a combination of familiar elements in earlier items (which make use of knowledge that participants bring to the task) and novel elements in later items (which require the use of knowledge that must be learnt from earlier items); and those of higher Gf perform better on Gf tasks, at least partly because they are able to benefit more from the learning opportunity provided by earlier items. It was found that proactive interference affects all participants when they learn from prior items. However, insufficient evidence was found to suggest that the ability to combat proactive interference contributes to individual differences in performance.