Scene Classification Using a Hybrid Generative/Discriminative Approach

We investigate whether dimensionality reduction using a latent generative model is beneficial for the task of weakly supervised scene classification. In detail, we are given a set of labeled images of scenes (for example, coast, forest, city, river, etc.), and our objective is to classify a new image into one of these categories. Our approach consists of first discovering latent ";topics"; using probabilistic Latent Semantic Analysis (pLSA), a generative model from the statistical text literature here applied to a bag of visual words representation for each image, and subsequently, training a multiway classifier on the topic distribution vector for each image. We compare this approach to that of representing each image by a bag of visual words vector directly and training a multiway classifier on these vectors. To this end, we introduce a novel vocabulary using dense color SIFT descriptors and then investigate the classification performance under changes in the size of the visual vocabulary, the number of latent topics learned, and the type of discriminative classifier used (k-nearest neighbor or SVM). We achieve superior classification performance to recent publications that have used a bag of visual word representation, in all cases, using the authors' own data sets and testing protocols. We also investigate the gain in adding spatial information. We show applications to image retrieval with relevance feedback and to scene classification in videos

Use of decision trees in colour feature selection: application to object recognition in outdoor scenes

A new method for the automated selection of colour features is described. The algorithm consists of two stages of processing. In the first, a complete set of colour features is calculated for every object of interest in an image. In the second stage, each object is mapped into several n-dimensional feature spaces in order to select the feature set with the smallest variables able to discriminate the remaining objects. The evaluation of the discrimination power for each concrete subset of features is performed by means of decision trees composed of linear discrimination functions. This method can provide valuable help in outdoor scene analysis where no colour space has been demonstrated as being the most suitable. Experiment results recognizing objects in outdoor scenes are reported

Monocular-based 3-D seafloor reconstruction and ortho-mosaicing by piecewise planar representation

Photo-mosaicing techniques have become popular for seafloor mapping in various marine science applications. However, the common methods cannot accurately map regions with high relief and topographical variations. Ortho-mosaicing borrowed from photogrammetry is an alternative technique that enables taking into account the 3-D shape of the terrain. A serious bottleneck is the volume of elevation information that needs to be estimated from the video data, fused, and processed for the generation of a composite ortho-photo that covers a relatively large seafloor area. We present a framework that combines the advantages of dense depth-map and 3-D feature estimation techniques based on visual motion cues. The main goal is to identify and reconstruct certain key terrain feature points that adequately represent the surface with minimal complexity in the form of piecewise planar patches. The proposed implementation utilizes local depth maps for feature selection, while tracking over several views enables 3-D reconstruction by bundle adjustment. Experimental results with synthetic and real data validate the effectiveness of the proposed approach

Systematic distortions of perceptual stability investigated using immersive virtual reality

Using an immersive virtual reality system, we measured the ability of observers to detect the rotation of an object when its movement was yoked to the observer's own translation. Most subjects had a large bias such that a static object appeared to rotate away from them as they moved. Thresholds for detecting target rotation were similar to those for an equivalent speed discrimination task carried out by static observers, suggesting that visual discrimination is the predominant limiting factor in detecting target rotation. Adding a stable visual reference frame almost eliminated the bias. Varying the viewing distance of the target had little effect, consistent with observers underestimating distance walked. However, accuracy of walking to a briefly presented visual target was high and not consistent with an underestimation of distance walked. We discuss implications for theories of a task-independent representation of visual space. © 2005 Elsevier Ltd. All rights reserved.

Human factors in cargo X-ray examination systems

The project investigated factors which may determine operator performance in X-ray screening of cargo, which is of similar importance to aviation security screening. A target detection paradigm showed significant difference in several measures of performance between baggage and postal contexts which provides evidence that these may constitute separate expertise domains.

Desempenho de escolares com e sem dificuldades de aprendizagem de ensino particular em habilidade fonológica, nomeação rápida, leitura e escrita

OBJETIVOS: caracterizar e comparar o desempenho de escolares com e sem dificuldades de aprendizagem no ensino particular em habilidades fonológicas, nomeação rápida, leitura e escrita. MÉTODOS: participaram desse estudo 60 escolares de 2ª a 4ª séries de escola de ensino particular, distribuídos em 6 grupos, sendo cada grupo composto por 10 escolares, sendo 3 grupos de escolares com dificuldades de aprendizagem e 3 grupos de escolares sem dificuldades de aprendizagem. Como procedimentos, foram realizadas a prova de nomeação automática rápida, a de consciência fonológica e a prova de leitura oral e escrita sob ditado. RESULTADOS: os resultados desse estudo evidenciaram desempenho superior dos escolares sem dificuldades de aprendizagem em relação àqueles com dificuldades. Os escolares com dificuldades de aprendizagem apresentaram maior relação velocidade/tempo em tarefas de nomeação e, conseqüentemente, desempenho inferior em tarefas de consciência fonológica e leitura e escrita de palavras isoladas quando comparados aos sem dificuldades de aprendizagem. CONCLUSÃO: os escolares com dificuldades de aprendizagem apresentaram comprometimento na relação entre as capacidades de nomeação e automatização dos estímulos apresentados com a capacidade de acesso lexical, discriminação visual, freqüência de uso dos estímulos e competição para a apresentação do menor tempo possível na nomeação dos códigos necessários para o estabelecimento do mecanismo de conversão fonema-grafema, exigido para a realização da leitura e escrita em um sistema alfabético como o português.

Desempenho de escolares em consciência fonológica, nomeação rápida, leitura e escrita

Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP)

A place preference test in the fish Nile tilapia

The Nile tilapia fish (Oreochromis niloticus) has a high potential to be used as a model in neuroscience studies. In the present study, the preference of the Nile tilapia between a gravel-enriched (GEE), a shelter-enriched (SEE) or a non-enriched (NEE) environment was determined, for developing a place preference model. Nile tilapia had an initial preference for GEE, but after 1 day of observation, the fish stabilized their frequency of visits among compartments. Hence, any stimulus motivating tilapia increase in compartment visiting indicates a positively reinforcing effect. This feature is very useful for the development of new behavioural paradigms for fish in tests using environmental discrimination, such as the conditioning place preference test. © 2006 Elsevier GmbH. All rights reserved.

Elaboração de programa de intervenção com as habilidades percepto-viso-motoras em escolares com dislexia do desenvolvimento

Pós-graduação em Fonoaudiologia - FFC

Formação de classes funcionais em cães domésticos (Canis familiaris)

O presente estudo investigou a possibilidade de ensinar dois cães domésticos (Canis familiaris) a responderem adequadamente a duas classes de três estímulos (caixas contendo pedaço de alimento quando com função positiva) através de um procedimento de mudanças sucessivas de discriminação simples (MSDS). A coleta de dados foi realizada em uma sala na qual duas, quatro ou seis caixas de madeira eram dispostas lado a lado, em posições fixas. Inicialmente, os sujeitos responderam a dois estímulos por sessão (um positivo[S+] e outro negativo[S-]), até que quatro das seis caixas apareciam com ambas as funções. Na fase seguinte, as quatro caixas foram apresentadas juntas, de maneira que duas caixas de uma classe eram S+ e as outras duas S-. Foram realizadas mudanças de função até que os sujeitos alcançassem o critério de aprendizagem em 11 de 12 sessões. Em seguida os sujeitos foram expostos a uma sessão ao reforçamento de respostas nas duas caixas ainda não utilizadas e depois a um treino de discriminação simples com as seis caixas. Foi realizado um teste consistindo em uma sessão de mudança com quatro caixas, após a qual eram re-inseridas as duas caixas restantes na sala experimental. Com os resultados negativos obtidos nesse teste, foi realizado um treino de MSDS com as seis caixas, após o qual o teste foi refeito. Ambos os sujeitos responderam adequadamente a todas as tentativas. Outros cinco testes foram realizados com as demais caixas, no mesmo formato do primeiro teste. Um sujeito respondeu adequadamente a dois desses cinco testes e o outro a três deles. Discute-se os efeitos da ênfase na discriminação visual e o uso de posições fixas das caixas sobre o desempenho dos sujeitos.

Desempenho de escolares de ensino público com e sem dificuldade de aprendizagem em habilidadefonológica, nomeação rápida, leitura e escrita

Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP)

Memory-enhancing effects of secreted forms of the β-amyloid precursor protein in normal and amnestic mice

When administered intracerebroventricularly to mice performing various learning tasks involving either short-term or long-term memory, secreted forms of the β-amyloid precursor protein (APPs751 and APPs695) have potent memory-enhancing effects and block learning deficits induced by scopolamine. The memory-enhancing effects of APPs were observed over a wide range of extremely low doses (0.05-5,000 pg intracerebroventricularly), blocked by anti-APPs antisera, and observed when APPs was administered either after the first training session in a visual discrimination or a lever-press learning task or before the acquisition trial in an object recognition task. APPs had no effect on motor performance or exploratory activity. APPs695 and APPs751 were equally effective in the object recognition task, suggesting that the memory-enhancing effect of APPs does not require the Kunitz protease inhibitor domain. These data suggest an important role for APPss on memory processes.

Motion perception: seeing and deciding.

The primate visual system offers unprecedented opportunities for investigating the neural basis of cognition. Even the simplest visual discrimination task requires processing of sensory signals, formation of a decision, and orchestration of a motor response. With our extensive knowledge of the primate visual and oculomotor systems as a base, it is now possible to investigate the neural basis of simple visual decisions that link sensation to action. Here we describe an initial study of neural responses in the lateral intraparietal area (LIP) of the cerebral cortex while alert monkeys discriminated the direction of motion in a visual display. A subset of LIP neurons carried high-level signals that may comprise a neural correlate of the decision process in our task. These signals are neither sensory nor motor in the strictest sense; rather they appear to reflect integration of sensory signals toward a decision appropriate for guiding movement. If this ultimately proves to be the case, several fascinating issues in cognitive neuroscience will be brought under rigorous physiological scrutiny.

Development of an oligonucleotide-based SNP detection method on lateral flow strips using hexapet tages

Detection of point mutations or single nucleotide polymorphisms (SNPs) is important in relation to disease susceptibility or detection in pathogens of mutations determining drug resistance or host range. There is an emergent need for rapid detection methods amenable to point-of-care applications. The purpose of this study was to reduce to practice a novel method for SNP detection and to demonstrate that this technology can be used downstream of nucleic acid amplification. The authors used a model system to develop an oligonucleotide-based SNP detection system on nitrocellulose lateral flow strips. To optimize the assay they used cloned sequences of the herpes simplex virus-1 (HSV-1) DNA polymerase gene into which they introduced a point mutation. The assay system uses chimeric polymerase chain reaction (PCR) primers that incorporate hexameric repeat tags ("hexapet tags"). The chimeric sequences allow capture of amplified products to predefined positions on a lateral flow strip. These "hexapet" sequences have minimal cross-reactivity and allow specific hybridization-based capture of the PCR products at room temperature onto lateral flow strips that have been striped with complementary hexapet tags. The allele-specific amplification was carried out with both mutant and wild-type primer sets present in the PCR mix ("competitive" format). The resulting PCR products carried a hexapet tag that corresponded with either a wild-type or mutant sequence. The lateral flow strips are dropped into the PCR reaction tube, and mutant sequence and wild-type sequences diffuse along the strip and are captured at the corresponding position on the strip. A red line indicative of a positive reaction is visible after 1 minute. Unlike other systems that require separate reactions and strips for each target sequence, this system allows multiplex PCR reactions and multiplex detection on a single strip or other suitable substrates. Unambiguous visual discrimination of a point mutation under room temperature hybridization conditions was achieved with this model system in 10 minutes after PCR. The authors have developed a capture-based hybridization method for the detection and discrimination of HSV-1 DNA polymerase genes that contain a single nucleotide change. It has been demonstrated that the hexapet oligonucleotides can be adapted for hybridization on the lateral flow strip platform for discrimination of SNPs. This is the first step in demonstrating SNP detection on lateral flow using the hexapet oligonucleotide capture system. It is anticipated that this novel system can be widely used in point-of-care settings.

Quantum of Vision

Visual inputs to artificial and biological visual systems are often quantized: cameras accumulate photons from the visual world, and the brain receives action potentials from visual sensory neurons. Collecting more information quanta leads to a longer acquisition time and better performance. In many visual tasks, collecting a small number of quanta is sufficient to solve the task well. The ability to determine the right number of quanta is pivotal in situations where visual information is costly to obtain, such as photon-starved or time-critical environments. In these situations, conventional vision systems that always collect a fixed and large amount of information are infeasible. I develop a framework that judiciously determines the number of information quanta to observe based on the cost of observation and the requirement for accuracy. The framework implements the optimal speed versus accuracy tradeoff when two assumptions are met, namely that the task is fully specified probabilistically and constant over time. I also extend the framework to address scenarios that violate the assumptions. I deploy the framework to three recognition tasks: visual search (where both assumptions are satisfied), scotopic visual recognition (where the model is not specified), and visual discrimination with unknown stimulus onset (where the model is dynamic over time). Scotopic classification experiments suggest that the framework leads to dramatic improvement in photon-efficiency compared to conventional computer vision algorithms. Human psychophysics experiments confirmed that the framework provides a parsimonious and versatile explanation for human behavior under time pressure in both static and dynamic environments.