Using Variational Multi-view Learning for Classification of Grocery Items.
ABSTRACT: An essential task for computer vision-based assistive technologies is to help visually impaired people to recognize objects in constrained environments, for instance, recognizing food items in grocery stores. In this paper, we introduce a novel dataset with natural images of groceries-fruits, vegetables, and packaged products-where all images have been taken inside grocery stores to resemble a shopping scenario. Additionally, we download iconic images and text descriptions for each item that can be utilized for better representation learning of groceries. We select a multi-view generative model, which can combine the different item information into lower-dimensional representations. The experiments show that utilizing the additional information yields higher accuracies on classifying grocery items than only using the natural images. We observe that iconic images help to construct representations separated by visual differences of the items, while text descriptions enable the model to distinguish between visually similar items by different ingredients.
Project description:OBJECTIVE:The present study aimed to determine the store types from which people in low-income neighbourhoods purchase most sugar-sweetened beverages (SSB) and to identify associations between purchasing location and demographic characteristics. DESIGN:Street-intercept surveys of passers-by near high foot-traffic intersections in 2016. Participants completed a beverage frequency questionnaire and identified the type of store (e.g. corner store, chain grocery) from which they purchased most SSB. SETTING:Eight low-income neighbourhoods in four Bay Area cities, California, USA.ParticipantsSample of 1132 individuals who reported consuming SSB, aged 18-88 years, who identified as African-American (41 %), Latino (29 %), White (17 %) and Asian (6 %). RESULTS:Based on surveys in low-income neighbourhoods, corner stores were the primary source from which most SSB were purchased (28 %), followed by discount stores (18 %) and chain groceries (16 %). In fully adjusted models, those with lower education were more likely to purchase from corner stores or discount groceries than all other store types. Compared with White participants, African-Americans purchased more frequently from corner stores, discount groceries and chain groceries while Latinos purchased more frequently from discount groceries. CONCLUSIONS:The wide range of store types from which SSB were purchased and demographic differences in purchasing patterns suggest that broader methodological approaches are needed to adequately capture the impact of SSB taxes and other interventions aimed at reducing SSB consumption, particularly in low-income neighbourhoods.
Project description:<h4>Introduction</h4>Research on urban food environments emphasizes limited access to healthy food, with fewer large supermarkets and higher food prices. Many residents of Hartford, Connecticut, which is often considered a food desert, buy most of their food from small and medium-sized grocery stores. We examined the food environment in greater Hartford, comparing stores in Hartford to those in the surrounding suburbs, and by store size (small, medium, and large).<h4>Methods</h4>We surveyed all small (over 1,000 ft2), medium, and large-sized supermarkets within a 2-mile radius of Hartford (36 total stores). We measured the distance to stores, availability, price and quality of a market basket of 25 items, and rated each store on internal and external appearance. Geographic Information System (GIS) was used for mapping distance to the stores and variation of food availability, quality, and appearance.<h4>Results</h4>Contrary to common literature, no significant differences were found in food availability and price between Hartford and suburban stores. However, produce quality, internal, and external store appearance were significantly lower in Hartford compared to suburban stores (all p<0.05). Medium-sized stores had significantly lower prices than small or large supermarkets (p<0.05). Large stores had better scores for internal (p<0.05), external, and produce quality (p<0.01). Most Hartford residents live within 0.5 to 1 mile distance to a grocery store.<h4>Discussion</h4>Classifying urban areas with few large supermarkets as 'food deserts' may overlook the availability of healthy foods and low prices that exist within small and medium-sized groceries common in inner cities. Improving produce quality and store appearance can potentially impact the food purchasing decisions of low-income residents in Hartford.
Project description:Working memory impairment is considered a core deficit in schizophrenia, but the precise nature of this deficit has not been determined. Multiple lines of evidence implicate deficits at the encoding stage. During encoding, information is held in a precategorical sensory store termed iconic memory, a literal image of the stimulus with high capacity but rapid decay. Pathologically increased iconic decay could reduce the number of items that can be transferred into working memory before the information is lost and could thus contribute to the working memory deficit seen in the illness. The current study used a partial report procedure to test the hypothesis that patients with schizophrenia (n = 37) display faster iconic memory decay than matched healthy control participants (n = 28). Six letters, arranged in a circle, were presented for 50 ms. Following a variable delay of 0-1000 ms, a central arrow cue indicated the item to be reported. In both patients and control subjects, recall accuracy decreased with increasing cue delay, reflecting decay of the iconic representation of the stimulus array. Patients displayed impaired memory performance across all cue delays, consistent with an impairment in working memory, but the rate of iconic memory decay did not differ between patients and controls. This provides clear evidence against faster loss of iconic memory representations in schizophrenia, ruling out iconic decay as an underlying source of the working memory impairment in this population. Thus, iconic decay rate can be added to a growing list of unimpaired cognitive building blocks in schizophrenia.
Project description:Iconic representations are ubiquitous; they fill children's cartoons, add humor to newspapers, and bring emotional tone to online communication. Yet, the communicative function they serve remains unaddressed by cognitive psychology. Here, we examined the hypothesis that iconic representations communicate emotional information more efficiently than their realistic counterparts. In Experiment 1, we manipulated low-level features of emotional faces to create five sets of stimuli that ranged from photorealistic to fully iconic. Participants identified emotions on briefly presented faces. Results showed that, at short presentation times, accuracy for identifying emotion on more "cartoonized" images was enhanced. In addition, increasing contrast and decreasing featural complexity benefited accuracy. In Experiment 2, we examined an event-related potential component, the P1, which is sensitive to low-level visual stimulus features. Lower levels of contrast and complexity within schematic stimuli were also associated with lower P1 amplitudes. These findings support the hypothesis that iconic representations differ from realistic images in their ability to communicate specific information, including emotion, quickly and efficiently, and that this effect is driven by changes in low-level visual features in the stimuli.
Project description:The ability to search for and precisely compare similar phenotypic appearances within and across species has vast potential in plant science and genetic research. The difficulty in doing so lies in the fact that many visual phenotypic data, especially visually observed phenotypes that often times cannot be directly measured quantitatively, are in the form of text annotations, and these descriptions are plagued by semantic ambiguity, heterogeneity, and low granularity. Though several bio-ontologies have been developed to standardize phenotypic (and genotypic) information and permit comparisons across species, these semantic issues persist and prevent precise analysis and retrieval of information. A framework suitable for the modeling and analysis of precise computable representations of such phenotypic appearances is needed.We have developed a new framework called the Computable Visually Observed Phenotype Ontological Framework for plants. This work provides a novel quantitative view of descriptions of plant phenotypes that leverages existing bio-ontologies and utilizes a computational approach to capture and represent domain knowledge in a machine-interpretable form. This is accomplished by means of a robust and accurate semantic mapping module that automatically maps high-level semantics to low-level measurements computed from phenotype imagery. The framework was applied to two different plant species with semantic rules mined and an ontology constructed. Rule quality was evaluated and showed high quality rules for most semantics. This framework also facilitates automatic annotation of phenotype images and can be adopted by different plant communities to aid in their research.The Computable Visually Observed Phenotype Ontological Framework for plants has been developed for more efficient and accurate management of visually observed phenotypes, which play a significant role in plant genomics research. The uniqueness of this framework is its ability to bridge the knowledge of informaticians and plant science researchers by translating descriptions of visually observed phenotypes into standardized, machine-understandable representations, thus enabling the development of advanced information retrieval and phenotype annotation analysis tools for the plant science community.
Project description:<h4>Background</h4>Diet is an important determinant of health, and food purchasing is a key antecedent to consumption.<h4>Objective</h4>We set out to evaluate the effectiveness of grocery store interventions to change food purchasing, and to examine whether effectiveness varied based on intervention components, setting, or socioeconomic status.<h4>Design</h4>We conducted a systematic review of randomized controlled trials (search performed June 2017). Studies must have: aimed to change food purchasing; been implemented in grocery stores (real or simulated); reported purchasing; and had a minimal control or compared interventions fulfilling our criteria. Searching, screening, bias assessment, and data extraction followed Cochrane methods. We grouped studies by intervention type (economic, environmental, swaps, and/or education), synthesized results narratively, and conducted an exploratory qualitative comparative analysis.<h4>Results</h4>We included 35 studies representing 89 interventions, >20,000 participants, and >800 stores. Risk of bias was mixed. Economic interventions showed the most promise, with 8 of the 9 studies in real stores and all 6 in simulated environments detecting an effect on purchasing. Swap interventions appeared promising in the 2 studies based in real stores. Store environment interventions showed mixed effects. Education-only interventions appeared effective in simulated environments but not in real stores. Available data suggested that effects of economic interventions did not differ by socioeconomic status, whereas for other interventions impact was variable. In our qualitative comparative analysis, economic interventions (regardless of setting) and environmental and swap interventions in real stores were associated with statistically significant changes in purchasing in the desired direction for ?1 of the foods targeted by the intervention, whereas education-only interventions in real stores were not.<h4>Conclusions</h4>Findings suggest that interventions implemented in grocery stores-particularly ones that manipulate price, suggest swaps, and perhaps manipulate item availability-have an impact on purchasing and could play a role in public health strategies to improve health. Review protocol registered at https://www.crd.york.ac.uk/PROSPERO/ as CRD42017068809.
Project description:Persistent neural activity that encodes online mental representations plays a central role in working memory (WM). However, there has been debate regarding the number of items that can be concurrently represented in this active neural state, which is often called the "focus of attention." Some models propose a strict single-item limit, such that just 1 item can be neurally active at once while other items are relegated to an activity-silent state. Although past studies have decoded multiple items stored in WM, these studies cannot rule out a switching account in which only a single item is actively represented at a time. Here, we directly tested whether multiple representations can be held concurrently in an active state. We tracked spatial representations in WM using alpha-band (8-12 Hz) activity, which encodes spatial positions held in WM. Human observers remembered 1 or 2 positions over a short delay while we recorded electroencephalography (EEG) data. Using a spatial encoding model, we reconstructed active stimulus-specific representations (channel-tuning functions [CTFs]) from the scalp distribution of alpha-band power. Consistent with past work, we found that the selectivity of spatial CTFs was lower when 2 items were stored than when 1 item was stored. Critically, data-driven simulations revealed that the selectivity of spatial representations in the two-item condition could not be explained by models that propose that only a single item can exist in an active state at once. Thus, our findings demonstrate that multiple items can be concurrently represented in an active neural state.
Project description:Sodium intakes, from foods and beverages, of 22,852 persons in the National Health and Nutrition Examination Surveys (NHANES 2003-2008) were examined by specific food source and by food location of origin. Analyses were based on a single 24-h recall. Separate analyses were conducted for children (6-11 years of age), adolescents (12-19), and adults (20-50 and ?51 years). Grouping of like foods (e.g., food sources) used a scheme proposed by the National Cancer Institute, which divides foods/beverages into 96 food subgroups (e.g., pizza, yeast breads or cold cuts). Food locations of origin were stores (e.g., grocery, convenience and specialty stores), quick-service restaurant/pizza (QSR), full-service restaurant (FSR), school, or other. Food locations of sodium were also evaluated by race/ethnicity amongst adults. Stores provided between 58.1% and 65.2% of dietary sodium, whereas QSR and FSR together provided between 18.9% and 31.8% depending on age. The proportion of sodium from QSR varied from 10.1% to 19.9%, whereas that from FSR varied from 3.4% to 13.3%. School meals provided 10.4% of sodium for 6-11 year olds and 6.0% for 12-19 year olds. Pizza from QSR, the top away from home food item, provided 5.4% of sodium in adolescents. QSR pizza, chicken, burgers and Mexican dishes combined provided 7.8% of total sodium in adult diets. Most sodium came from foods purchased in stores. Food manufacturers, restaurants, and grocery stores all have a role to play in reducing the amount of sodium in the American diet.
Project description:A dominant view of prefrontal cortex (PFC) function is that it stores task-relevant information in working memory. To examine this and determine how it applies when multiple pieces of information must be stored, we trained two subjects to perform a multi-item color change detection task and recorded activity of neurons in PFC. Few neurons encoded the color of the items. Instead, the predominant encoding was spatial: a static signal reflecting the item's position and a dynamic signal reflecting the subject's covert attention. These findings challenge the notion that PFC stores task-relevant information. Instead, we suggest that the contribution of PFC is in controlling the allocation of resources to support working memory. In support of this, we found that increased power in the alpha and theta bands of PFC local field potentials, which are thought to reflect long-range communication with other brain areas, was correlated with more precise color representations.
Project description:Although negative emotion can strengthen memory of an event it can also result in memory disturbances, as in post-traumatic stress disorder (PTSD). We examined the effects of negative item content on amygdalar and hippocampal function in memory for the items themselves and for the associations between them. During fMRI, we examined encoding and retrieval of paired associates made up of all four combinations of neutral and negative images. At test, participants were cued with an image and, if recognised, had to retrieve the associated (target) image. The presence of negative images increased item memory but reduced associative memory. At encoding, subsequent item recognition correlated with amygdala activity, while subsequent associative memory correlated with hippocampal activity. Hippocampal activity was reduced by the presence of negative images, during encoding and correct associative retrieval. In contrast, amygdala activity increased for correctly retrieved negative images, even when cued by a neutral image. Our findings support a dual representation account, whereby negative emotion up-regulates the amygdala to strengthen item memory but down-regulates the hippocampus to weaken associative representations. These results have implications for the development and treatment of clinical disorders in which diminished associations between emotional stimuli and their context contribute to negative symptoms, as in PTSD.