Intuitive Large Image Database Browsing Using Perceptual Similarity Enriched by Crowds

In this article, we evaluate the effectiveness of a pre-classification scheme for the fast retrieval of faces in a large image database. The studied approach is based on a partitioning of the face space through a clustering of face images. Mainly two issues are discussed. How to perform clustering with a non-trivial probabilistic measure of similarity between faces? How to assign face images to all clusters probabilistically to form a robust characterization vector? It is shown experimentally on the FERET face database that, with this simple approach, the cost of a search can be reduced by a factor 6 or 7 with no significant degradation of the performance.

Earlier a new rough set theory for very large databases was proposed by T.Y. Lin (1996). In this paper the authors attempt to evaluate the performance of such a rough set theory for a very large database. ORACLE, a relational database management system (RDBMS), is the market leader in open system databases. Windows NT has been growing into the data server environment and is a strong contender for decision support system applications with its open and cost-effective architecture. So ORACLE running under Windows NT was used in this report. The main goal of this research is to formulate a suitable rough set theory for very large databases.

Deep learning methodologies constitute nowadays the main approach for medical image analysis and disease prediction. Large annotated databases are necessary for developing these methodologies; such databases are difficult to obtain and to make publicly available for use by researchers and medical experts. In this paper, we focus on diagnosis of Covid-19 based on chest 3-D CT scans and develop a dual knowledge framework, including a large imaging database and a novel deep neural architecture. We introduce COV19-CT-DB, a very large database annotated for COVID-19 that consists of 7,750 3-D CT scans, 1,650 of which refer to COVID-19 cases and 6,100 to non-COVID19 cases. We use this database to train and develop the RACNet architecture. This architecture performs 3-D analysis based on a CNN-RNN network and handles input CT scans of different lengths, through the introduction of dynamic routing, feature alignment and a mask layer. We conduct a large experimental study that illustrates that the RACNet network has the best performance compared to other deep neural networks i) when trained and tested on COV19-CT-DB; ii) when tested, or when applied, through transfer learning, to other public databases.

Video cameras are widely deployed along city streets, interstate highways, traffic lights, stop signs and toll booths by entities that perform traffic monitoring and law enforcement. The videos captured by these cameras are typically compressed and stored in large databases. Performing a rapid search for a specific vehicle within a large database of compressed videos is often required and can be a time-critical life or death situation. In this paper, we propose video compression and decompression algorithms that enable fast and efficient vehicle or, more generally, event searches in large video databases. The proposed algorithm selects reference frames (i.e., I-frames) based on a vehicle having been detected at a specified position within the scene being monitored while compressing a video sequence. A search for a specific vehicle in the compressed video stream is performed across the reference frames only, which does not require decompression of the full video sequence as in traditional search algorithms. Our experimental results on videos captured in a local road show that the proposed algorithm significantly reduces the search space (thus reducing time and computational resources) in vehicle search tasks within compressed video streams, particularly those captured in light traffic volume conditions.

Large human motion databases contain variants of natural motions that are valuable for animation generation and synthesis. But retrieving visually similar motions is still a difficult and time-consuming problem. This paper provides methods for identifying visually and numerically similar motions in a large database given a query of motion segment. We propose an efficient indexing strategy that represents the motions compactly through a preprocessing. This representation scales down the range of searching the database. Motions in this range are possible candidates of the final matches. For detailed comparisons between the query and the candidates, we propose an algorithm that compares the motions’ curves swiftly. Our methods can apply to large human motion databases and achieve high performance and accuracy compared with previous work. We present experimental results on testing a database of about 2.9 million frames, or about 27 hours of motions played at 30 Hz.

Processing large databases is intricate for databases involving several types of textures. In particular, for content-based image retrieval, a query has to be compared with all the samples pertaining to the database in order to identify its content/class and this is time consuming. Furthermore, modeling of a large database is a difficult task for databases involving several types of textures since accurate models for certain textures are not guaranteed to be very relevant for other types of textures and vice-versa. In order to save computational time and increase performance in processing of large texture databases, the present paper proposes structuring texture databases by using stochasticity metaclasses.

Mining association rules in large databases is one of the most interesting data mining techniques in database communities. The number of association rules in a large database highly depends on rule’s support and confidence, and there might exist hundreds of rules in a very large database. The presentation of large number of rules in a very nice and noticeable way becomes highly challenging. In recent years researchers have developed several tools to visualize association rules. However, a large number of the tools cannot handle more than dozens of association rules. Furthermore, none of them can effectively manage association rules with multiple antecedents. Till now a uniform descriptive presentation technique has not been set up yet. We studied existing descriptive techniques in the context of visualization and introduced a graph-based technique as Unified Descriptive Language for Association Rules. The proposed technique can be used to extract the discovered rules of a very large database very conveniently and efficiently.

The purpose of this article was to discuss common issues associated with large databases and present possible solutions to improve the quality and usefulness of large database research. The volume of electronic healthcare-related data is growing exponentially. Some of these data are being stored in registries and administrative databases. These data repositories are increasingly common and can serve as sources of nurse-driven research and quality improvement activities. Although these large databases have a wealth of useful information, they have limitations that may bias results. These include missing data and cases, data accuracy and validity, and the statistical effect of large samples. Researchers using large databases to address quality, safety, clinical, or systems issues have a variety of available techniques to deal with data issues. Proper data cleaning activities such as screening, visualization, and outlier/inlier identification are essential for addressing inaccurate values within large data sets. Common methods for addressing missing data include case analyses and various imputation techniques. Statistical approaches such as risk reductions and effect size are also useful when working with large sample sizes. Registries and administrative databases provide healthcare researchers with increasing opportunities to address a wide variety of important practice and patient care questions. Healthcare researchers are encouraged to explore large data sets as they look for ways to improve patient safety and quality care, develop evidence-based practice guidelines, and fulfill regulatory and accreditation requirements.

A novel self-organizing map (SOM) based retrieval system is proposed for performing face matching in large database. The proposed system provides a small subset of faces that are most similar to a given query face, from which user can easily verify the matched images. The architecture of the proposed system consists of two major parts. First, the system provides a generalized integration of multiple feature-sets using multiple self-organizing maps. Multiple feature-sets are obtained from different feature extraction methods like Gabor filter, Local Autocorrelation Coefficients, etc. In this platform, multiple facial features are integrated to form a compressed feature vector without concerning scaling and length of individual feature set. Second, an SOM is trained to organize all the face images in a database through using the compressed feature vector. Using the organized map, similar faces to a query can be efficiently identified. Furthermore, the system includes a relevance feedback to enhance the face retrieval performance. The proposed method is computationally efficient. Comparative results show that the proposed approach is promising for identifying face in a given large image database.

BackgroundAlthough many studies have modelled the national burdens of hospitalizations and deaths due to influenza, few studies have considered the outpatient burden. To fill this gap for the United States (US), we applied traditional statistical modelling approaches to time series derived from large medical claims databases held in the private sector.MethodsWe accessed ICD-9-coded office visit data extracted from Truven Health Analytics’ MarketScan Commercial database covering about one third of the US population <65 years during 2001–2009, and Medicare Supplemental data covering about one fifth of US seniors 65+ during 2006–2009. We extracted weekly time series of visits due to respiratory diagnoses, otitis media (OM), and urinary tract infections (UTI), a “negative control”. We used multiple linear regression modelling to estimate age-specific influenza-related excess in office visits.ResultsIn the <65 year age group, in the 8 pre-pandemic seasons studied and for the broadest defined respiratory outcome, the model attributed an average of ~14.5 M (Standard deviation [SD] across seasons 3.9 million) office visits to influenza (rate of 5,581/100,000 population). Of these, ~80 % of visits occurred in the 5–17 and 18–49 age group. In school children aged 5–17 year olds and adult 18–64 year age groups the majority of visits were due to influenza B, while A/H3N2 explained most visits in children <5 year olds. The model further attributed ~2.2 M OM visits (SD across seasons 790,000) annually to influenza, of which 86 % of these occurred in children <18 years; this indicates that 6.4 % of all infants <2 years and 4.9 % of all toddlers aged 2–4 years in the US have an influenza-attributable outpatient visit with an OM diagnosis. In seniors 65 years and older, our model attributed ~0.7 M (SD across seasons 351,000) respiratory visits to influenza (rate of 1,887/100,000 population). The model identified no significant excess UTI (negative control) visits in most seasons.ConclusionsThis is to our knowledge a first study of the outpatient burden of influenza in the US in a large database. The model estimated that 10 % of all children <18 years and 4 % of the entire population <65 years seek outpatient care for respiratory illness attributable to influenza annually.Trial registrationClinicalTrial.gov, NCT02019732.

Artificial intelligence: Machine learning approach for screening large database and drug discovery

Editors’ introduction Renouf’s contribution, like those of Scott and Berber Sardinha, uses Corpus Linguistics (CL) techniques, analysing large amounts of text by computer. The focus in her paper is on the identification of signals of semantic relations. Thus the frame “or, more exactly” might signal a general-particular relationship in “we stopped by the side of a lake, or, more exactly a loch”. Lake is more general (a superordinate) than loch (a hyponym of lake). Renouf’s endeavour was to trace signals like “more exactly” in large text databases, looking at and teasing out the meaning relations which crop up. Her work thus complements Jordan’s analysis of Basis-Assessment in this volume, using quite different methods. At the same time, she attempts to find out how good a match may be made between lexical signal and meaning — it is quite possible in language that one form might carry numerous functions (and conversely that the same meaning relation can be realised by numerous forms). This implies a need to pin down the meanings identified. The context underlying this work is that of the influential work of Cruse (1986), which takes a non-CL view of meaning, heavily dependent on logic, and the notion of contextual normality by which Cruse means the test a near-native speaker or native speaker may make, as to whether a given string like “notable events such as a solar eclipse” seems normal. This is by no means a straightforward decision. Thus in (1) notable events such as meeting the President (2) notable events such as drinking tea the degree of contextual normality would vary in rather unpredictable ways, depending on whether or not one regularly drank tea or worked in the President’s office. The approach Renouf takes is to examine large numbers of texts using a computer, to see instead what forms are actually attested. In so doing, she finds that Cruse’s neat and logical patterns do not seem to be as neatly reflected in the evidence of large numbers of examples from newspaper text. These two methods are not really a simple matter of pre-CL and post-CL, nor are they alternatives, in our view. For a start, the notion of “attested” examples in Linguistics is not new at all, and much early work before computers went into collecting slips of paper with heard or read examples of words. Perhaps the best known example is the many thousands of slips collected by innumerable contributors, for the construction of the Oxford English Dictionary. Second, we would argue that to use CL techniques does not constitute a new Linguistics, any more than using a spade constitutes “Spade Gardening”. It is merely a matter of accessing resources. Third, it is not possible to use CL techniques without recourse to one’s intuitions, e.g. as to what is contextually normal. A CL method is only able to identify positive hits: when one finds numerous examples of a given string of words, one may conclude that it must be contextually normal. If one finds one only, one does not know whether it is a joke or nonce construction or is just rather unusual; and if the string is not found even in a large database, that in itself does not guarantee that it would be contextually abnormal, since new contextually normal strings can be created at any time. Renouf’s paper thus shows that lexical semantics needs insights from logic and intuitions of contextual normality, together with CL methods enabling access to large numbers of examples.

Shows that systems built on a simple statistical technique and a large training database can be automatically optimized to produce classification accuracies of 99% in the domain of handwritten digits. It is also shown that the performance of these systems scale consistently with the size of the training database, where the error rate is cut by more than half for every tenfold increase in the size of the training set from 10 to 100,000 examples. Three distance metrics for the standard nearest neighbor classification system are investigated: a simple Hamming distance metric, a pixel distance metric, and a metric based on the extraction of penstroke features. Systems employing these metrics were trained and tested on a standard, publicly available, database of nearly 225,000 digits provided by the National Institute of Standards and Technology. Additionally, a confidence metric is both introduced by the authors and also discovered and optimized by the system. The new confidence measure proves to be superior to the commonly used nearest neighbor distance. >

Tables constitute the majority of structured data in enterprise environments. While single-table question answering has received significant attention, research on multi-table reasoning remains limited. Compared to single-table QA, multi-table QA requires schema alignment, relational inference, and scalable context man- agement. We introduce tool-augmented reasoning as a paradigm for multi-table QA, and systematically study two complementary strategies: (1) free-form tool interaction, where models iteratively call exploration and computation tools, and(2) structured agent workflows, which stage tool-use into exploration, preparation, and analysis phases. Using the TQA-Bench dataset, we evaluate GPT-4o-mini and GPT-5-mini across both small (8k) and large (128k) database scales under varying tool context constraints. We show that tool augmentation substantially improves robustness and accuracy over direct prompting, with gains up to +28 percentage points. Structured workflows yield further benefits for weaker models on large databases, but regress for stronger models, revealing a scale–capacity trade-off in the value of structure. These results establish tool-augmented reasoning as a powerful paradigm for multi-table QA. Structure aids weaker models under scale but constrains stronger models, underscoring that tool-use strategies must be adapted jointly to model ability and database complexity.

Design Considerations for a Prolog Database Engine