Exemplar Samples Research Articles

Deep continual hashing for real-world multi-label image retrieval

Deep Hashing has achieved great success in large-scale image retrieval due to binary code’s storage and computation efficiency. However, its learning paradigm under real-world environments is less studied, and most existing approaches are developed in the closed scenario, e.g., simple and unchanging semantics. When images of new classes emerge, they have to retrain the model on all history training datasets, but the constant data uploading makes this impractical. This paper proposes a novel method, called continual deep semantic hashing (CDSH), for learning binary codes of multi-label images with increasing classes. The CDSH consists of two hashing networks. One learns to hash the increasing semantics of data, i.e., label, into the semantic codes and accumulates label-code pairs as long-term knowledge, incorporating empirically verified loss and designed special regularization to ensure encoding old labels unchanged. The other learns to map images to the corresponding semantic code from a probabilistic view and solid knowledge via retaining history exemplar samples and projecting model gradients. We theoretically prove this improves the probability of old data’s code unchanged after the model is updated. Extensive experiments on four widely used image datasets demonstrate that the CDSH method can continually learn hash functions and yield state-of-the-art retrieval performance.

Infrahuman madness: Mental health nursing and the discursive production of alterity.

By examining an exemplar sample of mental health nursing educational policies and related legislation, in this article, we trace the discursive production of madness as an "othered"identity category. We engage in a critical discourse analysis of mental health nursing education in Canada, drawing on provincial and federal policies and legislation as the main sources of data. Theoretically framed by critical posthumanism and mad studies, this article outlines how the mad subjectivity becomes decontextualized out of its identity-based understanding and recontextualized as an inferior category of "the human,"circulating within discourses of pedagogy, economics, law, and psychiatry. The article maps the intertextual nexus of the discourse of mental health nursing education, making visible the complex, the arbitrary, and the sometimes-contradictory nature of the discipline's grappling with identity-based mental health concepts. We close with several implications for nursing policy, education, and practice.

Active Sampling for Text Classification with Subinstance Level Queries

Active learning algorithms are effective in identifying the salient and exemplar samples from large amounts of unlabeled data. This tremendously reduces the human annotation effort in inducing a machine learning model as only a few samples, which are identified by the algorithm, need to be labeled manually. In problem domains like text mining and video classification, human oracles peruse the data instances incrementally to derive an opinion about their class labels (such as reading a movie review progressively to assess its sentiment). In such applications, it is not necessary for the human oracles to review an unlabeled sample end-to-end in order to provide a label; it may be more efficient to identify an optimal subinstance size (percentage of the sample from the start) for each unlabeled sample, and request the human annotator to label the sample by analyzing only the subinstance, instead of the whole data sample. In this paper, we propose a novel framework to address this challenging problem, in an effort to further reduce the labeling burden on the human oracles and utilize the available labeling budget more efficiently. We pose the sample and subinstance size selection as a constrained optimization problem and derive a linear programming relaxation to select a batch of exemplar samples, together with the optimal subinstance size of each, which can potentially augment maximal information to the underlying classification model. Our extensive empirical studies on six challenging datasets from the text mining domain corroborate the practical usefulness of our framework over competing baselines.

Meta-learning an intermediate representation for few-shot prediction of landslide susceptibility in large areas

Predicting a landslide susceptibility map (LSM) is essential for risk recognition and disaster prevention. Despite the successful application of data-driven approaches for LSM prediction, most methods generally apply a single global model to predict the LSM for an entire target region. However, in large-scale areas with significant environmental change, various parts of the region hold different landslide-inducing environments, and therefore, should be predicted with respective models. This study first segmented target scenarios into blocks for individual analysis. Then, the critical problem is that in each block with limited samples, conducting training and testing a model is impossible for a satisfactory LSM prediction, especially in dangerous mountainous areas where landslide surveying is costly. To solve the problem, we trained an intermediate representation by the meta-learning paradigm, which is superior for capturing information valuable for few-shot adaption from LSM tasks. We hypothesized that there are more general and vital concepts concerning landslide causes and are sensitive to variations in input features. Thus, we can quickly adapt the models from the intermediate representation for different blocks or even unseen tasks using very few exemplar samples. Experimental results on the two study areas demonstrated the validity of our block-wise analysis in large scenarios and revealed the top few-shot adaption performances of the proposed methods.

Memory-Efficient Class-Incremental Learning for Image Classification

With the memory-resource-limited constraints, class-incremental learning (CIL) usually suffers from the "catastrophic forgetting" problem when updating the joint classification model on the arrival of newly added classes. To cope with the forgetting problem, many CIL methods transfer the knowledge of old classes by preserving some exemplar samples into the size-constrained memory buffer. To utilize the memory buffer more efficiently, we propose to keep more auxiliary low-fidelity exemplar samples, rather than the original real-high-fidelity exemplar samples. Such a memory-efficient exemplar preserving scheme makes the old-class knowledge transfer more effective. However, the low-fidelity exemplar samples are often distributed in a different domain away from that of the original exemplar samples, that is, a domain shift. To alleviate this problem, we propose a duplet learning scheme that seeks to construct domain-compatible feature extractors and classifiers, which greatly narrows down the above domain gap. As a result, these low-fidelity auxiliary exemplar samples have the ability to moderately replace the original exemplar samples with a lower memory cost. In addition, we present a robust classifier adaptation scheme, which further refines the biased classifier (learned with the samples containing distillation label knowledge about old classes) with the help of the samples of pure true class labels. Experimental results demonstrate the effectiveness of this work against the state-of-the-art approaches. We will release the code, baselines, and training statistics for all models to facilitate future research.

Asking the Right Questions to the Right Users: Active Learning with Imperfect Oracles

Active learning algorithms automatically identify the salient and exemplar samples from large amounts of unlabeled data and tremendously reduce human annotation effort in inducing a machine learning model. In a traditional active learning setup, the labeling oracles are assumed to be infallible, that is, they always provide correct answers (in terms of class labels) to the queried unlabeled instances. However, in real-world applications, oracles are often imperfect and provide incorrect label annotations. Oracles also have diverse expertise and while they may be noisy, certain oracles may provide accurate annotations to certain specific instances. In this paper, we propose a novel framework to address the challenging problem of active learning in the presence of multiple imperfect oracles. We pose the optimal sample and oracle selection as a constrained optimization problem and derive a linear programming relaxation to select a batch of (sample-oracle) pairs, which can potentially augment maximal information to the underlying classification model. Our extensive empirical studies on 9 challenging datasets (from a variety of application domains) corroborate the usefulness of our framework over competing baselines.

Exploring Relational Reconciliation Processes in Christian, Jewish, and Muslim Families

ObjectiveTo explore personal and interpersonal processes that encourage relational reconciliation in nonclinical religious families.BackgroundFew studies have addressed what reconciliation is and when it is beneficial. Although intergroup reconciliation has been well documented, little research has addressed relational reconciliation in families. We focus on relational reconciliation pertaining to more normative and typical relational hurts and offenses that nearly all families experience, rather than severe offenses that might be best addressed in a clinical setting.MethodUsing systematic qualitative methods, in‐depth interviews from a nonclinical, exemplar sample of 198 religiously, ethnically, and geographically diverse mothers, fathers, and adolescents were coded and analyzed. Research questions focused on what circumstances led to a need for reconciliation, what motivated families to reconcile, how families reconciled, and what benefits families received from reconciling.ResultsFamilies were reportedly motivated to reconcile (a) because of their religious beliefs, (b) because they could see a “bigger picture” beyond the immediate conflict, and (c) because they felt that they had been recipients of God's love and forgiveness. The process of reconciliation involved (a) praying to God for help (spiritual processes), (b) admitting mistakes and taking responsibility (personal processes), (c) forgiving and being forgiven (relational processes), and (d) working to fix problems and make amends (practical processes).ConclusionConsistent with previous research, these processes reportedly fostered self‐healing following relational distance and led to positive relational outcomes.ImplicationsWhen self‐healing does not occur, clinicians, pastoral counselors, and family life educators may play an important role in helping individuals and families develop and incorporate the transformative processes of reconciliation identified in our data.

Medication Exposure Patterns in Primary Care Patients Prescribed Pharmacogenetically Actionable Opioids

Current approaches to assessing medication exposure fail to capture the complexity of the phenomenon and the context in which it occurs. This study's purpose was to develop a typology of subgroups of patients who share common patterns of medication exposure. To create the typology, we used an exemplar sample of 30 patients in a large public healthcare system who had been prescribed the pharmacogenetically actionable opioids codeine or tramadol. Data related to medication exposure were drawn from large data repositories. Using a person-oriented qualitative approach, eight subgroups of patients who shared common patterns of medication exposure were identified. The subgroups had one of five opioid prescription patterns (i.e., singular, episodic, switching, sustained, multiplex), and one of three types of primary foci of medical care (i.e., pain, comorbidities, both). The findings reveal medication exposure patterns that are dynamic, multidimensional, and complex, and the typology offers an innovative approach to assessing medication exposure.

Encoding Microreactors with Droplet Chains in Microfluidics.

Droplet-based high throughput biomolecular screening and combinatorial synthesis entail a viable indexing strategy to be developed for the identification of each microreactor. Here, we propose a novel indexing scheme based on the generation of droplet sequences on demand to form unique encoding droplet chains in fluidic networks. These codes are represented by multiunit and multilevel droplets packages, with each code unit possessing several distinct signal levels, potentially allowing large encoding capacity. For proof of concept, we use magnetic nanoparticles as the encoding material and a giant magnetoresistance (GMR) sensor-based active sorting system supplemented with an optical detector to generate and decode the sequence of one exemplar sample droplet reactor and a 4-unit quaternary magnetic code. The indexing capacity offered by 4-unit multilevel codes with this indexing strategy is estimated to exceed 104, which holds great promise for large-scale droplet-based screening and synthesis.

SR-NBS: A fast sparse representation based N-best class selector for robust phoneme classification

Although exemplar based approaches have shown good accuracy in classification problems, some limitations are observed in the accuracy of exemplar based automatic speech recognition (ASR) applications. The main limitation of these algorithms is their high computational complexity which makes them difficult to extend to ASR applications. In this paper, an N-best class selector is introduced based on sparse representation (SR) and a tree search strategy. In this approach, the classification is fulfilled in three steps. At first, the set of similar training samples for the specific test sample is selected by k-dimensional (KD) tree search algorithm. Then, an SR based N-best class selector is used to limit the classification among certain classes. This makes the classifier adapt to each test sample and reduces the empirical risk. Finally, a well known low error rate classifier is trained by the selected exemplar samples and the trained classifier is employed to classify among the candidate classes. The algorithm is applied to phoneme classification and it is compared with some well-known phoneme classifiers according to accuracy and complexity issues. By this approach, we obtain competitive classification rate with promising computational complexity in comparison with the state of the art phoneme classifiers in clean and well known acoustic noisy environments which causes this approach become a suitable candidate for ASR applications.

Phylogeny and classification of Hypherpes auctorum (Coleoptera: Carabidae: Pterostichini: Pterostichus)

Based on an exemplar sample of pterostichine species (Carabidae: Pterostichini), 28S rDNA and COI and COII mtDNA sequence data are used to reconstruct a phylogenetic hypothesis for generic and subgeneric taxa putatively in or related to the subgenus Hypherpes Chaudoir (Coleoptera: Carabidae: Pterostichus Bonelli). The monophyly of Pterostichus is equivocal as the position of the subgenus Bothriopterus Chaudoir varies depending on methods of sequence alignment and gap region treatment. Pterostichus is found to be monophyletic in the combined data analysis if Cyclotrachelus Chaudoir and Tapinopterus Schaum are included in a larger concept of the genus. It is recommended that these be treated as subgenera of Pterostichus. Taxa currently included in Hypherpes are found to form a monophyletic group. No taxon previously suggested as a close relative of Hypherpes was found to be in, or closely related to Hypherpes. The sister-group of Hypherpes remains unclear, but there is some support for a clade of Pseudoferonina Ball Cryobius Chaudoir as the adelphotaxon. Taxa included in current classifications of Hypherpes compose a group that is in fact a complex of Hypherpes sensu stricto and two other subgenera, Leptoferonia Casey and Anilloferonia Van Dyke, which have been treated as junior synonyms of Hypherpes. Our analyses show that these three taxa are well supported as subgenera and reciprocally monophyletic, with the only change to previous taxonomic concepts of included species being the transfer of Pterostichus rothi (Hatch) from Anilloferonia to Leptoferonia. It is recommended that all three of these subgenera be recognized rather than being subsumed under Hypherpes. In Leptoferonia the DNA data support all species groups that were established by Hacker using morphological characters, with the exception of the inopinus-group. Significant reduction of the compound eyes has occurred independently at least five and possibly seven times in the Hypherpes complex. As many as five separate instances of eye reduction may have occurred in Leptoferonia alone. Maddison's concentrated changes test was used to show that there is a significant correlation between microphthalmy and autapomorphic sequence data as represented by longer than average terminal branch lengths based on Bayesian estimates of change per site. However, taxon pair contrasts show no consistent pattern of absolute difference of evolutionary rate or directionality of differences between small-eyed taxa and their sister species or sister clade. Repeated patterns of allopatric distributions are found for species-pairs of Leptoferonia, which consist of divisions along a north/south axis near the Pacific Coast and in the Sierra Nevada Range, or east/west divisions between coastal species and inland or Sierran species. In addition to allopatric biogeographic patterns, instances of sympatry in closely related species are interpreted to have been the result of two reduced-eye species moving into the deep litter and soil layer, thereby ecologically differentiating from near-surface leaf-litter and log dwelling species. Pterostichus morionides (Chaudoir), which is restricted to the Sierra Nevada Mountains in western North America, is found to be sister to P. adoxus (Say) and P. tristis (Dejean), the only species of Hypherpes in eastern North America. This grouping (mta-clade) was further tested by using a subset of taxa for 18S rDNA, CAD and wg sequence data and was found in some or all most-parsimonious trees for these data. In cases where they did not form a clade, they usually formed a convex group. Although counterintuitive due to the unusual disjunct biogeographic connection of these two areas and the generally dissimilar form of the adults, the mta-clade is very well supported by the DNA sequence data.

Exemplar Sampling

An exemplar sample is a small subset of a dataset that forms a characterization of the dataset and can also be used as a list of targets for further inspection. Traditionally exemplars are encountered in cluster analysis techniques such as k-medoids, where each exemplar represents a different cluster. This article extends the concept of exemplar sampling by constructing four criteria for choosing a sample: similarity, distinctness, exhaustiveness, and typicalness. Each criterion illustrates a different way of characterizing the dataset. We describe an adapted version of the CLARANS algorithm which can be used to efficiently draw exemplar samples, and show some examples on test datasets. Exemplar sampling is proposed for use in finite population inference, as an alternative to methods such as stratification or balancing.

Using Hydrophilic Adhesive Tape for Collection of Evidence for Forensic DNA Analysis

Known exemplar samples of human DNA have traditionally been body fluids, such as blood, saliva, and semen. In each case, the presence of water is a risk for the bacterial growth, which may degrade the DNA evidence. In this study, the authors have developed a method that employed a hydrophilic adhesive tape (HAT) for collecting DNA evidence. The HAT method was used to remove surface cells from relatively hairless areas on the body. The area examined were ankle, arm, behind the ear, between fingers and back of the neck. The HAT was then dissolved in the extraction buffer. DNA typing was performed at vWA, THo1, F13A1, and FES loci using the short tandem repeat (STR) analysis. Our results show that the samples collected from ear give the best results with a success rate of 100%. All subjects tested by this method had known STR genotypes established from buccal swabs. The authors' results suggest that the HAT method can be used as a less invasive method for collecting biological evidence for forensic DNA analysis. In addition, this collection method should reduce the risk of DNA degradation due to the moisture, which is encountered using conventional collecting methods.

Genetic Variation among South American Populations of Tetraploid Chenopodium sect. Chenopodium subsect. Cellulata

Electrophoretic analysis of South American Chenopodium (sect. Che- nopodium subsect. Cellulata, 2n = 4x = 36) reveals multi-locus isozyme systems for leucine aminopeptidase (LAP) and phosphoglucoisomerase (PGI) that appear to be the result of tetraploid gene duplication. The bi-locus LAP system is monomorphic for active alleles with putative null alleles at each locus that are not present in pop- ulations at the northern range extreme. The null allele at Lap-i appears to be sporadically distributed in the highlands of Peru, Bolivia, and Argentina, whereas the null variant at Lap-2 shows a clinal increase from north to south with fixation in the Argentine lowlands. All material (62 population samples from Chile, Argen- tina, Bolivia, and Peru, and 206 exemplar samples from Ecuador, Peru, and Bolivia) is monomorphic at Pg'-I and essentially monomorphic at the other two PGI loci, with the exception of low elevation populations of Argentina, which show five active variants and one null. Aside from clinal trends at two loci, the observed pattern of genetic variation is not consistent with extreme discontinuities in selective regime, habitat, and distribution. It is proposed that an archaic genotype has been conserved in the Andean area of chenopod cultivation through interpopulation gene flow me- diated by weed-crop genetic interaction and human dispersal. A relatively high level of inter-population reproductive isolation in lowland Argentina has possibly allowed differentiation of PGI alleles and silencing of one LAP locus through fixation of null alleles. Populations from lowland Argentina include individuals showing electro- phoretic diploidization in both isozyme systems, thus producing the typical pheno- type of diploid Chenopodium. The percentage of heterozygous plants per population and level of detected out-crossing indicate a high potential for gene flow within populations of lowland Argentina, although genetic interaction with sympatric C. album was not detected.