Precision-recall Trade-off Research Articles

MODEL&CO: exoplanet detection in angular differential imaging by learning across multiple observations

ABSTRACT Direct imaging of exoplanets is particularly challenging due to the high contrast between the planet and the star luminosities, and their small angular separation. In addition to tailored instrumental facilities implementing adaptive optics and coronagraphy, post-processing methods combining several images recorded in pupil tracking mode are needed to attenuate the nuisances corrupting the signals of interest. Most of these post-processing methods build a model of the nuisances from the target observations themselves, resulting in strongly limited detection sensitivity at short angular separations due to the lack of angular diversity. To address this issue, we propose to build the nuisance model from an archive of multiple observations by leveraging supervised deep learning techniques. The proposed approach casts the detection problem as a reconstruction task and captures the structure of the nuisance from two complementary representations of the data. Unlike methods inspired by reference differential imaging, the proposed model is highly non-linear and does not resort to explicit image-to-image similarity measurements and subtractions. The proposed approach also encompasses statistical modelling of learnable spatial features. The latter is beneficial to improve both the detection sensitivity and the robustness against heterogeneous data. We apply the proposed algorithm to several data sets from the VLT/SPHERE instrument, and demonstrate a superior precision-recall trade-off compared to the PACO algorithm. Interestingly, the gain is especially important when the diversity induced by ADI is the most limited, thus supporting the ability of the proposed approach to learn information across multiple observations.

Comparative Analysis of Machine Learning Algorithms for Detecting Fake News: Efficacy and Accuracy in the Modern Information Ecosystem

In an era where the spread of fake news poses a significant threat to the integrity of the information landscape, the need for effective detection tools is paramount. This study evaluates the efficacy of three machine learning algorithms—Multinomial Naive Bayes, Passive Aggressive Classifier, and Logistic Regression—in distinguishing fake news from genuine articles. Leveraging a balanced dataset, meticulously processed and vectorized through Term Frequency-Inverse Document Frequency (TF-IDF), we subjected each algorithm to a rigorous classification process. The algorithms were evaluated on metrics such as precision, recall, and F1-score, with the Passive Aggressive Classifier outperforming others, achieving a remarkable 0.99 in both precision and recall. Logistic Regression followed with an accuracy of 0.98, while Multinomial Naive Bayes displayed robust recall at 1.00 but lower precision at 0.91, resulting in an accuracy of 0.95. These metrics underscored the nuanced capabilities of each algorithm in correctly identifying fake and real news, with the Passive Aggressive Classifier demonstrating superior balance in performance. The study's findings highlight the potential of employing machine learning techniques in the fight against fake news, with the Passive Aggressive Classifier showing promise due to its high accuracy and balanced precision-recall trade-off. These insights contribute to the ongoing efforts in digital media to develop advanced, ethical, and accurate tools for maintaining information veracity. Future research should continue to refine these models, ensuring their applicability in diverse and evolving news ecosystems.

CRISPR-DIPOFF: an interpretable deep learning approach for CRISPR Cas-9 off-target prediction.

CRISPR Cas-9 is a groundbreaking genome-editing tool that harnesses bacterial defense systems to alter DNA sequences accurately. This innovative technology holds vast promise in multiple domains like biotechnology, agriculture and medicine. However, such power does not come without its own peril, and one such issue is the potential for unintended modifications (Off-Target), which highlights the need for accurate prediction and mitigation strategies. Though previous studies have demonstrated improvement in Off-Target prediction capability with the application of deep learning, they often struggle with the precision-recall trade-off, limiting their effectiveness and do not provide proper interpretation of the complex decision-making process of their models. To address these limitations, we have thoroughly explored deep learning networks, particularly the recurrent neural network based models, leveraging their established success in handling sequence data. Furthermore, we have employed genetic algorithm for hyperparameter tuning to optimize these models' performance. The results from our experiments demonstrate significant performance improvement compared with the current state-of-the-art in Off-Target prediction, highlighting the efficacy of our approach. Furthermore, leveraging the power of the integrated gradient method, we make an effort to interpret our models resulting in a detailed analysis and understanding of the underlying factors that contribute to Off-Target predictions, in particular the presence of two sub-regions in the seed region of single guide RNA which extends the established biological hypothesis of Off-Target effects. To the best of our knowledge, our model can be considered as the first model combining high efficacy, interpretability and a desirable balance between precision and recall.

Benchmarking of computational methods for m6A profiling with Nanopore direct RNA sequencing.

N6-methyladenosine (m6A) is the most abundant internal eukaryotic mRNA modification, and is involved in the regulation of various biological processes. Direct Nanopore sequencing of native RNA (dRNA-seq) emerged as a leading approach for its identification. Several software were published for m6A detection and there is a strong need for independent studies benchmarking their performance on data from different species, and against various reference datasets. Moreover, a computational workflow is needed to streamline the execution of tools whose installation and execution remains complicated. We developed NanOlympicsMod, a Nextflow pipeline exploiting containerized technology for comparing 14 tools for m6A detection on dRNA-seq data. NanOlympicsMod was tested on dRNA-seq data generated from in vitro (un)modified synthetic oligos. The m6A hits returned by each tool were compared to the m6A position known by design of the oligos. In addition, NanOlympicsMod was used on dRNA-seq datasets from wild-type and m6A-depleted yeast, mouse and human, and each tool's hits were compared to reference m6A sets generated by leading orthogonal methods. The performance of the tools markedly differed across datasets, and methods adopting different approaches showed different preferences in terms of precision and recall. Changing the stringency cut-offs allowed for tuning the precision-recall trade-off towards user preferences. Finally, we determined that precision and recall of tools are markedly influenced by sequencing depth, and that additional sequencing would likely reveal additional m6A sites. Thanks to the possibility of including novel tools, NanOlympicsMod will streamline the benchmarking of m6A detection tools on dRNA-seq data, improving future RNA modification characterization.

Innovative deep learning techniques for monitoring aggressive behavior in social media posts

The study aims to evaluate and compare the performance of various machine learning (ML) classifiers in the context of detecting cyber-trolling behaviors. With the rising prevalence of online harassment, developing effective automated tools for aggression detection in digital communications has become imperative. This research assesses the efficacy of Random Forest, Light Gradient Boosting Machine (LightGBM), Logistic Regression, Support Vector Machine (SVM), and Naive Bayes classifiers in identifying cyber troll posts within a publicly available dataset. Each ML classifier was trained and tested on a dataset curated for the detection of cyber trolls. The performance of the classifiers was gauged using confusion matrices, which provide detailed counts of true positives, true negatives, false positives, and false negatives. These metrics were then utilized to calculate the accuracy, precision, recall, and F1 scores to better understand each model’s predictive capabilities. The Random Forest classifier outperformed other models, exhibiting the highest accuracy and balanced precision-recall trade-off, as indicated by the highest true positive and true negative rates, alongside the lowest false positive and false negative rates. LightGBM, while effective, showed a tendency towards higher false predictions. Logistic Regression, SVM, and Naive Bayes displayed identical confusion matrix results, an anomaly suggesting potential data handling or model application issues that warrant further investigation. The findings underscore the effectiveness of ensemble methods, with Random Forest leading in the cyber troll detection task. The study highlights the importance of selecting appropriate ML algorithms for text classification tasks in social media contexts and emphasizes the need for further scrutiny into the anomaly observed among the Logistic Regression, SVM, and Naive Bayes results. Future work will focus on exploring the reasons behind this occurrence and the potential of deep learning techniques in enhancing detection performance.

PhosBoost: Improved phosphorylation prediction recall using gradient boosting and protein language models.

Protein phosphorylation is a dynamic and reversible post-translational modification that regulates a variety of essential biological processes. The regulatory role of phosphorylation in cellular signaling pathways, protein-protein interactions, and enzymatic activities has motivated extensive research efforts to understand its functional implications. Experimental protein phosphorylation data in plants remains limited to a few species, necessitating a scalable and accurate prediction method. Here, we present PhosBoost, a machine-learning approach that leverages protein language models and gradient-boosting trees to predict protein phosphorylation from experimentally derived data. Trained on data obtained from a comprehensive plant phosphorylation database, qPTMplants, we compared the performance of PhosBoost to existing protein phosphorylation prediction methods, PhosphoLingo and DeepPhos. For serine and threonine prediction, PhosBoost achieved higher recall than PhosphoLingo and DeepPhos (.78, .56, and .14, respectively) while maintaining a competitive area under the precision-recall curve (.54, .56, and .42, respectively). PhosphoLingo and DeepPhos failed to predict any tyrosine phosphorylation sites, while PhosBoost achieved a recall score of .6. Despite the precision-recall tradeoff, PhosBoost offers improved performance when recall is prioritized while consistently providing more confident probability scores. A sequence-based pairwise alignment step improved prediction results for all classifiers by effectively increasing the number of inferred positive phosphosites. We provide evidence to show that PhosBoost models are transferable across species and scalable for genome-wide protein phosphorylation predictions. PhosBoost is freely and publicly available on GitHub.

Evaluation of Attention-Based LSTM and Bi-LSTM Networks For Abstract Text Classification in Systematic Literature Review Automation

Systematic Review (SR) presents the highest form of evidence in research for decision and policy-making. Nonetheless, the structured steps involved in carrying out SRs make it demanding for reviewers. Many studies have projected the abstract screening stage in the SR process to be the most burdensome for reviewers, thus automating this stage with artificial intelligence (AI). However, majority of these studies focus on using traditional machine learning classifiers for the abstract classification. Thus, there remain a gap to explore the potential of deep learning techniques for this task. This study seeks to bridge the gap by exploring how LSTM and Bi-LSTM models together with GloVe for vectorisation can accelerate this stage. As a further aim to increase precision while sustaining a recall >= 95% due to precision-recall trade-off, attention mechanics is added to these classifiers. The final experimental results obtained showed that Bi-LSTM with attention has the capacity to expedite citation screening.

RB-CCR: Radial-Based Combined Cleaning and Resampling algorithm for imbalanced data classification

Real-world classification domains, such as medicine, health and safety, and finance, often exhibit imbalanced class priors and have asynchronous misclassification costs. In such cases, the classification model must achieve a high recall without significantly impacting precision. Resampling the training data is the standard approach to improving classification performance on imbalanced binary data. However, the state-of-the-art methods ignore the local joint distribution of the data or correct it as a post-processing step. This can causes sub-optimal shifts in the training distribution, particularly when the target data distribution is complex. In this paper, we propose Radial-Based Combined Cleaning and Resampling (RB-CCR). RB-CCR utilizes the concept of class potential to refine the energy-based resampling approach of CCR. In particular, RB-CCR exploits the class potential to accurately locate sub-regions of the data-space for synthetic oversampling. The category sub-region for oversampling can be specified as an input parameter to meet domain-specific needs or be automatically selected via cross-validation. Our 5times 2 cross-validated results on 57 benchmark binary datasets with 9 classifiers show that RB-CCR achieves a better precision-recall trade-off than CCR and generally out-performs the state-of-the-art resampling methods in terms of AUC and G-mean.

Differentiation of Active Corneal Infections from Healed Scars Using Deep Learning

To develop and evaluate an automated, portable algorithm to differentiate active corneal ulcers from healed scars using only external photographs. A convolutional neural network was trained and tested using photographs of corneal ulcers and scars. De-identified photographs of corneal ulcers were obtained from the Steroids for Corneal Ulcers Trial (SCUT), Mycotic Ulcer Treatment Trial (MUTT), and Byers Eye Institute at Stanford University. Photographs of corneal ulcers (n= 1313) and scars (n= 1132) from the SCUT and MUTT were used to train a convolutional neural network (CNN). The CNN was tested on 2 different patient populations from eye clinics in India (n= 200) and the Byers Eye Institute at Stanford University (n= 101). Accuracy was evaluated against gold standard clinical classifications. Feature importances for the trained model were visualized using gradient-weighted class activation mapping. Accuracy of the CNN was assessed via F1 score. The area under the receiver operating characteristic (ROC) curve (AUC) was used to measure the precision-recall trade-off. The CNN correctly classified 115 of 123 active ulcers and 65 of 77 scars in patients with corneal ulcer from India (F1 score, 92.0% [95% confidence interval (CI), 88.2%-95.8%]; sensitivity, 93.5% [95% CI, 89.1%-97.9%]; specificity, 84.42% [95% CI, 79.42%-89.42%]; ROC: AUC, 0.9731). The CNN correctly classified 43 of 55 active ulcers and 42 of 46 scars in patients with corneal ulcers from Northern California (F1 score, 84.3% [95% CI, 77.2%-91.4%]; sensitivity, 78.2% [95% CI, 67.3%-89.1%]; specificity, 91.3% [95% CI, 85.8%-96.8%]; ROC: AUC, 0.9474). The CNN visualizations correlated with clinically relevant features such as corneal infiltrate, hypopyon, and conjunctival injection. The CNN classified corneal ulcers and scars with high accuracy and generalized to patient populations outside of its training data. The CNN focused on clinically relevant features when it made a diagnosis. The CNN demonstrated potential as an inexpensive diagnostic approach that may aid triage in communities with limited access to eye care.

SSG-LUGIA: Single Sequence based Genome Level Unsupervised Genomic Island Prediction Algorithm.

Genomic Islands (GIs) are clusters of genes that are mobilized through horizontal gene transfer. GIs play a pivotal role in bacterial evolution as a mechanism of diversification and adaptation to different niches. Therefore, identification and characterization of GIs in bacterial genomes is important for understanding bacterial evolution. However, quantifying GIs is inherently difficult, and the existing methods suffer from low prediction accuracy and precision-recall trade-off. Moreover, several of them are supervised in nature, and thus, their applications to newly sequenced genomes are riddled with their dependency on the functional annotation of existing genomes. We present SSG-LUGIA, a completely automated and unsupervised approach for identifying GIs and horizontally transferred genes. SSG-LUGIA is a novel method based on unsupervised anomaly detection technique, accompanied by further refinement using cues from signal processing literature. SSG-LUGIA leverages the atypical compositional biases of the alien genes to localize GIs in prokaryotic genomes. SSG-LUGIA was assessed on a large benchmark dataset `IslandPick' and on a set of 15 well-studied genomes in the literature and followed by a thorough analysis on the well-understood Salmonella typhi CT18 genome. Furthermore, the efficacy of SSG-LUGIA in identifying horizontally transferred genes was evaluated on two additional bacterial genomes, namely, those of Corynebacterium diphtheria NCTC13129 and Pseudomonas aeruginosa LESB58. SSG-LUGIA was examined on draft genomes and was demonstrated to be efficient as an ensemble method. Our results indicate that SSG-LUGIA achieved superior performance in comparison to frequently used existing methods. Importantly, it yielded a better trade-off between precision and recall than the existing methods. Its nondependency on the functional annotation of genomes makes it suitable for analyzing newly sequenced, yet uncharacterized genomes. Thus, our study is a significant advance in identification of GIs and horizontally transferred genes. SSG-LUGIA is available as an open source software at https://nibtehaz.github.io/SSG-LUGIA/.

Detecting Submerged Objects Using Active Acoustics and Deep Neural Networks: A Test Case for Pelagic Fish

The accurate detection and quantification of submerged targets has been recognized as a key challenge in marine exploration, one that traditional census approaches cannot handle efficiently. Here we present a deep learning approach to detect the pattern of a moving fish from the reflections of an active acoustic emitter. To allow for real-time detection, we use a convolutional neural network, which provides the simultaneous labeling of a large buffer of signal samples. This allows to capture the structure of the reflecting signal from the moving target and to separate it from clutter reflections. We evaluate system performance both on synthetic (simulated) data, as well as on real data recorded over 50 sea experiments in a variety of sea conditions. When tested on real signals, the network trained on simulated patterns showed non-trivial detection capabilities, suggesting that transfer learning can be a viable approach in these scenarios, where tagged data is often lacking. However, training the network directly on the real reflections with data augmentation techniques allowed to reach a more favorable precision-recall trade-off, approaching an ideal detection bound. We also evaluate an alternative model based on recurrent neural networks which, despite exhibiting slightly inferior performance, could be applied in scenarios requiring on-line processing of the reflection sequence.

Identifying current Juul users among emerging adults through Twitter feeds

IntroductionJuul is the most popular electronic cigarette on the market. Amid concerns around uptake of e-cigarettes by never smokers, can we detect whether someone uses Juul based on their social media activities? This is the central premise of the effort reported in this paper. Several recent social media-related studies on Juul use tend to focus on the characterization of Juul-related messages on social media. In this study, we assess the potential in using machine learning methods to automatically identify Juul users (past 30-day usage) based on their Twitter data. MethodsWe obtained a collection of 588 instances, for training and testing, of Juul use patterns (along with associated Twitter handles) via survey responses of college students. With this data, we built and tested supervised machine learning models based on linear and deep learning algorithms with textual, social network (friends and followers), and other hand-crafted features. ResultsThe linear model with textual and follower network features performed best with a precision-recall trade-off such that precision (PPV) is 57 % at 24 % recall (sensitivity). Hence, at least every other college-attending Twitter user flagged by our model is expected to be a Juul user. Additionally, our results indicate that social network features tend to have a large impact (positive) on classification performance. ConclusionThere are enough latent signals from social feeds for supervised modeling of Juul use, even with limited training data, implying that such models are highly beneficial to very focused intervention campaigns. This initial success indicates potential for more involved automated surveillance of Juul use based on social media data, including Juul usage patterns, nicotine dependence, and risk awareness.

Industrial Anomaly Detection: A Comparison of Unsupervised Neural Network Architectures

Anomaly detection is critical to the efficient and secure operation of industrial equipment, integrated sensors, and the overall production process. Cyber-physical systems and supervisory control and data acquisition systems often embed algorithms for anomaly detection. Anomalies are, by definition, rare events and labor-intensive to identify. Pipelined deep learning techniques such as recurrent neural networks (RNN), 1-D convolution neural networks (1DCNN), and generative adversarial networks create unique anomaly detection architectures. These architectures are unsupervised, and labeled anomaly data is not required. With two large, multivariate time-series testbed experimental datasets, comparisons are made between the neural network architectures and traditional machine learning (TML) baseline techniques. Evaluation metrics appropriate for imbalanced anomaly datasets include the area under the receiver operating characteristic curve and the area under the precision-recall tradeoff curve. Experimental results demonstrate that pipelined neural network architectures exhibit higher detection rates than the TML techniques. 1DCNN can substitute for RNNs when performance and long time-sequences are essential.

Efficient pothole detection using smartphone sensors

Road safety remains a casualty in India, with potholes wrecking asphalt pavements by the dozens. A study in 2017 recorded that potholes caused the budget for road safety to increase by a whopping 100.4 per cent, and even doubled the death toll from that of the year prior. To address this situation, an effective solution is required that ensures the drivers’ safety and can prove beneficial for long term measures. This can be established by employing an apt pothole detection system which is simple yet functional. In this paper, the method for such a system is described which uses accelerometer and gyroscope, both built in the modern day smartphones, to sense potholes. Pothole induced vibrations can be measured on the axis reading, making them distinguishable. Our proposed Neural Network model is trained and evaluated on the data acquired from the sensors and classifies the potholes from the non-potholes. The neural network gives a classification accuracy of 94.78 per cent. It also presents a solid precision-recall trade-off with 0.71 precision and 0.81 recall, considerably high for a problem with class imbalance. The results indicate that the method is suitable for creating an accurate and sensitive supervised model for pothole detection.

Large-scale directed network inference with multivariate transfer entropy and hierarchical statistical testing.

Network inference algorithms are valuable tools for the study of large-scale neuroimaging datasets. Multivariate transfer entropy is well suited for this task, being a model-free measure that captures nonlinear and lagged dependencies between time series to infer a minimal directed network model. Greedy algorithms have been proposed to efficiently deal with high-dimensional datasets while avoiding redundant inferences and capturing synergistic effects. However, multiple statistical comparisons may inflate the false positive rate and are computationally demanding, which limited the size of previous validation studies. The algorithm we present—as implemented in the IDTxl open-source software—addresses these challenges by employing hierarchical statistical tests to control the family-wise error rate and to allow for efficient parallelization. The method was validated on synthetic datasets involving random networks of increasing size (up to 100 nodes), for both linear and nonlinear dynamics. The performance increased with the length of the time series, reaching consistently high precision, recall, and specificity (>98% on average) for 10,000 time samples. Varying the statistical significance threshold showed a more favorable precision-recall trade-off for longer time series. Both the network size and the sample size are one order of magnitude larger than previously demonstrated, showing feasibility for typical EEG and magnetoencephalography experiments.

SonicParanoid: fast, accurate and easy orthology inference.

MotivationOrthology inference constitutes a common base of many genome-based studies, as a pre-requisite for annotating new genomes, finding target genes for biotechnological applications and revealing the evolutionary history of life. Although its importance keeps rising with the ever-growing number of sequenced genomes, existing tools are computationally demanding and difficult to employ.ResultsHere, we present SonicParanoid, which is faster than, but comparably accurate to, the well-established tools with a balanced precision-recall trade-off. Furthermore, SonicParanoid substantially relieves the difficulties of orthology inference for those who need to construct and maintain their own genomic datasets.Availability and implementationSonicParanoid is available with a GNU GPLv3 license on the Python Package Index and BitBucket. Documentation is available at http://iwasakilab.bs.s.u-tokyo.ac.jp/sonicparanoid.Supplementary information Supplementary data are available at Bioinformatics online.

Dissection of Influenza Infection InVivo by Single-Cell RNA Sequencing.

SummaryThe influenza virus is a major cause of morbidity and mortality worldwide. Yet, both the impact of intracellular viral replication and the variation in host response across different cell types remain uncharacterized. Here we used single-cell RNA sequencing to investigate the heterogeneity in the response of lung tissue cells to in vivo influenza infection. Analysis of viral and host transcriptomes in the same single cell enabled us to resolve the cellular heterogeneity of bystander (exposed but uninfected) as compared with infected cells. We reveal that all major immune and non-immune cell types manifest substantial fractions of infected cells, albeit at low viral transcriptome loads relative to epithelial cells. We show that all cell types respond primarily with a robust generic transcriptional response, and we demonstrate novel markers specific for influenza-infected as opposed to bystander cells. These findings open new avenues for targeted therapy aimed exclusively at infected cells.

Adaptive Image Mismatch Removal With Vector Field Interpolation Based on Improved Regularization and Gaussian Kernel Function

When the regularized kernel methods are utilized in the mismatch removal problem, the regularization coefficient and the choice of kernel function will seriously affect the performance of the methods. In this paper, we propose a method that combines an improved regularization and an adaptive Gaussian kernel function to interpolate the vector fields so as to overcome the issue. We formulated the problem as a modified maximum a posterior estimation of a Bayesian model. In this model, a two-order term of the regularization coefficient is introduced into the regularized risk function in order that the coefficient can be adaptively estimated in the expectation–maximization algorithm. In addition, an adaptive Gaussian kernel function also is imposed to construct the regularization, in which the width of the kernel function is adaptively determined by the diagonal length of the maximum enveloping rectangle of the sample set. Our experimental results verified that our method was robust to large outlier percentages and was slightly superior to some state-of-the-art methods in precision-recall tradeoff and efficiency. The evidence that the performance of our method was insensitive to the remaining inner parameters verified its good self-adaptability. Finally, airborne image pairs were used to demonstrate that our method can establish the feature correspondences even under a discontinuous vector field scene. In addition, we found that our method can obtain higher precision given a residual threshold for special applications such as robust epipolar geometry estimation in computer vision and photogrammetry.

Automated Identification of Diabetic Retinopathy Using Deep Learning

Diabetic retinopathy (DR) is one of the leading causes of preventable blindness globally. Performing retinal screening examinations on all diabetic patients is an unmet need, and there are many undiagnosed and untreated cases of DR. The objective of this study was to develop robust diagnostic technology to automate DR screening. Referral of eyes with DR to an ophthalmologist for further evaluation and treatment would aid in reducing the rate of vision loss, enabling timely and accurate diagnoses. We developed and evaluated a data-driven deep learning algorithm as a novel diagnostic tool for automated DR detection. The algorithm processed color fundus images and classified them as healthy (noretinopathy) or having DR, identifying relevant cases for medical referral. A total of 75 137 publicly available fundus images from diabetic patients were used to train and test an artificial intelligence model to differentiate healthy fundi from those with DR. A panel of retinal specialists determined the ground truth for our data set before experimentation. We also tested our model using the public MESSIDOR 2 and E-Ophtha databases for external validation. Information learned in our automated method was visualized readily through an automatically generated abnormality heatmap, highlighting subregions within each input fundus image for further clinical review. We used area under the receiver operating characteristic curve (AUC) as a metric to measure the precision-recall trade-off of our algorithm, reporting associated sensitivity and specificity metrics on the receiver operating characteristic curve. Our model achieved a 0.97 AUC with a 94% and 98% sensitivity and specificity, respectively, on 5-fold cross-validation using our local data set. Testing against the independent MESSIDOR 2 and E-Ophtha databases achieved a 0.94 and 0.95 AUC score, respectively. A fully data-driven artificial intelligence-based grading algorithm can be used to screen fundus photographs obtained from diabetic patients and to identify, with high reliability, which cases should be referred to an ophthalmologist for further evaluation and treatment. The implementation of such an algorithm on a global basis could reduce drastically the rate of vision loss attributed to DR.

Incremental Loop Closure Verification by Guided Sampling

Loop closure detection, which is the task of identifying locations revisited by a robot in a sequence of odometry and perceptual observations, is typically formulated as a combination of two subtasks: (1) bag-of-words image retrieval and (2) post-verification using random sample consensus (RANSAC) geometric verification. The main contribution of this study is the proposal of a novel post-verification framework that achieves good precision recall trade-off in loop closure detection. This study is motivated by the fact that not all loop closure hypotheses are equally plausible (e.g., owing to mutual consistency between loop closure constraints) and that if we have evidence that one hypothesis is more plausible than the others, then it should be verified more frequently. We demonstrate that the loop closure detection problem can be viewed as an instance of a multi-model hypothesize-and-verify framework. Thus, we can build guided sampling strategies on this framework where loop closures proposed using image retrieval are verified in a planned order (rather than in a conventional uniform order) to operate in a constant time. Experimental results using a stereo simultaneous localization and mapping (SLAM) system confirm that the proposed strategy, the use of loop closure constraints and robot trajectory hypotheses as a guide, achieves promising results despite the fact that there exists a significant number of false positive constraints and hypotheses.