Confidence Threshold Value Research Articles

Training a Hyperdimensional Computing Classifier Using a Threshold on Its Confidence.

Hyperdimensional computing (HDC) has become popular for light-weight and energy-efficient machine learning, suitable for wearable Internet-of-Things devices and near-sensor or on-device processing. HDC is computationally less complex than traditional deep learning algorithms and achieves moderate to good classification performance. This letter proposes to extend the training procedure in HDC by taking into account not only wrongly classified samples but also samples that are correctly classified by the HDC model but with low confidence. We introduce a confidence threshold that can be tuned for each data set to achieve the best classification accuracy. The proposed training procedure is tested on UCIHAR, CTG, ISOLET, and HAND data sets for which the performance consistently improves compared to the baseline across a range of confidence threshold values. The extended training procedure also results in a shift toward higher confidence values of the correctly classified samples, making the classifier not only more accurate but also more confident about its predictions.

A novel deep learning-based method for detection of weeds in vegetables.

Precision weed control in vegetable fields can substantially reduce the required weed control inputs. Rapid and accurate weed detection in vegetable fields is a challenging task due to the presence of a wide variety of weed species at various growth stages and densities. This paper presents a novel deep-learning-based method for weed detection that recognizes vegetable crops and classifies all other green objects as weeds. The optimal confidence threshold values for YOLO-v3, CenterNet, and Faster R-CNN were 0.4, 0.6, and 0.4/0.5, respectively. These deep-learning models had average precision (AP) above 97% in the testing dataset. YOLO-v3 was the most accurate model for detection of vegetables and yielded the highest score of 0.971, along with high precision and recall values of 0.971 and 0.970, respectively. The inference time of YOLO-v3 was similar to CenterNet, but significantly shorter than that of Faster R-CNN. Overall, YOLO-v3 showed the highest accuracy and computational efficiency among the deep-learning architectures evaluated in this study. These results demonstrate that deep-learning-based methods can reliably detect weeds in vegetable crops. The proposed method avoids dealing with various weed species, and thus greatly reduces the overall complexity of weed detection in vegetable fields. Findings have implications for advancing site-specific robotic weed control in vegetable crops.

Accuracy Improvement of Age Prediction Model Based on AutoML

Recommendations from websites are sometimes not suitable enough to fit the needs of different groups of people. One element in considering the recommended algorithm is the age range of users. To approach the goal of improving recommended algorithms in general, age prediction based on machine learning became the target of this research. We started by using AutoML from Google Cloud as the training tool and proportionally choosing 4,908 facial images of Asian people, which were categorized into six groups: images of the age below 10; images of the age from 10 to 20; images of the age from 20 to 30; images of the age from 30 to 40; images of the age from 40 to 50; images of the age from 50 to 60; and images of the age above 60. From the first training, we demonstrated 64.89 percentage of precision with the confidence threshold value of 0.5. After changing the threshold to 0.71, we optimized the result to 69.83 percentage of precision. By observing the result given by the deployment from Cloud training to an Android device, we found that the model differentiated Asian people aged from 10 to 40 most precisely, while it was relatively weak to distinguish people of the age from 50 to 60 and people of the age above 60. These results show an intelligent work of the prediction model on sifting people by ages and the strong relationship between the precision and the quantity of corresponding data. Its focus on distinguishing young people is suitable for our goal as young people are the main force of Internet users. By adjusting the composition of the dataset, we are looking for broader usage in other fields.

Concept Lattice Method for Spatial Association Discovery in the Urban Service Industry

A relative lag in research methods, technical means and research paradigms has restricted the rapid development of geography and urban computing. Hence, there is a certain gap between urban data and industry applications. In this paper, a spatial association discovery framework for the urban service industry based on a concept lattice is proposed. First, location data are used to form the formal context expressed by 0 and 1. Frequent closed itemsets and a concept lattice are computed on the basis of the formal context of the urban service industry. Frequent closed itemsets can filter out redundant information in frequent itemsets, uniquely determine the complete set of all frequent itemsets, and be orders of magnitude smaller than the latter. Second, spatial frequent closed itemsets and association rules discovery algorithms are designed and built based on the formal context. The inputs of the frequent closed itemsets discovery algorithms include the given formal context and frequent threshold value, while the outputs are all frequent closed itemsets and the partial order relationship between them. Newly added attributes create new concepts to guarantee the uniqueness of the new spatial association concept. The inputs of spatial association rules discovery algorithms include frequent closed itemsets and confidence threshold values, and a rule is confident when and only if its confidence degree is not less than the confidence threshold value. Third, the spatial association of the urban service industry in Nanning, China is taken as a case to verify the method. The results are basically consistent with the spatial distribution of the urban service industry in Nanning City. This study enriches the theories and methods of geography as well as urban computing, and these findings can provide guidance for location-based service planning and management of urban services.

Automated Tool for Resume Classification Using Sementic Analysis

Recruitment in the IT sector has been on the rise in recent times. Software companies are on the hunt to recruit raw talent right from the colleges through job fairs. The process of allotment of projects to the new recruits is a manual affair, usually carried out by the Human Resources department of the organization. This process of project allotment to the new recruits is a costly affair for the organization as it relies mostly on human effort. In the recent times, software companies round the world are leveraging the advances in machine learning and Artificial intelligence in general to automate routine tasks in the enterprise to increase the productivity. In the paper, we discuss the design and implementation of a resume classifier application which employs an ensemble learning based voting classifier to classify a profile of a candidate into a suitable domain based on his interest, work-experience and expertise mentioned by the candidate in the profile. The model employs topic modelling techniques to introduce a new domain to the list of domains upon failing to achieve the threshold value of confidence for the classification of the candidate profile. The Stack-Overflow REST APIs are called for the profiles which fail on the confidence threshold test set in the application. The topics returned by the APIs are subjected to topic modelling to obtain a new domain, on which the voting classifier is retrained after a fixed interval to improve the accuracy of the model.Overall, emphasis is laid out on building a dynamic machine learning automation tool which is not solely dependent on the training data in allotment of projects to the new recruits. We extended our previous work withnew learning model that has the ability to classify the resumes with better accuracy and support more new domains

"Host-insect and host-plant associated diversity in microbiota isolated from most important Oriental-Australian region egg parasitoid"

Host-insect and host-plant associated differentiation of genetically divergent microbiota were recorded from economically important egg parasitoid collected from 26 locations in India, Trichogramma chilonis constituted 86.8% of the populations collected. It was recorded from 14 host-insects, 14 different crops and weed plants from 12 states. Nine species of yeast were recorded from parasitoid from 5 host-insects with Wickerhamomyces anamalus was isolated from 36.4% samples and highest numbers were recorded from parasitoid collected on sugarcane. Bacillus cereus, Pseudomonas sp. and Stenotrophomonas maltophilia from T. chilonis constitute 64.3% of bacterial diversity based on their 16S rDNA sequences. For taxonomic identification using 16s rDNA and ITS sequences, we performed taxonomic classification of total 33 ITS isolates against UNITE Fungal ITS database and assigned taxonomy hierarchy to the sequences. Also, a total of 13 isolates 16s rDNA sequences were taxonomically assigned against RDP 16s rDNA database using RDP Naive Bayesian rRNA Classifier Version 2.1. Most of the species are correctly identified in the respective species members with high confidence threshold value support.

Chinese text classification by the Naïve Bayes Classifier and the associative classifier with multiple confidence threshold values

Each type of classifier has its own advantages as well as certain shortcomings. In this paper, we take the advantages of the associative classifier and the Naïve Bayes Classifier to make up the shortcomings of each other, thus improving the accuracy of text classification. We will classify the training cases with the Naïve Bayes Classifier and set different confidence threshold values for different class association rules (CARs) to different classes by the obtained classification accuracy rate of the Naïve Bayes Classifier to the classes. Since the accuracy rates of all selected CARs of the class are higher than that obtained by the Naïve Bayes Classifier, we could further optimize the classification result through these selected CARs. Moreover, for those unclassified cases, we will classify them with the Naïve Bayes Classifier. The experimental results show that combining the advantages of these two different classifiers better classification result can be obtained than with a single classifier.

Association classification algorithm based on structure sequence in protein secondary structure prediction

Objective To propose a novel associate classification algorithm SAC (structural association classification) and develop a compound pyramid model for accurate and precise protein secondary structure prediction. Method Based on the slide window theory, the protein sequence was treated as a window with length of 13, in which the target amino acid resided in the center, while the remaining area was targeted as secondary amino acid structures. To the head and tail of the sequence, the mirror method was employed to fill the space with an opposite- position structure in relation to the central position. In the mining process, the KDD ∗ model not only focuses on the high support and confidence rules, but also pay attention to high confidence and low support rules, which is called ‘knowledge in shortage’. Towards the end of the mining process, sets H, E and C, consisted of rule sets whose consequents are α-helix, β-sheet and C-coil, were created respectively to meet the basic requirements for the protein secondary structure prediction. The knowledge base of protein secondary structure was then established with these three newly-acquired rule sets. Through the CMAR (Classification based on Multiple Association rules) algorithm, a novel multi-classifier was developed to determine the best likelihood of a given window to the secondary structure through the adjacent information on amino acid sequential window and screening of three different rule sets. Result The protein knowledge base consisted of 8049 rules corresponding to sets H, E and C with 2642, 1895 and 3512 rules, respectively, was obtained. Experiment shows, theoretically, accuracy ratio exceeded 85% when confidence threshold value was 70% and 90%. Through the classification process using the multi-classifier SAC developed in four experiments, the significantly high accuracy and recall ratios up to 83.06% (According to Q 3 criterion, followed by abbreviation) in RS126 ( Chen & Chaudhari, 2007; Guo et al., 2004; Hu et al., 2004; Liu et al., 2004) and 80.49% in CB513 ( Guo et al., 2004; Liu et al., 2004; Wang & Liu (2004)), respectively, were demonstrated. Conclusion The structural association classification algorithm with pyramid classification developed in the present study demonstrated significantly high accuracy in the protein secondary structure prediction. The study results suggest a highly reliable and accurate alternative in the contemporary protein structure prediction.

PARTS: Probabilistic Alignment for RNA joinT Secondary structure prediction

A novel method is presented for joint prediction of alignment and common secondary structures of two RNA sequences. The joint consideration of common secondary structures and alignment is accomplished by structural alignment over a search space defined by the newly introduced motif called matched helical regions. The matched helical region formulation generalizes previously employed constraints for structural alignment and thereby better accommodates the structural variability within RNA families. A probabilistic model based on pseudo free energies obtained from precomputed base pairing and alignment probabilities is utilized for scoring structural alignments. Maximum a posteriori (MAP) common secondary structures, sequence alignment and joint posterior probabilities of base pairing are obtained from the model via a dynamic programming algorithm called PARTS. The advantage of the more general structural alignment of PARTS is seen in secondary structure predictions for the RNase P family. For this family, the PARTS MAP predictions of secondary structures and alignment perform significantly better than prior methods that utilize a more restrictive structural alignment model. For the tRNA and 5S rRNA families, the richer structural alignment model of PARTS does not offer a benefit and the method therefore performs comparably with existing alternatives. For all RNA families studied, the posterior probability estimates obtained from PARTS offer an improvement over posterior probability estimates from a single sequence prediction. When considering the base pairings predicted over a threshold value of confidence, the combination of sensitivity and positive predictive value is superior for PARTS than for the single sequence prediction. PARTS source code is available for download under the GNU public license at http://rna.urmc.rochester.edu.

The effect of threshold values on association rule based classification accuracy

Classification Association Rule Mining (CARM) systems operate by applying an Association Rule Mining (ARM) method to obtain classification rules from a training set of previously classified data. The rules thus generated will be influenced by the choice of ARM parameters employed by the algorithm (typically support and confidence threshold values). In this paper we examine the effect that this choice has on the predictive accuracy of CARM methods. We show that the accuracy can almost always be improved by a suitable choice of parameters, and describe a hill-climbing method for finding the best parameter settings. We also demonstrate that the proposed hill-climbing method is most effective when coupled with a fast CARM algorithm such as the TFPC algorithm which is also described.

A computer-assisted method for making linear radiographic measurements using stored regions of interest.

Accurate detection of changes in alveolar bone height requires radiographic measuring methods with a reliability of standard deviation (SD) of 0.15 mm or better. No periodontal digital imaging system has reached this reliability, although 3 analogue methods have achieved this goal. However, existing linear methods are time-consuming, difficult to use, unsuitable for measuring all possible (unsharp) anatomical sites and do not provide a confidence estimate for sites of change observed in serial standardized films. A rapid computer-assisted method using stored image regions of interest (ROI) has been developed which allows retest measurements for all possible sites and automatically calculates a 90%, 95%, 98%, or 99% confidence threshold value, derived from duplicate measurement variation, for sites of apparent crest height change. 28 examiners, with minimal training in operating the system, measured 14 different cement-enamel junction to crest height distances from a standard bitewing image, with and without the ROI method. The measurements were repeated 4 weeks later. 13/14 sites achieved an intra-examiner SD threshold of less than or equal to 0.15 mm with the ROI method but 0/14 without. A higher inter-examiner SD threshold of less than or equal to 0.22 mm was achieved for 13/14 sites using ROI and 0/14 without (p less than 0.001). The measurement of crest height changes in a region of previous severe vertical bone loss is demonstrated using serial films. The potential for using trabecular bone patterns as reference sites in regions where traditional measurement points are absent is demonstrated.(ABSTRACT TRUNCATED AT 250 WORDS)