Cyclical Curriculum Learning.

Artificial neural networks (ANNs) are inspired by human learning. However, unlike human education, classical ANN does not use a curriculum. Curriculum learning (CL) refers to the process of ANN training in which samples are used in a meaningful order. When using CL, training begins with a subset of the dataset and new samples are added throughout the training, or training begins with the entire dataset and the number of samples used is reduced. With these changes in training dataset size, better results can be obtained with curriculum, anti-curriculum, or random-curriculum methods than the vanilla method. However, a generally efficient CL method for various architectures and datasets is not found. In this article, we propose cyclical CL (CCL), in which the data size used during training changes cyclically rather than simply increasing or decreasing. Instead of using only the vanilla method or only the curriculum method, using both methods cyclically like in CCL provides more successful results. We tested the method on 18 different datasets and 15 architectures in image and text classification tasks and obtained more successful results than no-CL and existing CL methods. We also have shown theoretically that it is less erroneous to apply CL and vanilla cyclically instead of using only CL or only the vanilla method. The code of the cyclical curriculum is available at https://github.com/CyclicalCurriculum/Cyclical-Curriculum.

Performance evaluation of the machine learning algorithms used in inference mechanism of a medical decision support system.

The importance of the decision support systems is increasingly supporting the decision making process in cases of uncertainty and the lack of information and they are widely used in various fields like engineering, finance, medicine, and so forth, Medical decision support systems help the healthcare personnel to select optimal method during the treatment of the patients. Decision support systems are intelligent software systems that support decision makers on their decisions. The design of decision support systems consists of four main subjects called inference mechanism, knowledge-base, explanation module, and active memory. Inference mechanism constitutes the basis of decision support systems. There are various methods that can be used in these mechanisms approaches. Some of these methods are decision trees, artificial neural networks, statistical methods, rule-based methods, and so forth. In decision support systems, those methods can be used separately or a hybrid system, and also combination of those methods. In this study, synthetic data with 10, 100, 1000, and 2000 records have been produced to reflect the probabilities on the ALARM network. The accuracy of 11 machine learning methods for the inference mechanism of medical decision support system is compared on various data sets.

Effects of different topical agents on enamel demineralization around orthodontic brackets: an in vivo and in vitro study.

BACKGROUND: The aim of this study was to evaluate the in vivo and in vitro effects of a casein phosphopeptide-amorphous calcium phosphate (CPP-ACP) and fluoride containing topical agents in reducing enamel demineralization around orthodontic brackets, and to compare this with a control group. METHODS: Twenty-one patients and 60 extracted premolars were divided into three groups: two experimental and one control. Tooth Mousse® (CPP-ACP gel; GC-Corp, Tokyo, Japan) and Fluoridin N5® (Fluoride gel; Voco-GmbH, Cuxhaven, Germany) were applied to tooth surfaces around orthodontic brackets in the experimental groups. Teeth were extracted after 60 days to evaluate the in vivo effects of the testing materials. For the in vitro experiment, samples were cycled for 14 days through a daily procedure of demineralization. All teeth were sectioned and evaluated by superficial microhardness analysis. An indentation was made from two positions (occlusal-cervical) and one depth (10 µm). RESULTS: Comparisons of occlusal and cervical microhardness scores for all specimens showed no statistically significant side differences. A multiple comparison test showed that the use of CPP-ACP and fluoride containing topical gels were more significantly efficient than the control group (p<0.001). No significant differences were detected between CPP-ACP and the fluoride groups against demineralization. CONCLUSIONS: In vivo and in vitro evaluations indicated that CPP-ACP and fluoride containing agents successfully inhibited caries around orthodontic brackets.

In vivo effects of amorphous calcium phosphate-containing orthodontic composite on enamel demineralization around orthodontic brackets.

BACKGROUND: The aim of this study was to evaluate the in vivo effects of an amorphous calcium phosphate-containing orthodontic composite in reducing enamel demineralization around orthodontic brackets, and to compare it with the control. METHODS: Fourteen orthodontic patients were divided into two equal groups. They received brackets fitted to all first premolars, bonded with either Aegis Ortho® (The Bosworth Co.), an ACP-containing orthodontic composite (experimental group), or Concise® (3M Dental Products), a resin-based orthodontic composite (control group). After 30 days, the teeth were extracted and longitudinally sectioned, and evaluated by superficial-microhardness analysis. The determinations were made at the bracket edge cementing limits and at occlusal and cervical points 100 and 200 µm away from the edge. In all of these positions, indentations were made at depths of 10, 20, 30, 50, 70, and 90 µm from the enamel surface. Analysis of variance (ANOVA) and Tukey post hoc test was used. The statistical significance level was set at p<0.05. RESULTS: The ANOVA showed statistically significant differences for position, material, depth, and their interactions (p<0.001). The multiple comparison test showed that the ACP-containing orthodontic composite was significantly more efficient than the control composite, reducing enamel demineralization in almost all evaluations (p<0.001). CONCLUSIONS: Present results indicated that ACP-containing orthodontic composite for bonding orthodontic brackets successfully inhibited demineralization in vivo. This effect was localized to the area around the brackets and was statistically significant after 30 days.

Effect of periosteal stimulation therapy on bone formation in orthopedically expanded suture in rats.

OBJECTIVE: The aim of this experimental study was to evaluate the effects of periosteal stimulation therapy (PST), on bone regeneration in response to expansion of the interpremaxillary suture, in rats. MATERIAL AND METHODS: Sixteen male, 50-60 days old Wistar rats were divided into two equal groups (control and experimental). Both groups were subjected to expansion for 5 days, and 30 cN of force was applied to the maxillary incisors with helical spring. On the 2nd day of the expansion procedure in the experimental group, the periosteum over the interpremaxillary suture was stimulated locally by intraperiosteal penetrations with a hypodermic needle. Bone regeneration in the interpremaxillary suture was evaluated by bone histomorphometry and the new bone area, bone perimeter, feret's diameter and new bone/old bone percentage parameters were evaluated. Mann-Whitney U-test was used for statistical evaluation at p < 0.05 level. RESULTS: Significant differences were found between groups for all investigated histomorphometric parameters. New bone area (p < 0.01), bone perimeter (p < 0.05), feret's diameter (p < 0.01) and new bone/old bone percentage (p < 0.01) measurements were significantly higher in the experimental group than that in the control. Bone histomorphometric measurements revealed that bone architecture in the PST group was improved. CONCLUSIONS: Local stimulation of the periosteum of the sutural area during expansion stimulates bone formation and improves healing.

Microleakage between composite/wire and composite/enamel interfaces of flexible spiral wire retainers: direct versus indirect application methods.

OBJECTIVE: The aim of this in vitro study was to compare the microleakage of flexible spiral wire retainers (FSWR) at composite/wire and composite/enamel interfaces produced by an indirect application method to that of a conventional direct application method. MATERIALS AND METHODS: Forty freshly extracted human mandibular incisor teeth were randomly divided into two equal groups. Group 1 was bonded directly according to the manufacturer's recommendations. Group 2 consisted of 20 teeth bonded indirectly with Transbond LR as the adhesive and Sondhi Rapid Set A/B Primer (3M Unitek, Monrovia, CA, USA), a filled resin primer. After bonding, specimens were further sealed with nail varnish, stained with 0.5% basic fuchsine for 24 h, sectioned and examined under a stereomicroscope that measured microleakage at the composite/wire and composite/enamel interfaces from both mesial and distal margins. Statistical analyses were performed using Wilcoxon and Mann-Whitney U-tests with Bonferroni correction. RESULTS: Statistical comparisons indicated that no statistically significant differences were observed between composite/wire and composite/enamel interfaces for direct vs. indirect application procedures (Direct method p = 0.630 and Indirect method p = 0.930). Comparisons of the microleakage scores between direct and indirect FSWR application groups at composite/wire and composite/enamel interfaces indicated no statistically significant microleakage differences at mesial or distal margins (Composite/enamel interface p = 0.361 and Composite/wire interface p = 0.270). CONCLUSION: The type of FSWR application procedures (direct vs. indirect) did not significantly affect the amount of microleakage at the enamel/composite/wire complex.

Cline: a new decision-tree family.

A new family of algorithm called Cline that provides a number of methods to construct and use multivariate decision trees is presented. We report experimental results for two types of data: synthetic data to visualize the behavior of the algorithms and publicly available eight data sets. The new methods have been tested against 23 other decision-tree construction algorithms based on benchmark data sets. Empirical results indicate that our approach achieves better classification accuracy compared to other algorithms.