The system that automatically identifies the anthropometric fingerprint is one of systems interact directly with user, which every day will be provided a diverse database. This requires to optimized handle process meet needs users such as fast processing time, almost absolute accuracy, no errors in real process. Therefore, this paper, we propose application machine learning methods develop classification algorithms based on singularity feature. goal paper reduce number comparisons automatic...

We present a new light supervision method to derive additional acoustic training data automatically for broadcast news transcription systems. A subset of the TDT corpus, which consists audio with corresponding closed-caption (CC) transcripts, is identified by aligning CC transcripts and hypotheses generated lightly-supervised decoding. Phrases three or more contiguous words, on both decoder's agree, are selected. The selection yields 702 hours, 72% captioned data. When adding 700 hours...

We describe a software agent that learns to find information on the World Wide Web (WWW), deciding what new pages might interest user. The maintains separate hotlist (for links were interesting) and coldlist not for each topic. By analyzing immediately accessible from link, types of user is interested in. This can be used inform when interesting page becomes available or order user's exploration unseen existing so more promising ones are investigated first. compare four different learning...

Distance learning has dramatically increased in recent years because of advanced technology. In addition, numerous universities had to offer courses online mode 2020 and 2021 the COVID-19 pandemic. However, there are more challenges distance than traditional method (e.g., feedback interaction). Recently, researchers started using simple EEG headsets identify confused students during based on machine approaches. they faced unpleasant accuracy algorithms or nondeep neural networks. this paper,...

Early prediction of the quality software modules prior to testing and operations can yield great benefits development teams, especially those high-assurance mission-critical systems. Such an estimation allows effective use resources improve system that need it most achieve high reliability. To reliability, by means predictive methods, several tools are available. Software classification models provide a class module, i.e., fault-prone or not fault-prone. Recent advances in data mining field...

Anti-theft problem has been challenging since it mainly depends on the existence of external devices to defend from thefts. Recently, driver behavior analysis using supervised learning investigated with goal detect burglary by identifying drivers. In this paper, we propose a data-driven technique, LiveDI, which uses driving removing use in order identify The built model utilizes Gated Recurrent Unit (GRU) and Fully Convolutional Networks (FCN) learn long-short term patterns behaviors...

We applied neural network language model (NNLM) on Chinese by training and testing it 2011 GALE Mandarin evaluation task. Exploiting the fact that there are no word boundaries in written Chinese, we trained various NNLMs using either word, character, or both, including a wordcharacter hybrid-input NNLM which accepts both character as input. Our best result showed up to 0.6% absolute (6.3% relative) Character Error Rate (CER) reduction compared an un-pruned 4-gram standard 0.2% (2.6% CER...

The novel approach for automatic detection and classification of road defects is proposed based on shape texture features analysis. system includes three main steps: position detection, feature contour extraction followed by defects. implem ented in Matlab digital images analysis combined with machine learning algorithms such as the random forest algorithm boosting. Segmentation implemented using graph-cuts method Markov fields. efficiency demonstrated real data set.

The objective of this paper is to propose a robust approach building computer vision system detect and classify pavement defects based on features, such as the contour feature (chain code histogram, Hu-moment), shape an object (length, width, area). In paper, we present method build automated different types rupture road edge, potholes, subsidence depressions image processing techniques machine learning methods. That includes following steps. First step defect position (ROI) then described...

This study investigates the attribute selection problem for reducing number of software metrics (program attributes) used by a case-based reasoning (CBR) quality classification model. The are selected using Kolmogorov-Smirnov (K-S) two sample test. "modified expected cost misclassification" measure, recently proposed our research team, is as performance measure to select, evaluate, and compare models. procedure presented in this paper can assist development organization determining that...

The increased reliance on computer systems in the modern world has created a need for engineering reliability control of to highest possible standards. This is especially crucial high-assurance and mission critical systems. Software quality classification models are one important tools achieving high reliability. They can be used calibrate software metrics-based detect fault-prone modules. Timely use such greatly aid detecting faults early life cycle product. Individual classifiers (models)...

This paper describes the BBN real-time recognition systems used in 2004 Rich Transcription (RT) benchmark test for English Conversational Telephone Speech (CTS) and Broadcast News (BN) tasks. We describe system architecture, along withthe algorithms weused inorder to reduce computation with minimal impact on accuracy. Particular choices design of thefinal are analyzed toshow trade-offs between speed also present recently developed new architecture systems, which outperforms we submitted RT04...

In this paper, we present a method to extract probabilistic acoustic features by using the Adaptive Boosting algorithm (AdaBoost). We build phoneme Gaussian mixture classifiers, and use AdaBoost enhance classification performance. The outputs from are posterior probabilities for each frame given all phonemes. Those then used train new model in similar way as original features. gains obtained when combine them with baseline PLP. Adaboost systems both Arabic Mandarin have contributed on final...