Film industry becomes one of the largest amongst the other economic sectors worldwide nowadays. This industry also has a significant impact in the global economic market. Many movies have been produced each year. Howe...
详细信息
Learning user preferences become very important as the personalization systems grow rapidly in this current era. Offering special and personal services can be an added value for the companies to maintain their custome...
详细信息
This paper tackles the short-term hydro-power unit commitment problem in a multi-reservoir system - a cascade-based operation scenario. For this, we propose a new mathematical modelling in which the goal is to maximiz...
详细信息
Context: Recent laws to ensure the security and protection of personal data establish new software requirements. Consequently, new technologies are needed to guarantee software quality under the perception of privacy ...
详细信息
Due to advances in deep learning, the performance of automatic beat and downbeat tracking in musical audio signals has seen great improvement in recent years. In training such deep learning based models, data augmenta...
详细信息
Preference Elicitation is now considered a crucial stage in recommender system. That is the stage where we collect and query preferences of the users as part of interactive decision support system. Preference elicitat...
详细信息
ISBN:
(数字)9781665404228
ISBN:
(纸本)9781665404235
Preference Elicitation is now considered a crucial stage in recommender system. That is the stage where we collect and query preferences of the users as part of interactive decision support system. Preference elicitation can be performed using many ways, such as asking the user explicitly to give ratings or feedbacks for some products, or implicitly observing their behaviour. Once this preference has been collected by the system, it can then be used to generate item recommendation. Using explicit feedback itself, such as rating only, has several drawbacks. Using ratings is very subjective. Two users with similar taste might give different rating to the same item. A user also may not be consistent in giving number to express their liking. In other case, a user may like an item more than the other items, but feels difficult to give lower rating on the other item, because it means that the particular item is just as bad as the one he/she does not like. By showing two different items as a pair, it will be easier for the users to pick which one is the best. In this paper, we show the result of our experiment in predicting user pairwise preferences by using their movie ratings data. We evaluated five different machine learning algorithms to predict the data in pairwise preference format. The result shows that K-Nearest Neighbour and Random Forest outperformed the other algorithms.
This paper presents a novel system architecture that integrates blind source separation with joint beat and downbeat tracking in musical audio signals. The source separation module segregates the percussive and non-pe...
详细信息
In this study, deep embedding of acoustic and articulatory features are combined for speaker identification. First, a convolutional neural network (CNN)-based universal background model (UBM) is constructed to generat...
ISBN:
(数字)9781509066315
ISBN:
(纸本)9781509066322
In this study, deep embedding of acoustic and articulatory features are combined for speaker identification. First, a convolutional neural network (CNN)-based universal background model (UBM) is constructed to generate acoustic feature (AC) embedding. In addition, as the articulatory features (AFs) represent some important phonological properties during speech production, a multilayer perceptron (MLP)-based AF embedding extraction model is also constructed for AF embedding extraction. The extracted AC and AF embeddings are concatenated as a combined feature vector for speaker identification using a fully-connected neural network. This proposed system was evaluated by three corpora consisting of King-ASR, LibriSpeech and SITW, and the experiments were conducted according to the properties of the datasets. We adopted all three corpora to evaluate the effect of AF embedding, and the results showed that combining AF embedding into the input feature vector improved the performance of speaker identification. The LibriSpeech corpus was used to evaluate the effect of the number of enrolled speakers. The proposed system achieved an EER of 7.80% outperforming the method based on x-vector with PLDA (8.25%). And we further evaluated the effect of signal mismatch using the SITW corpus. The proposed system achieved an EER of 25.19%, which outperformed the other baseline methods.
This paper aims to improve speaker embedding representation based on x-vector for extracting more detailed information for speaker verification. We propose a statistics pooling time delay neural network (TDNN), in whi...
ISBN:
(数字)9781509066315
ISBN:
(纸本)9781509066322
This paper aims to improve speaker embedding representation based on x-vector for extracting more detailed information for speaker verification. We propose a statistics pooling time delay neural network (TDNN), in which the TDNN structure integrates statistics pooling for each layer, to consider the variation of temporal context in frame-level transformation. The proposed feature vector, named as statsvector, are compared with the baseline x-vector features on the VoxCeleb dataset and the Speakers in the Wild (SITW) dataset for speaker verification. The experimental results showed that the proposed stats-vector with score fusion achieved the best performance on VoxCeleb1 dataset. Furthermore, considering the interference from other speakers in the recordings, we found that the proposed statsvector efficiently reduced the interference and improved the speaker verification performance on the SITW dataset.
The scale of modem Artificial Intelligence systems has been growing and entering more research territories by incorporating Deep Learning (DL) and Deep Reinforcement Learning (DRL) methods. More specifically, multi-ag...
详细信息
ISBN:
(数字)9781728187082
ISBN:
(纸本)9781728187099
The scale of modem Artificial Intelligence systems has been growing and entering more research territories by incorporating Deep Learning (DL) and Deep Reinforcement Learning (DRL) methods. More specifically, multi-agent DRL methods have been widely applied to address the problems of high-dimensional computation, which interpret the conditions that real-world systems mainly encounter and the issues that require resolving. However, the current approaches of DL and DRL are often challenged for their untransparent and time-consuming modeling processes in their attempt to achieve a practical and applicable inference based on human-level perspective and acceptance. This paper presents an explainable and adaptable augmented knowledge attention network for multi-agent DRL systems, which uses game theory simulation to tackle the problem of non-stationarity at the beginning, while improving the learning exploration built upon the strategic ontology to achieve the learning convergence more efficiently for autonomous agents. We anticipate that our approach will facilitate future research studies and potential research inspections of emerging multi-agent DRL systems for increasingly complex and autonomous environments.
暂无评论