scholarly journals A novel model for voice command fingerprinting using deep learning

2022 ◽  
Vol 65 ◽  
pp. 103085
Jianghan Mao ◽  
Chenyu Wang ◽  
Yanhui Guo ◽  
Guoai Xu ◽  
Shoufeng Cao ◽  
Trung Minh Nguyen ◽  
Thien Huu Nguyen

The previous work for event extraction has mainly focused on the predictions for event triggers and argument roles, treating entity mentions as being provided by human annotators. This is unrealistic as entity mentions are usually predicted by some existing toolkits whose errors might be propagated to the event trigger and argument role recognition. Few of the recent work has addressed this problem by jointly predicting entity mentions, event triggers and arguments. However, such work is limited to using discrete engineering features to represent contextual information for the individual tasks and their interactions. In this work, we propose a novel model to jointly perform predictions for entity mentions, event triggers and arguments based on the shared hidden representations from deep learning. The experiments demonstrate the benefits of the proposed method, leading to the state-of-the-art performance for event extraction.

Jun Xiao ◽  
Hao Ye ◽  
Xiangnan He ◽  
Hanwang Zhang ◽  
Fei Wu ◽  

Factorization Machines (FMs) are a supervised learning approach that enhances the linear regression model by incorporating the second-order feature interactions. Despite effectiveness, FM can be hindered by its modelling of all feature interactions with the same weight, as not all feature interactions are equally useful and predictive. For example, the interactions with useless features may even introduce noises and adversely degrade the performance. In this work, we improve FM by discriminating the importance of different feature interactions. We propose a novel model named Attentional Factorization Machine (AFM), which learns the importance of each feature interaction from data via a neural attention network. Extensive experiments on two real-world datasets demonstrate the effectiveness of AFM. Empirically, it is shown on regression task AFM betters FM with a 8.6% relative improvement, and consistently outperforms the state-of-the-art deep learning methods Wide&Deep [Cheng et al., 2016] and DeepCross [Shan et al., 2016] with a much simpler structure and fewer model parameters. Our implementation of AFM is publicly available at:

2020 ◽  
Vol 12 (20) ◽  
pp. 3314
Zhang Yubo ◽  
Yan Zhuoran ◽  
Yang Jiuchun ◽  
Yang Yuanyuan ◽  
Wang Dongyan ◽  

In recent decades, land use/cover change (LUCC) due to urbanization, deforestation, and desertification has dramatically increased, which changes the global landscape and increases the pressure on the environment. LUCC not only accelerates global warming but also causes widespread and irreversible loss of biodiversity. Therefore, LUCC reconstruction has important scientific and practical value for studying environmental and ecological changes. The commonly used LUCC reconstruction models can no longer meet the growing demand for uniform and high-resolution LUCC reconstructions. In view of this circumstance, a deep learning-integrated LUCC reconstruction model (DLURM) was developed in this study. Zhenlai County of Jilin Province (1986–2013) was taken as an example to verify the proposed DLURM. The average accuracy of the DLURM reached 92.87% (compared with the results of manual interpretation). Compared with the results of traditional models, the DLURM had significantly better accuracy and robustness. In addition, the simulation results generated by the DLURM could match the actual land use (LU) map better than those generated by other models.

Yantao Yu ◽  
Zhen Wang ◽  
Bo Yuan

Factorization machines (FMs) are a class of general predictors working effectively with sparse data, which represents features using factorized parameters and weights. However, the accuracy of FMs can be adversely affected by the fixed representation trained for each feature, as the same feature is usually not equally predictive and useful in different instances. In fact, the inaccurate representation of features may even introduce noise and degrade the overall performance. In this work, we improve FMs by explicitly considering the impact of individual input upon the representation of features. We propose a novel model named \textit{Input-aware Factorization Machine} (IFM), which learns a unique input-aware factor for the same feature in different instances via a neural network. Comprehensive experiments on three real-world recommendation datasets are used to demonstrate the effectiveness and mechanism of IFM. Empirical results indicate that IFM is significantly better than the standard FM model and consistently outperforms four state-of-the-art deep learning based methods.

2021 ◽  
Nour Salim Nassar

Abstract Recommender systems are everywhere books, products, movies, and more. Traditional recommender systems typically use a single criterion in the recommendation, while studies have shown that multi-criteria recommending is more accurate. Novel deep learning techniques have produced remarkable achievements in many fields. The use of such techniques in recommendation systems has started to get attention recently, and several models of recommendation have been proposed based on deep learning. However, there is still no work for using deep learning in hybrid multi-criteria recommender systems. In this work, a model for a hybrid deep multi-criteria recommender system was presented. The model mainly includes two major parts: In the first one, the model obtains the user ID, item ID, and the item metadata to be used as input to a deep neural network in order to predict the criteria ratings. In the second part, the obtained ratings act as an input to another deep neural network, where the overall rating is predicted. Our experiments were conducted on a real-world dataset. They demonstrated the superiority of the proposed novel model over the other models in all measures used to evaluate performance. This indicates the successful use of hybrid deep multi-criteria in the recommendation systems.

2022 ◽  
pp. 125-142
Vijay Srinivas Srinivas Tida ◽  
Raghabendra Shah ◽  
Xiali Hei

The laser-based audio signal injection can be used for attacking voice controllable systems. An attacker can aim an amplitude-modulated light at the microphone's aperture, and the signal injection acts as a remote voice-command attack on voice-controllable systems. Attackers are using vulnerabilities to steal things that are in the form of physical devices or the form of virtual using making orders, withdrawal of money, etc. Therefore, detection of these signals is important because almost every device can be attacked using these amplitude-modulated laser signals. In this project, the authors use deep learning to detect the incoming signals as normal voice commands or laser-based audio signals. Mel frequency cepstral coefficients (MFCC) are derived from the audio signals to classify the input audio signals. If the audio signals are identified as laser signals, the voice command can be disabled, and an alert can be displayed to the victim. The maximum accuracy of the machine learning model was 100%, and in the real world, it's around 95%.

Fuxing Hong ◽  
Dongbo Huang ◽  
Ge Chen

Factorization Machine (FM) is a widely used supervised learning approach by effectively modeling of feature interactions. Despite the successful application of FM and its many deep learning variants, treating every feature interaction fairly may degrade the performance. For example, the interactions of a useless feature may introduce noises; the importance of a feature may also differ when interacting with different features. In this work, we propose a novel model named Interaction-aware Factorization Machine (IFM) by introducing Interaction-Aware Mechanism (IAM), which comprises the feature aspect and the field aspect, to learn flexible interactions on two levels. The feature aspect learns feature interaction importance via an attention network while the field aspect learns the feature interaction effect as a parametric similarity of the feature interaction vector and the corresponding field interaction prototype. IFM introduces more structured control and learns feature interaction importance in a stratified manner, which allows for more leverage in tweaking the interactions on both feature-wise and field-wise levels. Besides, we give a more generalized architecture and propose Interaction-aware Neural Network (INN) and DeepIFM to capture higher-order interactions. To further improve both the performance and efficiency of IFM, a sampling scheme is developed to select interactions based on the field aspect importance. The experimental results from two well-known datasets show the superiority of the proposed models over the state-of-the-art methods.

2021 ◽  
Vol 21 (1) ◽  
Chenxi Sun ◽  
Hongna Dui ◽  
Hongyan Li

Abstract Background Disease prediction based on electronic health records (EHRs) is essential for personalized healthcare. But it’s hard due to the special data structure and the interpretability requirement of methods. The structure of EHR is hierarchical: each patient has a sequence of admissions, and each admission has some co-occurrence diagnoses. However, the existing methods only partially model these characteristics and lack the interpretation for non-specialists. Methods This work proposes a time-aware and co-occurrence-aware deep learning network (TCoN), which is not only suitable for EHR data structure but also interpretable: the co-occurrence-aware self-attention (CS-attention) mechanism and time-aware gated recurrent unit (T-GRU) can model multilevel relations; the interpretation path and the diagnosis graph can make the result interpretable. Results The method is tested on a real-world dataset for mortality prediction, readmission prediction, disease prediction, and next diagnoses prediction. Experimental results show that TCoN is better than baselines with 2.01% higher accuracy. Meanwhile, the method can give the interpretation of causal relationships and the diagnosis graph of each patient. Conclusions This work proposes a novel model—TCoN. It is an interpretable and effective deep learning method, that can model the hierarchical medical structure and predict medical events. The experiments show that it outperforms all state-of-the-art methods. Future work can apply the graph embedding technology based on more knowledge data such as doctor notes.

2021 ◽  
Vol 14 (1) ◽  
pp. 326
Bingqing Huang ◽  
Haonan Zheng ◽  
Xinbo Guo ◽  
Yi Yang ◽  
Ximing Liu

Deep learning models are playing an increasingly important role in time series forecasting with their excellent predictive ability and the convenience of not requiring complex feature engineering. However, the existing deep learning models still have shortcomings in dealing with periodic and long-distance dependent sequences, which lead to unsatisfactory forecasting performance on this type of dataset. To handle these two issues better, this paper proposes a novel periodic time series forecasting model based on DA-RNN, called DA-SKIP. Using the idea of task decomposition, the novel model, based on DA-RNN, GRU-SKIP and autoregressive component, breaks down the prediction of periodic time series into three parts: linear forecasting, nonlinear forecasting and periodic forecasting. The results of the experiments on Solar Energy, Electricity Consumption and Air Quality datasets show that the proposed model outperforms the three comparison models in capturing periodicity and long-distance dependence features of sequences.

Sign in / Sign up

Export Citation Format

Share Document