The Infona portal uses cookies, i.e. strings of text saved by a browser on the user's device. The portal can access those files and use them to remember the user's data, such as their chosen settings (screen view, interface language, etc.), or their login data. By using the Infona portal the user accepts automatic saving and using this information for portal operation purposes. More information on the subject can be found in the Privacy Policy and Terms of Service. By closing this window the user confirms that they have read the information on cookie usage, and they accept the privacy policy and the way cookies are used by the portal. You can change the cookie settings in your browser.
In this work, we carry out a first exploration of the possibility of increasing the performance of Deep Neural Networks (DNNs) by applying diversity techniques to them. Since DNNs are usually very strong, weakening them can be important for this purpose. This paper includes experimental evidence of the effectiveness of binarizing multi-class problems to make beneficial the application of bagging to...
Convolutional Neural Network (CNN) is a type of feed-forward artificial neural network, exploiting the unknown structure in input distribution to discover good representations with multiple layers of small neuron collections. CNN uses relatively little pre-processing compared to other classification algorithms, usually uses gradient decent to updates the parameters in the network. Since CNN was introduced...
The article based on analyzing the theory of stock investment and the stock price prediction method, starting from the practical point of view, by describing the background and significance in Qinghai province listing Corporation stock price forecasting, which makes people aware of the importance of the stock prediction, introduces the stock prediction theory and the theory of BP neural network. This...
Since the audio recapture can be used to assist audio splicing, it is important to identify whether a suspected audio recording is recaptured or not. However, few works on such detection have been reported. In this paper, we propose an method to detect the recaptured audio based on deep learning and we investigate two deep learning techniques, i.e., neural network with dropout method and stack auto-encoders...
In this paper, we present a novel framework for handwritten numeral recognition. Considering unconstrained handwritten numerals as numeral feature vectors in the corresponding numeral vector space, we commence by reducing the coordinate dimensionalityof vector space by employing Spectral Regression Discriminant Analysis (SRDA). We then calculate the local density for all numeral classes. For each...
With low cost, ease of use, and scalability, online power estimation, which uses data obtained from battery monitoring unit (BMU) to estimate power consumption, could be a potential power estimation method for commercial smartphones. However, existing online power estimation methods exhibit high errors compared with the use of external power monitors. This is because they do not tackle three main...
Statistical learning has recently seen an expansion of applications in different areas of science, finance and industry, as it plays a great role within the fields of statistics, data mining and artificial intelligence. Hence, it intersects with areas of engineering and other disciplines as well. It is used for both regression and classification problems. Solving these problems usually involves building/training...
This paper presents a comparative pilot usability study of Dasher and an on-screen keyboard on a head-mounted display. Interaction logging data was captured along with subjective responses (via the SUS questionnaire). The results indicate that there is a strong need to develop text entry systems for smart glasses rather to simply adopt those that are already available. However, both approaches are...
In this paper we study keystroke dynamics as an authentication mechanism for touch screen based devices. The authentication process decides whether the identity of a given person is accepted or rejected. This can be easily implemented by using a two-class classifier which operates with the help of positive samples (belonging to the authentic person) and negative ones. However, collecting negative...
One of the biggest challenges in on-line signature verification is the detection of skilled forgeries. In this paper, we propose a novel scheme, based on the Kinematic Theory of rapid human movements and its associated Sigma LogNormal model, to improve the performance of on-line signature verification systems. The approach combines the high performance of DTW-based systems in verification tasks, with...
Machine Learning solutions for concept drift detection problems try to decide to what extent a particular set of examples still represents the current concept rather than treating all data equally. Monitoring the set of relevant features used to generate the classification model may be an effective strategy for concept drift detection. This paper focuses on analyzing the possibility of detecting drifts...
This paper presents a high-performance two-stage cascade CNN model. The main idea behind the cascade CNN model is complementary classification objectives between Stage I and Stage II. Discriminative learning is introduced to train Stage II by feeding back poorly recognized training samples. Experiments have been conducted on the competitive MNIST handwritten digit database. The cascade model achieved...
In this paper, automatic speaker verification using whispered speech is explored. In the past, whispered speech has been shown to convey relevant speaker identity and gender information, nevertheless it is not clear how to efficiently use this information in speech-based biometric systems. This study compares the performance of three different speaker verification systems trained and tested under...
A training protocol for learning deep neural networks, called greedy layer-wise training, is applied to the evolution of a hierarchical, feed-forward Genetic Programming based system for feature construction and object recognition. Results on a popular handwritten digit recognition benchmark clearly demonstrate that two layers of feature transformations improves generalisation compared to a single...
Dynamic motion of human shows kinematic aspects related to storing elastic energy in skeletal muscle. This results from joint stiffness modulation and as a consequence, countermovement which is opposite to the intended motion is observed. We propose a segmentation algorithm based on a hidden semi-Markov model that infers dynamic motion phases probabilistically from sEMG observations during countermovement...
Fingerprint recognition systems are vulnerable to spoof attacks, which consist in presenting forged fingerprints to the sensor. Typical anti-spoofing mechanism is fingerprint liveness detection. Existing liveness detection methods are still not robust to spoofing materials, datasets and sensor variations. In particular, the performance of a liveness detection algorithm remarkably drops upon encountering...
Masked implementations of cryptographic algorithms are often used in commercial embedded cryptographic devices to increase their resistance to side channel attacks. In this work we show how neural networks can be used to both identify the mask value, and to subsequently identify the secret key value with a single attack trace with high probability. We propose the use of a pre-processing step using...
In this paper, we propose a new method that allows, by finding the convex hull of a character image, to set out in one pass only, the control parameters of a particular character distortion process. This character distortion method can then be applied to normalize the character image, i.e. to reduce the within-class scatter of images of handwritten characters, which could lead to a significant improvement...
This work presents training methods and recognition experiments for phoneme-wise tied mixture densities in hidden Markov models (HMM). The system trains speaker dependent, but vocabulary independent, phoneme models for the recognition of Finnish words. The Learning Vector Quantization (LVQ) methods are applied to increase the discrimination between the phoneme models. A segmental LVQ3 training is...
We investigate the use of deep neural nets (DNN) to provide initial speaker change points in a speaker diarization system. The DNN trains states that correspond to the location of the speaker change point (SCP) in the speech segment input to the DNN. We model these different speaker change point locations in the DNN input by 10 to 20 states. The confidence in the SCP is measured by the number of frame...
Set the date range to filter the displayed results. You can set a starting date, ending date or both. You can enter the dates manually or choose them from the calendar.