Detonation Classification From Acoustic Signature With the Restricted Boltzmann Machine

Abstract

We compare the recently proposed Discriminative Restricted Boltzmann Machine (DRBM) to the classical Support Vector Machine (SVM) on a challenging classification task consisting in identifying weapon classes from audio signals. The three weapon classes considered in this work (mortar, rocket, and rocket‐propelled grenade), are difficult to reliably classify with standard techniques because they tend to have similar acoustic signatures. In addition, specificities of the data available in this study make it challenging to rigorously compare classifiers, and we address methodological issues arising from this situation. Experiments show good classification accuracy that could make these techniques suitable for fielding on autonomous devices. DRBMs appear to yield better accuracy than SVMs, and are less sensitive to the choice of signal preprocessing and model hyperparameters. This last property is especially appealing in such a task where the lack of data makes model validation difficult.

Short-term output power forecasting of photovoltaic systems based on the deep belief net

“…DBN model is stacked by the restricted Boltzmann machine (RBM) based on energy function. 25,26 For a given set of states (v, h), v represents the visible layer, h represents the hidden layer, and its energy function is defined as follows…”

Enhancing Single Speaker Recognition Using Deep Belief Network

“…In such case, the ensemble learning can be applied that can learn any nonlinear boundary through appropriately combining the simple classifiers. It has potential ability to reduce over fitting problems greatly, to decrease the risk of a single classifier, and to obtain better performance than its single classifiers [20]. The usual ensemble classifiers are boost-based, bagging-based approaches [21], random subspace [22], and so forth.…”

Avaliação da Qualidade da Voz em Serviços de Comunicação usando Deep Learning

“…RBMé uma RNA estocástica generativa, treinada de forma não supervisionada e para problemas que necessitam de classificação,é necessário adicionar um método de aprendizagem supervisionado, classificando as amostras com base nas características extraídas pelo RBM. Estudos [Bengio et al 2012] sobre identificação de classes em sinais de voz demonstram uma acurácia superior da Máquina de Boltzmann Restrita Discriminativa (DRBM, do inglês Discriminative Restricted Boltzmann Machine) em relação ao clássico uso da máquina de vetor de suporte (SVM, do inglês Support Vector Machine). De acordo com [Bengio et al 2012], técnicas padrões de aprendizagem de máquina possuem dificuldades em classificar assinaturas acústicas similares.…”

“…Estudos [Bengio et al 2012] sobre identificação de classes em sinais de voz demonstram uma acurácia superior da Máquina de Boltzmann Restrita Discriminativa (DRBM, do inglês Discriminative Restricted Boltzmann Machine) em relação ao clássico uso da máquina de vetor de suporte (SVM, do inglês Support Vector Machine). De acordo com [Bengio et al 2012], técnicas padrões de aprendizagem de máquina possuem dificuldades em classificar assinaturas acústicas similares.…”