This paper reviews recent developments in deep learning and unsupervised feature learning for time-series modeling. It highlights the unique challenges of time-series data, such as high dimensionality, noise, and temporal dependencies, and discusses traditional approaches like autoregressive models and Hidden Markov Models (HMMs). The paper introduces unsupervised feature learning techniques, including Restricted Boltzmann Machines (RBM), Conditional RBM (cRBM), Gated RBM (GRBM), Auto-encoders, Recurrent Neural Networks (RNN), and Convolutional Neural Networks (CNN). These methods are designed to capture temporal relationships and handle complex, high-dimensional data. The paper also reviews classical time-series problems, such as video analysis, stock market prediction, speech recognition, music recognition, and motion capture data modeling, where these techniques have been applied. It emphasizes the importance of selecting appropriate models based on the properties of the data and the specific task requirements, and suggests future directions for improving these models.This paper reviews recent developments in deep learning and unsupervised feature learning for time-series modeling. It highlights the unique challenges of time-series data, such as high dimensionality, noise, and temporal dependencies, and discusses traditional approaches like autoregressive models and Hidden Markov Models (HMMs). The paper introduces unsupervised feature learning techniques, including Restricted Boltzmann Machines (RBM), Conditional RBM (cRBM), Gated RBM (GRBM), Auto-encoders, Recurrent Neural Networks (RNN), and Convolutional Neural Networks (CNN). These methods are designed to capture temporal relationships and handle complex, high-dimensional data. The paper also reviews classical time-series problems, such as video analysis, stock market prediction, speech recognition, music recognition, and motion capture data modeling, where these techniques have been applied. It emphasizes the importance of selecting appropriate models based on the properties of the data and the specific task requirements, and suggests future directions for improving these models.