AKAnil Kumar

Anil Kumar

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Anil Kumar has a total of 10 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Deep Learning Approach with Attention Mechanism for Multivariate Mixed Traffic Flow Prediction

    by Anil Kumar, Shiv Kumar Verma, Raju Ranjan

    Traffic congestion is a major challenge in urban areas, primarily driven by industrialization, urbanization, and population growth. Governments and transportation agencies worldwide are actively seeking effective solutions to mitigate this issue. This study addresses traffic prediction under mixed traffic conditions by categorizing vehicles, which is essential for understanding and managing heterogeneous traffic flows. We propose a predictive model that forecasts traffic in terms of categorized vehicle speed and volume. The model integrates a Convolutional Long Short-Term Memory (ConvLSTM) neural network with an LSTM-based attention mechanism to enhance spatiotemporal learning capabilities. It is trained on seven multivariate features, including categorized vehicular speeds, time of day, and day of the week, using the same set of features as input for forecasting. The performance of the proposed model is evaluated against various baseline neural network architectures, including Dense, Gated Recurrent Unit (GRU), Long Short-Term Memory (LSTM), and ConvLSTM. Experimental results demonstrate that the proposed model outperforms these baselines in terms of prediction accuracy and stability. In particular, compared with the ConvLSTM model, the proposed approach achieves improvements of 48.06% in mean squared error (MSE), 14.40% in mean absolute error (MAE), and 14.51% in mean pinball loss (MPL), demonstrating the effectiveness of incorporating the attention mechanism for improved traffic forecasting under mixed traffic conditions.

  • Open Access

    Frequency-attention based Non-autoregressive Transformer (FANT) for long-sequence time series forecasting

    by Preeti, Vanshika Singh, Geethanjali Kher, Manisha Bansal, Vikash Chaudhary, Aruna Jain, Anil Kumar, Rajni Bala, Ram Pal Singh

    Long-sequence multivariate time series forecasting is widely applicable across various real-world scenarios. The Vanilla transformer has recently been used to predict the long-sequence time series. Transformers are initially designed for tasks like machine translation, where the order of the input sequence is crucial, but the specific temporal dependencies of the time series may not be fully captured. Moreover, its autoregressive nature poses limitations. This paper proposes a novel approach named FANT (frequency-attention-based non-autoregressive transformer) tailored for accurately forecasting long-sequence multivariate time series. The FANT method decomposes the time series into trend and seasonality components and enriches the input data with a calculated time vector using TimetoVec. This augmentation aids in capturing both short-term and long-term dependencies within the time series, facilitating the model’s comprehension of intricate temporal relationships. The approach employs the Fourier transform to shift the data into the frequency domain, where self-attention is subsequently calculated to discern patterns and periodicity within the data. Importantly, the FANT model is non-autoregressive, enabling simultaneous generation of all sequence elements. This characteristic minimises error propagation and reduces computational time, enhancing overall efficiency. The empirical evaluation of the FANT model encompasses seven datasets from financial and energy domains. Results indicate its superiority over the vanilla Transformer, showcasing a minimum relative improvement of 18.74% across all datasets. Additionally, an ablation study underscores the effectiveness of the frequency-attention model compared to self-attention.

  • Open Access

    Intelligent Traffic Identification System Powered byConvolutional Neural Networks

    by Anil Kumar, Raju Ranjan

    In model cities, managing traffic flow is paramount, particularly during peak hours and festive periods. Accurate and real-time traffic information is crucial for effective urban planning and transportation management. In response to this challenge, our research introduces a novel solution based on Convolutional Neural Networks (CNNs) to identify traffic flow patterns in urban environments using 2D traffic images as the primary dataset. This dataset comprises two categories: "High Traffic" and "Normal Traffic," which are strategically partitioned into a 70:30 training and testing ratio to facilitate the training of our proposed CNN models. The performance of various CNN models is assessed through a rigorous evaluation process, and the best-performing model is identified based on evaluation metrics, including accuracy, precision, recall, F1-Score, and ROC values computed from confusion matrices. Our results showcase the feasibility of CNNs in addressing the traffic flow identification problem, providing valuable insights for urban planners, traffic engineers, and local authorities. The proposed Convolutional Neural network with three layers using the cropped dataset of images (CNN3*) based solution has the potential to significantly enhance traffic management efficiency and alleviate congestion in model cities, ultimately contributing to improved urban liveability and mobility. This research underscores the importance of leveraging deep learning techniques to tackle complex urban challenges and paves the way for future advancements in traffic flow analysis and smart city development.