Browsing by Author "Wijesinghe, R"

Filter results by typing the first few letters
Now showing 1 - 5 of 5
  • Thumbnail Image
    item: Article-Full-text
    End-to-end data-dependent routing in multi-path neural networks
    (Springer, 2025) Tissera, D; Wijesinghe, R; Vithanage, K; Xavier, A; Fernando, S; Rodrigo, R
    Neural networks are known to give better performance with increased depth due to their ability to learn more abstract features. Although the deepening of networks has been well established, there is still room for efficient feature extraction within a layer, which would reduce the need for mere parameter increment. The conventional widening of networks by having more filters in each layer introduces a quadratic increment of parameters. Having multiple parallel convolutional/dense operations in each layer solves this problem, but without any context-dependent allocation of input among these operations: the parallel computations tend to learn similar features making the widening process less effective. Therefore, we propose the use of multipath neural networks with data-dependent resource allocation from parallel computations within layers, which also lets an input be routed end-to-end through these parallel paths. To do this, we first introduce a crossprediction based algorithm between parallel tensors of subsequent layers. Second, we further reduce the routing overhead by introducing feature-dependent cross-connections between parallel tensors of successive layers. Using image recognition tasks, we show that our multi-path networks show superior performance to existing widening and adaptive feature extraction, even ensembles, and deeper networks at similar complexity.
  • Thumbnail Image
    item: Conference-Full-text
    Modification of Thermal Conductivity of Cotton Fabric Using Graphene
    (2015-08-03) Gunasekera, U; Perera, N; Perera, S; Hareendra, Y; Somaweera, L; De Silva, N; Tissera, N; Wijesinghe, R
    This paper investigates the possibility to use Graphene coated fabric for thermal conductive purposes. Existing methods of thermal conductivity improvement of textiles require extrusion facilities and the end products are heavy due to insertion of metal wires. By using Graphene Oxide application, these disadvantages can be eliminated. Graphene Oxide was applied onto the fabric using a simple dip and dry method and an exhaustion dyeing method. Then the coated fabric was allowed to react with an aqueous solution of reducing agents. The test results showed that the most suitable method to obtain a significant improvement in thermal conductivity of textiles is to follow an exhaustion dyeing method with reducing agents. The reduction process should be maintained at 700C for 30min duration with the addition of Sodium Hydrosulphide. Exhaust dyed fabric which was treated with reducing agents showed an improvement of 55% in thermal conductivity compared to the uncoated fabric, which is a significant improvement in the thermal conductivity of textiles. In addition to the thermal conductivity values, few related textile properties were tested to evaluate the effect of the Graphene coating on the fabric characteristics.
  • Thumbnail Image
    item: Conference-Abstract
    Modification of thermal conductivity of cotton fabric using graphene
    Gunasekera, U; Perera, N; Perera, S; Hareendra, Y; Somaweera, L; De Silva, N; Tissera, N; Wijesinghe, R
    This paper investigates the possibility to use Graphene coated fabric for thermal conductive purposes. Existing methods of thermal conductivity improvement of textiles require extrusion facilities and the end products are heavy due to insertion of metal wires. By using Graphene Oxide application, these disadvantages can be eliminated. Graphene Oxide was applied onto the fabric using a simple dip and dry method and an exhaustion dyeing method. Then the coated fabric was allowed to react with an aqueous solution of reducing agents. The test results showed that the most suitable method to obtain a significant improvement in thermal conductivity of textiles is to follow an exhaustion dyeing method with reducing agents. The reduction process should be maintained at 700C for 30min duration with the addition of Sodium Hydrosulphide. Exhaust dyed fabric which was treated with reducing agents showed an improvement of 55% in thermal conductivity compared to the uncoated fabric, which is a significant improvement in the thermal conductivity of textiles. In addition to the thermal conductivity values, few related textile properties were tested to evaluate the effect of the Graphene coating on the fabric characteristics.
  • Thumbnail Image
    item: Article-Full-text
    Neural mixture models with expectation-maximization for end-to-end deep clustering
    (Elsevier, 2022) Tissera, D; Vithanage, K; Wijesinghe, R; Xavier, A; Jayasena, S; Fernando, S; Rodrigo, R
    Any clustering algorithm must synchronously learn to model the clusters and allocate data to those clusters in the absence of labels. Mixture model-based methods model clusters with pre-defined statistical distributions and allocate data to those clusters based on the cluster likelihoods. They iteratively refine those distribution parameters and member assignments following the Expectation-Maximization (EM) algorithm. However, the cluster representability of such hand-designed distributions that employ a limited amount of parameters is not adequate for most real-world clustering tasks. In this paper, we realize mixture model-based clustering with a neural network where the final layer neurons, with the aid of an additional transformation, approximate cluster distribution outputs. The network parameters pose as the parameters of those distributions. The result is an elegant, much-generalized representation of clusters than a restricted mixture of hand-designed distributions. We train the network end-to-end via batch-wise EM iterations where the forward pass acts as the E-step and the backward pass acts as the M-step. In image clustering, the mixture-based EM objective can be used as the clustering objective along with existing representation learning methods. In particular, we show that when mixture-EM optimization is fused with consistency optimization, it improves the sole consistency optimization performance in clustering. Our trained networks outperform single-stage deep clustering methods that still depend on k-means, with unsupervised classification accuracy of 63.8% in STL10, 58% in CIFAR10, 25.9% in CIFAR100, and 98.9% in MNIST.
  • Thumbnail Image
    item: Conference-Full-text
    Real-time upper body motion tracking using computer vision for improved human-robot interaction and teleoperation
    (IEEE, 2023-12-09) Nandasena, NASN; Vimukthi, WAA; Herath, HMKKMB; Wijesinghe, R; Yasakethu, SLP; Abeysooriya, R; Adikariwattage, V; Hemachandra, K
    Upper body motion tracking mapping is crucial for robot control because it gives the machine a better understanding of how a human operator moves, allowing it to react instinctively and naturally. Most current research has focused on using wearable sensors and remote controls to enhance communication between robots and humans. However, this research aims to address the issue by embracing a nonwearable sensor-based strategy to promote more natural and spontaneous interactions between humans and robots. Moreover, A 3-DoF manipulator was also designed and developed utilizing robotics technologies. The vision system captured a human operator's upper body movements in realtime video footage. Computer vision approaches were used to extract positional and orientation information from the upper body in this setting. The system combines the MediaPipe pose model with kinematics theories to estimate the hands' position and movement in real-time. According to the experiment results, the system's overall accuracy is 94.1 (±1.2) %, and the motion tracking system's accuracy is 96.5 (±2.0) %.