M Tech Dissertations

Permanent URI for this collectionhttp://drsr.daiict.ac.in/handle/123456789/3

Browse

Author: search.filters.author.Soni, Maulik

Search Results

Now showing 1 - 3 of 3
  • Thumbnail Image
    ItemOpen Access
    Algorithmic Trading using Machine Learning
    (Dhirubhai Ambani Institute of Information and Communication Technology, 2021) Soni, Maulik; Majumder, Prasenjit
    This works aims to understand the different factors which influence the stock market and further use those factors to forecast its future values. The Nifty50 and NiftyBANK under National Stock Exchange India(NSE India) are considered the target indexes in this research. The parameters which affects the Stock Market are diverse in nature which mainly categorise in two types: Trend in historical stock prices and events happening around the world available in unstructured format. The later part generally involves like social blogs, news etc, which can change the perspective of investors towards the Stock Market. Additionally, we have explore two different entities, Gold and Rupee-Dollar exchange, and tried to Anderstand how its impact the stock market. In addition to this, the Retrospective Analysis between the Stock Market and Nifty50 carried out which inspect the particular events responsible for changes happened in the Stock movement. As Machine Learning proves very successful in solving complex problems in past, we also have leveraged the advantage of popular algorithms/models to forecast the market.
  • Thumbnail Image
    ItemOpen Access
    Acyclic edge colouring of subgraphs of hypercubes
    (Dhirubhai Ambani Institute of Information and Communication Technology, 2020) Soni, Maulik; Muthu, Rahul
    Acyclic edge colouring is a very important combinatorial optimization problem, which many researchers are currently working on. Acyclic edge colouring is assignment of colours to vertices or edges of the graph such that, no two colour induced subgraph contains any cycle. In its most general form, it is a very difficult problem and thus research work on this problem, usually focuses on some limited aspect. In particular one approach to this problem is to find exact values for the acyclic chromatic index for special families of graphs. One might also look for corresponding efficient algorithms, for such optimal colourings. However, it is essential to study a few papers on this problem of acyclic edge colouring in order to develop one’s techniques and solve a new aspect of this problem. We aim to obtain optimal bounds and algorithms for the acyclic edge colouring problem on the class of subgraphs of hypercubes. We have given the best colour bound of D + 1 (where D is maximum degree of graph), for three categories of subgraphs of hypercubes. First is arbitrary subgraph of any dimensional hypercube, second is (d �� 1)-regular subgraphs of d-dimensional hypercube with constraints of limited 2 and 4 cross edges. and third is 3-regular subgraphs of any dimensional hypercube. We have given the some approaches proven effective from mathematical or simulation point of view. All the colour bounds given by us, are supported by well known, Acyclic Edge Colouring Conjecture. The approach of colouring 3-regular subgraphs of any dimensional hypercube, is not just limited to subgraphs of the hypercube, but it can also be considered for any regular bipartite graph with regularity equals to 3.
  • Thumbnail Image
    ItemOpen Access
    Distributed caching mechanism for video on demand on the Internet
    (Dhirubhai Ambani Institute of Information and Communication Technology, 2006) Soni, Maulik; Srivastava, Sanjay; Lenin, R. B.
    Video on Demand applications in the Internet are delay sensitive and highly resource intensive applications. Such applications require streaming of large video files in a time bound manner. Because large volume in terms of data and high transmission rate video streaming requires high bandwidth and storage space than data transfer applications. Caching an ongoing stream at an intermediate node between the content server, which provides video on demand service, and the client(s) is a viable solution to save resources. The proposed scheme is such a caching scheme, which caches video content among the intermediate nodes in a distributed manner. The streams are cached among the dynamically selected caching nodes and the clients are served the video content from these nodes. The caching nodes contact each other and collaboratively act together to provide the service. The scheme also proposes a heuristic to estimate resource requirements for the video on demand systems. The behavior of the scheme is demonstrated through simulation. The quality of service using the proposed scheme is compared with some of the existing schemes.