Hi

I feel that there is still scope in implementing metaheuristic algorithms for Bigdata. There are some interesting extension with respect to Bigdata clustering and search.

Dear Sabarinath Pandurengan,

There are huge scope for it....every year hundreds of research articles are coming on metaheuristic optimization.

Kindly have a look on the link given below:

https://www.researchgate.net/publication/280244004_A_comparative_study_of_metaheuristic_optimization_approaches_for_directional_overcurrent_relays_coordination 

Article A comparative study of metaheuristic optimization approaches...

Unfortunately they all are variations of old methods that have not been very well received, as they do not actually guarantee to find even a near-optimal solution.

I would think that the future of metaheuristics is going to be rather sad, as they are mostly based on non-mathematical methodologies: they are not guided by optimality principles, and therefore they will fail, and they will - because of it - become extinct within a few years.

I disagree with the arguments advanced by Michael Patriksson. You should separate the good grain from the bad in the literature on heuristics, but should not go to trivial conclusions and disregard the whole field. For sure, most recent "nature-inspired" methods rely on artistic/fancy terminologies and boil down to a few core optimization concepts which are already known. This proliferation of methods seriously discredits the field, and simply exists because of our current evaluation system for research, which favors the quantity of papers over the content, and the publishers who see an opportunitie to capture a bigger market. I recommend to read this position paper first:

Article Metaheuristics -- the metaphor exposed

This being said, metaheuristics remain essential for solving most complex and large engineering applications nowadays. In the machine learning field, to make simple examples, K-means, CART, Lasso are all simple local search of construction methods, which have been used thousands of times. Heuristics are ubiquitous and even found within so-called "exact" MIP solvers, where they guide branching decisions, separate cuts or produce good primal solutions. For NP-hard problems, heuristics produce solutions on non-controlled quality in a controlled time, whereas exact methods based on mathematical programming give solutions of controlled quality in a non-controlled time (usually far too large for large application cases). Both have their strengths and weaknesses.

You can also find stunningly good papers focused on the computational complexity of local searches and metaheuristics, e.g., considering how to systematically search a wider set of solutions (possibly exponential) with the smallest possible complexity. Many fascinating research questions are open:

Johnson, D. S., Papadimitriou, C. H., & Yannakakis, M. (1988). How easy is local search? Journal of Computer and System Sciences, 37(1), 79–100.

Chandra, B., Karloff, H., & Tovey, C. (1999). New results on the old k-opt algorithm for the traveling salesman problem. SIAM Journal on Computing (Vol. 28).

Deineko, V. G. V. L., & Woeginger, G. J. G. J. (2000). A study of exponential neighborhoods for the travelling salesman problem and for the quadratic assignment problem. Mathematical Programming, 87(3), 519–542.

Englert, M., Röglin, H., & Vöcking, B. (2014). Worst case and probabilistic analysis of the 2-opt algorithm for the TSP. Algorithmica, 68(1), 190–264.

De Berg, M., Buchin, K., Jansen, B. M. P., & Woeginger, G. (2018). Fine-grained complexity analysis of two classic TSP variants.

Have a good reading !