Majid Jahani

Research

My research mostly focuses on Large scale convex and non-convex optimization, and design and analysis of algorithms for machine learning. Prior coming to Lehigh, I was working on Network Optimization and its applications in public transportation.

Here is a list of my working papers and publications

  • Majid Jahani, Naga Venkata C. Gudapati, Chenxin Ma, Rachael Tappenden, Martin Takáč. “Fast and safe: accelerated gradient methods with optimality certificates and underestimate sequences.” Computational Optimization and Applications [PDF]
  • Soheil Sadeghi Eshkevari, Martin Takáč, Shamim N. Pakzad and Majid Jahani. “DynNet: Physics-based neural architecture design for linear and nonlinear structural response modeling and prediction,” [PDF]
  • Majid Jahani, Mohammadreza Nazari, Rachael Tappenden, Albert S. Berahas and Martin Takáč. “SONIA: A Symmetric Blockwise TruncatedOptimization Algorithm,” [PDF]
  • Peter Richtárik, Majid Jahani, Selin Damla Ahipasaoglu and Martin Takáč. “Alternating Maximization: Unifying Framework for 8 Sparse PCA Formulations and Efficient Parallel Codes,” Optimization and Engineering (OPTE), [PDF]
  • Majid Jahani, Xi He, Chenxin Ma, Aryan Mokhtari, Dheevatsa Mudigere, Alejandro Ribeiro and Martin Takáč. “Efficient Distributed Hessian Free Algorithm for Large-scale ERM via Accumulating Sample Strategy,” AISTATS 2020 [PDF]
  • Albert S. Berahas, Majid Jahani and Martin Takáč. “Quasi-Newton Methods for Deep Learning, ” Workshop @ OPT 2019[PDF]
  • Majid Jahani, Xi He, Chenxin Ma, Aryan Mokhtari, Dheevatsa Mudigere, Alejandro Ribeiro and Martin Takáč. “Grow Your Samples and Optimize Better via Distributed Newton CG and Accumulating Strategy,” Beyond First Order Methods in ML Workshop @ NeurIPS 2019 [PDF]
  • Mohammadreza Nazari, Majid Jahani, Lawrence V. Snyder, and Martin Takáč. “Don’t Forget Your Teacher: A Corrective Reinforcement Learning Framework,” arXiv preprint [PDF]
  • Majid Jahani, Mohammadreza Nazari, Sergey Rusakov, Albert S. Berahas and Martin Takáč. “Scaling Up Quasi-Newton Algorithms: Communication Efficient Distributed SR1,” The Sixth International Conference on Machine Learning, Optimization, and Data Science (LOD), 2020 [PDF]
  • Albert S. Berahas, Majid Jahani and Martin Takáč. “Quasi-Newton Methods for Deep Learning: Forget the Past, Just Sample,” arXiv preprint [PDF]
  • Chenxin Ma, Naga Venkata C. Gudapati, Majid Jahani, Rachael Tappenden, Martin Takáč. Underestimate Sequences via Quadratic Averaging. arXiv preprint [PDF]
  • Majid Jahani, S. Mehdi Hashemi, Mehdi Ghatee, Mohsen Jahanshahi. A novel model for bus stop location appropriate for Public Transit Network Design: The case of Central Business Districts (CBD) of Tehran .  International Journal of Smart Electrical Engineering [PDF]
  • Roozbeh Ebrazi , S. Mehdi Hashemi, Majid Jahani. Dynamic Location-Allocation Model of the Traffic Surveillance System. 9th International Conference of Iranian Operations Research Society, Shiraz, Iran, 2016.
  • Majid Jahani, S. Mehdi Hashemi, Roozbeh Ebrazi. A novel model for bus stop location appropriate for Public Transit Network Design. The 5th Iranian conference on Applied Mathematics, Hamedan, Iran, 2013.
  • Majid Jahani, S. Mehdi Hashemi, Roozbeh Ebrazi, A Novel Heuristic Algorithm for Transit Network Design Problem Using Combinatorial Optimization. The 14 th International Conference on Traffic and Transportation Engineering, Tehran, Iran, 2015.

Technical Reports

  • Public Transportation Network Design with respect to Efficient Times
    • Majid Jahani,
    • Master Thesis, Amirkabir University of Technology, Supervisor Prof. S. M. Hashemi