Articles

Tutorial on Distributed Strategic Learning For Engineers

 
Materials:  
Audience: Engineers, Masters, Ph.D. Students, Researchers 
 
This tutorial course will revisit the fundamental tools  of distributed strategic learning in view of their applications to wireless networks. The course will be subdivided into a theoretical part where the classical methods and results for distributed learning are introduced, and an application part where practical considerations in engineering are visited. Each part will decline successively the basic tools and applications, the advanced methods and results known today, as well as current research activities. 
 
 
 
A precise outline is given below:
 
puceblackPart I: THEORETICAL NOTIONS
  • puceblackMarkov trees
  • puceblackDynamical systems
  • puceblackStochastic approximation
  • puceblackDifferential inclusion
  • puceblackBasics of matrix games
  • puceblackBasics of random matrix games (RMGs)
  • puceblackBasics of robust games
  • puceblackBasics of dynamic robust games
  • puceblackDynamic solution concepts/stationary solution concepts
  • puceblackResearch today: 
    • puceblack fast convergence of iterative learning patterns,
    •  puceblack hitting time to a set, 
    •  puceblack frequency of visits
  • puceblackIntroduction to strategy learning 
puceblack(STRATEGY LEARNING and PERFORMANCE ESTIMATIONS)
  • puceblackStrategy-learning schemes [for equilibrium, Pareto point, stackelberg solution, satisfactory point, coalition selection]
  • puceblackPayoff-learning,
  • puceblackCombined learning (CODIPAS)
  • puceblackHeterogeneous learning
  • puceblackHybrid learning
  • puceblackLearning with random updates
  • puceblackRisk-sensitive strategic learning in dynamic robust games
  • puceblackCombined learning for continuous action space
  • puceblackMean field learning
  • puceblack Learning in random matrix games (RMGs). CODIPAS for mean-variance learning
  • puceblackResearch today:
    • puceblack Best learning algorithms for stability/performance tradeoff,
    • puceblack How to learn global optima in a fully distributed way?
puceblackPart II: APPLICATIONS TO WIRELESS NETWORKS, COMMUNICATIONS AND NETWORK ECONOMICS
 
puceblack (WIRELESS NETWORKS and COMMUNICATIONS)
  • puceblackDistributed learning for parallel routing and frequency selection
  • puceblackStrategic Learning in user-centric network selection
  • puceblackCombined learning under noise in WLAN
  • puceblackCost of learning and Quality of Experience (QoE) in LTE
  • puceblackDistributed Learning for Network Security
  • puceblackLearning under uncertainty for Network MIMO
  • puceblackCoalitional learning for cognitive radios
  • puceblackResearch today: 
    •  puceblack Why should we care on distributed strategic learning?
    •  puceblack How to extract useful information from outdated and noisy measurements?
    • puceblack How to extract useful information from big data?
 
puceblack (ECONOMICS OF NETWORKS)
  • puceblackMean field learning for the smart grid
  • puceblackHierarchical learning for  network design
  • puceblackRisk-sensitive learning for the economics of cloud computing
  • puceblackResearch today: 
    • puceblack Simultaneous, coalitional and mutual learning 
    • puceblack Learn how to flatten the peaks in large-scale networks