InfoCoBuild

CS 224: Advanced Algorithms

CS 224: Advanced Algorithms (Fall 2014, Harvard Univ.). Instructor: Professor Jelani Nelson. An algorithm is a well-defined procedure for carrying out some computational task. Typically the task is given, and the job of the algorithmist is to find such a procedure which is efficient, for example in terms of processing time and/or memory consumption. CS 224 is an advanced course in algorithm design, and topics we will cover include the word RAM model, data structures, amortization, online algorithms, linear programming, semidefinite programming, approximation algorithms, hashing, randomized algorithms, fast exponential time algorithms, graph algorithms, and computational geometry.

Lecture 24 - Fast Exponential-Time Algorithms: TSP, 3-SAT, K-Colorability


Go to the Course Home or watch other lectures:

Lecture 01 - Course Introduction, Word RAM and van Emde Boas Trees
Lecture 02 - Fusion Trees, Word-Level Parallelism
Lecture 03 - Hashing: Load Balancing, K-wise Independence, Dictionary
Lecture 04 - Hashing: Linear Probing (5-wise Indep.), Bloom Filters, Cuckoo Hashing, Bloomier Filters
Lecture 05 - Hashing: Cuckoo Hashing Analysis, Power of Two Choices
Lecture 06 - Amortized Analysis, Binomial Heaps, Fibonacci Heaps
Lecture 07 - Splay Trees
Lecture 08 - Online Algorithms, Competitive Analysis, Move-to-Front, Paging
Lecture 09 - Randomized Paging, Primal/Dual Online Algorithms
Lecture 10 - Online Primal/Dual: e/(e-1) Ski Rental, Approximation Algorithms via Dual Fitting
Lecture 11 - Approximation Algorithms via Dual Fitting, LP Integrality Gaps, PTAS/FPTAS/FPRAS
Lecture 12 - FPTAS (Knapsack), FPRAS (DNF Counting), Semidefinite Programming
Lecture 13 - Learning Topic Models; Machine Learning Problems
Lecture 14
Lecture 15 - Linear Programming: Standard Form, Vertices, Bases, Simplex
Lecture 16 - Simplex, Strong Duality, Complementary Slackness, Ellipsoid Algorithm
Lecture 17 - Path-Following Interior Point, First Order Methods (Gradient Descent)
Lecture 18 - Second Order Methods (Newton's Method), Path-Following Interior Point
Lecture 19 - Multiplicative Weights, Learning from Experts
Lecture 20 - Applying Multiplicative Weights to Linear Programmings
Lecture 21 - Faster s-t Max Flow: Scaling for Max Flow, Blocking Flows
Lecture 22 - Link-Cut Trees
Lecture 23 - O(log2 n) Amortized Analysis of Link-Cut Trees, Min Cost Max Flow
Lecture 24 - Fast Exponential-Time Algorithms: TSP, 3-SAT, K-Colorability
Lecture 25 - Zeta Transform, Mobius Inversion, Streaming Algorithms
Lecture 26 - Streaming Algorithms, Power of Random Signs