Machine Learning is the study of methods for programming computers to learn. Computers are applied to a wide range of tasks, and for most of these it is relatively easy for programmers to design and implement the necessary software. However, there are many tasks for which this is difficult or impossible. These can be divided into four general categories. First, there are problems for which there exist no human experts. For example, in modern automated manufacturing facilities, there is a need to predict machine failures before they occur by analyzing sensor readings. Because the machines are new, there are no human experts who can be interviewed by a programmer to provide the knowledge necessary to build a computer system. A machine learning system can study recorded data and subsequent machine failures and learn prediction rules. Second, there are problems where human experts exist, but where they are unable to explain their expertise. This is the case in many perceptual tasks, such as speech recognition, hand-writing recognition, and natural language understanding. Virtually all humans exhibit expert-level abilities on these tasks, but none of them can describe the detailed steps that they follow as they perform them. Fortunately, humans can provide machines with examples of the inputs and correct outputs for these tasks, so machine learning algorithms can learn to map the inputs to the outputs. Third, there are problems where phenomena are changing rapidly. In finance, for example, people would like to predict the future behavior of the stock market, of consumer purchases, or of exchange rates. These behaviors change frequently, so that even if a programmer could construct a good predictive computer program, it would need to be rewritten frequently. A learning program can relieve the programmer of this burden by constantly modifying and tuning a set of learned prediction rules. Fourth, there are applications that need to be customized for each computer user separately. Consider, for example, a program to filter unwanted electronic mail messages. Different users will need different filters. It is unreasonable to expect each user to program his or her own rules, and it is infeasible to provide every user with a software engineer to keep the rules up-to-date. A machine learning system can learn which mail messages the user rejects and maintain the filtering rules automatically. Machine learning addresses many of the same research questions as the fields of statistics, data mining, and psychology, but with differences of emphasis. Statistics focuses on understanding the phenomena that have generated the data, often with the goal of testing different hypotheses about those phenomena. Data mining seeks to find patterns in the data that are understandable by people. Psychological studies of human learning aspire to understand the mechanisms underlying the various learning behaviors exhibited by people (concept learning, skill acquisition, strategy change, etc.).

Machine learning

We will review some of the major results in random graphs and some of the more challenging open problems. We will cover algorithmic and structural questions. We will touch on newer models, including those related to the WWW.

Random graphs

Many practical computing problems concern large graphs. Standard examples include the Web graph and various social networks. The scale of these graphs - in some cases billions of vertices, trillions of edges - poses challenges to their efficient processing. In this paper we present a computational model suitable for this task. Programs are expressed as a sequence of iterations, in each of which a vertex can receive messages sent in the previous iteration, send messages to other vertices, and modify its own state and that of its outgoing edges or mutate graph topology. This vertex-centric approach is flexible enough to express a broad set of algorithms. The model has been designed for efficient, scalable and fault-tolerant implementation on clusters of thousands of commodity computers, and its implied synchronicity makes reasoning about programs easier. Distribution-related details are hidden behind an abstract API. The result is a framework for processing large graphs that is expressive and easy to program.

/pdf/pregel-a-system-for-large-scale-graph-processing-30lgjenxxg.pdf

Pregel: a system for large-scale graph processing

where A represents the magnetic vector potential, is an integral of the hydromagnetic equations. This -integral made it possible to formulate a variational principle for the force-free magnetic fields. The integral expresses the fact that motions cannot transform a given field in an entirely arbitrary different field, if the conductivity of the medium isconsidered infinite. In this paper we shall show that the full set of hydromagnetic equations admit five more integrals, besides the energy integral, if dissipative processes are absent. These integrals, as we shall presently verify, are I2 =fbHvdV, (2)

Proceedings of the National Academy of Sciences

Computational geometry

Near-optimal gossiping algorithms are given for two- and higher dimensional tori. It is assumed that the amount of data each PU is contributing is so large, that start-up time may be neglected. For two-dimensional tori, an earlier algorithm achieved optimality in an intricate way, with a time-dependent routing pattern. In our algorithms, in all steps, the PUs forward the received packets in the same way.

/pdf/gossiping-large-packets-on-full-port-tori-1olvj90ce2.pdf

Gossiping Large Packets on Full-Port Tori

In the past a number of I/O-efficient algorithms were designed to solve a problem on a static data set. However, many data sets like social networks or web graphs change their shape frequently. We provide experimental results of the first external-memory dynamic breadth-first search (BFS) implementation based on earlier theoretical work [13] that crucially relies on a randomized clustering. We refine this approach using a new I/O-efficient deterministic clustering, which groups vertices in a level-aligned hierarchy and facilitates easy access to clusters of changing sizes during the BFS updates. In most cases the new external memory dynamic BFS implementation is significantly faster than recomputing the BFS levels after an edge insertion from scratch.

An Implementation of I/O-Efficient Dynamic Breadth-First Search Using Level-Aligned Hierarchical Clustering

We initiate a study of algorithms with a focus on the computational complexity of individual elements, and introduce the fragile complexity of comparison-based algorithms as the maximal number of comparisons any individual element takes part in. We give a number of upper and lower bounds on the fragile complexity for fundamental problems, including Minimum, Selection, Sorting and Heap Construction. The results include both deterministic and randomized upper and lower bounds, and demonstrate a separation between the two settings for a number of problems. The depth of a comparator network is a straight-forward upper bound on the worst case fragile complexity of the corresponding fragile algorithm. We prove that fragile complexity is a different and strictly easier property than the depth of comparator networks, in the sense that for some problems a fragile complexity equal to the best network depth can be achieved with less total work and that with randomization, even a lower fragile complexity is possible.

/pdf/fragile-complexity-of-comparison-based-algorithms-1xs0vg62bi.pdf

Fragile complexity of comparison-based algorithms

We propose a variation of online paging in two-level memory systems where pages in the fast cache get modified and therefore have to be explicitly written back to the slow memory upon evictions. For increased performance, up to ? arbitrary pages can be moved from the cache to the slow memory within a single joint eviction, whereas fetching pages from the slow memory is still performed on a one-by-one basis. The main objective in this new ?-paging scenario is to bound the number of evictions. After providing experimental evidence that ?-paging can improve the performance of flash-memory devices in the context of translation layers we turn to the theoretical connections between ?-paging and standard paging. We give lower bounds for deterministic and randomized ?-paging algorithms. For deterministic algorithms, we show that an adaptation of LRU is strongly competitive, while for the randomized case we show that by adapting the classical Mark algorithm we get an algorithm with a competitive ratio larger than the lower bound by a multiplicative factor of approximately 1.7.

/pdf/online-paging-for-flash-memory-devices-3mklsxcju6.pdf

Online Paging for Flash Memory Devices

We present a simple algorithm which maintains the topological order of a directed acyclic graph with n nodes under an online edge insertion sequence in O(n^{2.75}) time, independent of the number of edges m inserted. For dense DAGs, this is an improvement over the previous best result of O(min(m^{3/2} log(n), m^{3/2} + n^2 log(n)) by Katriel and Bodlaender. We also provide an empirical comparison of our algorithm with other algorithms for online topological sorting. Our implementation outperforms them on certain hard instances while it is still competitive on random edge insertion sequences leading to complete DAGs.

Ulrich Meyer

Papers

Gossiping Large Packets on Full-Port Tori

An Implementation of I/O-Efficient Dynamic Breadth-First Search Using Level-Aligned Hierarchical Clustering

Fragile complexity of comparison-based algorithms

Online Paging for Flash Memory Devices

An O(n^{2.75}) algorithm for online topological ordering