Mahendran Velauthapillai

Scheduling Broadcasts in Wireless Networks

We consider problems involving how to schedule broadcasts in a pulled-based data-dissemination service, such as the DirecPC system, where data requested by the clients is delivered via broadcast. In particular, we consider the case where all the data items are of approximately equal in size and preemption is not allowed. We give an offline O(1)-speed O(1)-approximation algorithm for the problem of minimizing the average response time. We provide worst-case analysis, under various objective functions, of the online algorithms that have appeared in the literature, namely, Most Requests First, First Come First Served, and Longest Wait First.

Breaking the probability ½ barrier in FIN-type learning

We show that for every probabilistic FIN-type learner with success ratio greater than 24/49, there is another probabilistic FIN-type learner with success ratio 1/2 that simulates the former. We will also show that this simulation result is tight. We obtain as a consequence of this work a characterization of FIN-type team learning with success ratio between 24/49 and 1/2. We conjecture that the learning capabilities of probabilistic FIN-type learners for probabilities beginning at probability 1/2 are delimited by the sequence 8n/17n-2 for n > 2, which has an accumulation point at 8/17.

On-Demand Broadcasting Under Deadline

In broadcast scheduling multiple users requesting the same information can be satisfied with one single broadcast. In this paper we study preemptive on-demand broadcast scheduling with deadlines on a single broadcast channel. We will show that the upper bound results in traditional real-time scheduling does not hold under broadcast scheduling model. We present two easy to implement online algorithms BCast and its variant BCast2. Under the assumption the requests are approximately of equal length (say k), we show that BCast is O(k) competitive. We establish that this bound is tight by showing that every online algorithm is Ω(k) competitive even if all requests are of same length k. We then consider the case where the laxity of each request is proportional to its length. We show that BCast is constant competitive if all requests are approximately of equal length. We then establish that BCast2 is constant competitive for requests with arbitrary length. We also believe that a combinatorial lemma that we use to derive the bounds can be useful in other scheduling system where the deadlines are often changing (or advanced).

Trade-off among parameters affecting inductive inference

Abstract This paper is concerned with the algorithmic learning, by example in the limit, of programs that compute recursive functions. The particular focus is on the relationship of three parameters that effect inferribility: the number of experimental trials, the plurality of approaches to the particular learning problem, and the accuracy of the final result. Each of these parameters has been examined extensively before. However, the precise characterization of the three-way interaction between these parameters is still not known. This paper makes significant progress toward a complete solution.

Transcriptional profiling of Francisella tularensis infected peripheral blood mononuclear cells: a predictive tool for tularemia

In this study, we analyzed temporal gene expression patterns in human peripheral blood mononuclear cells (PBMCs) infected with the Francisella tularensis live vaccine strain from 1 to 24 h utilizing a whole human Affymetrix gene chip. We found that a considerable number of induced genes had similar expression patterns and functions as reported previously for gene expression profiling in patients with ulceroglandular tularemia. Among the six uniquely regulated genes reported for tularemia patients as being part of the alarm signal gene cluster, five, namely caspase 1, PSME2, TAP-1, GBP1, and GCH1, were induced in vitro. We also detected four out of the seven potential biomarkers reported in tularemia patients, namely TNFAIP6 at 4 h and STAT1, TNFSF10, and SECTM1 at 16 and 24 h. These observations underscore the value of using microarray expression profiling as an in vitro tool to identify potential biomarkers for human infection and disease. Our results indicate the potential involvement of several host pathways/processes in Francisella infection, notably those involved in calcium, zinc ion binding, PPAR signaling, and lipid metabolism, which further refines the current knowledge of F. tularensis infection and its effects on the human host. Ultimately, this study provides support for utilizing in vitro microarray gene expression profiling in human PBMCs to identify biomarkers of infection and predict in vivo immune responses to infectious agents.

Relations between probabilistic and team one-shot learners (extended abstract)

A typical way to increase the power of a learning paradigm is to allow randomization and require successful learning only with some probability p. Another standard approach is to allow a team of s learners working in parallel and to demand only that at least r of them correctly learn. These two variants are compared for the model of learning of total recursive functions where the learning algorithm is allowed an unbounded but finite amount of computation, and must halt with a correct program after receiving only a finite number of values of the function to be learned.

Inductive inference with bounded number of mind changes

Inductive inference machines (IIMs) synthesize programs, given their intended input-output behavior. The program synthesis is viewed as a potentially infinite process of learning by example. Smith [20] studied team learning and obtained results that characterized trade-offs between the number of machines and resources in the learning process. Pitt [16] defined probabilistic learning and showed that ‘probabilistic learning’ is the same as ‘team learning’. Later [17] introduced probabilistic team learning and compared probabilistic team learning and team learning. However, for any given team when we restrict amount of resources allotted for each IIM, then most of the above results fail to hold. This paper studies the relationships between team learning, probabilistic learning and probabilistic team learning when limited resources are available. Some preliminary results obtained indicates a very interesting relationship between them. The proofs for some of the preliminary results, used n-ary recursion theorems, and some complex diagonalization.

Classification using information

AbstractLet

Asking questions versus verifiability

\mathcal{A}