Learning with continuous experts using drifting games

We consider the problem of learning to predict as well as the best in a group of experts making continuous predictions. We assume the learning algorithm has prior knowledge of the maximum number of mistakes of the best expert. We propose a new master strategy that achieves the best known performance for on-line learning with continuous experts in the mistake bounded model. Our ideas are based on drifting games, a generalization of boosting and on-line learning algorithms. We prove new lower bounds based on the drifting games framework which, though not as tight as previous bounds, have simpler proofs and do not require an enormous number of experts. We also extend previous lower bounds to show that our upper bounds are exactly tight for sufficiently many experts. A surprising consequence of our work is that continuous experts are only as powerful as experts making binary or no prediction in each round.     

Proactive thermal management in green datacenters

The increasing demand for faster computing and high storage capacity has resulted in an increase in energy consumption and heat generation in datacenters. Because of the increase in heat generation, cooling requirements have become a critical concern, both in terms of growing operating costs as well as their environmental and societal impacts. Presently, thermal management techniques make an effort to thermally profile and control datacenters’ cooling equipment to increase their efficiency. In conventional thermal management techniques, cooling systems are triggered by the temperature crossing predefined thresholds. Such reactive approaches result in delayed response as the temperature may already be too high, which can result in performance degradation of hardware.     

Rolling element bearing design through genetic algorithms

The design of rolling element bearings has been a challenging task in the field of mechanical engineering. While most of the real aspects of the design are never disclosed by bearing manufacturers, the common engineer is left with no other alternative than to refer to standard tables and charts containing the bearing performance characteristics. This paper presents a more viable method to solve this problem using genetic algorithms (GAs). Since the algorithm is basically a guided random search, it weakens the chances of getting trapped in local maxima or minima. The method used has yielded improved performance parameters than those catalogued in standard tables.     

Experimental study of wheel wear in electrolytic in-process dressing and grinding

Profile accuracy of components ground with ultra-precision machine tools is primarily dependent on wheel wear. Quantitative analysis of wheel wear is therefore an important aspect for precision grinding with electrolytic in-process dressing (ELID). In this paper, wheel wear is measured from ELID grinding experiments with different dressing and machining parameters. The grinding forces and dressing current characteristics of the experiments are also compared. Based on the results, a benchmark function is defined for wheel wear rate. A relation for identifying insufficient dressing from sufficient dressing for particular machining conditions is also identified. It is found that insufficient dressing produces pitting and/or arcing on the wheel surface, and wheel wear can be linearly correlated to ELID grinding conditions when the wheels are sufficiently dressed.     

Efficient dressing of the wheel in ELID grinding by controllable voltage with force feed back

Electrolytic in-process dressing (ELID) grinding is a new method for achieving ultraprecision surface on hard and brittle materials. In ELID grinding, pulsed direct current voltage is applied on the metal-bonded diamond wheels to ensure constant protrusion of sharp cutting grits throughout the grinding cycle. The peak dressing voltage is kept constant irrespective of the wheel sharpness in conventional ELID grinding, which may lead to overdressing of the grinding wheel. Grinding force ratio (which is also known as K value) is an indicator for the grit sharpness. In this paper, a new approach of wheel dressing has been proposed where the peak dressing voltage is varied according to the change in the K value during grinding. A Kistler three-component dynamometer has been used to monitor the grinding force ratio in real time for this purpose. The methodology to implement the new “dressing-on-demand” concept has been discussed thoroughly in this paper. The experimental comparison with the conventional ELID grinding has also been carried out to prove the advantages of the proposed system.     

A comprehensive panel of turn-on caspase biosensors for investigating caspase specificity and caspase activation pathways

Caspases play a central role in apoptosis, differentiation, and proliferation, and represent important therapeutic targets for treating cancer and inflammatory disorders. Toward the goal of developing new tools to probe caspase substrate cleavage specificity as well as to systematically interrogate caspase activation pathways, we have constructed and investigated a comprehensive panel of caspase biosensors with a split-luciferase enabled bioluminescent read out. We first interrogated the panel of caspase biosensors for substrate cleavage specificity of caspase 1-10 in widely utilized in vitro translation systems, namely, rabbit reticulocyte lysate (RRL) and wheat germ extract (WGE). Commercial RRL was found to be unsuitable for investigating caspase specificity, owing to surprising levels of endogenous caspase activity, while specificity profiles of the caspase sensors in WGE agree very well with traditional peptide probes. The full panel of biosensors was utilized for studying caspase activation and inhibition in several mammalian cytosolic extracts, clearly demonstrating that they can be utilized to directly monitor activation or inhibition of procaspase 3/7. Furthermore, the complete panel of caspase biosensors also provided new insights into caspase activation pathways wherein we surprisingly discovered the activation of procaspase 3/7 by caspase 4/5.     

Low frequency oscillations in power systems: A physical account and adaptive stabilizers

This paper presents a physical explanation of the phenomenon of low frequency oscillations experienced in power systems. A brief account of the present practice of providing fixed gain power system stabilizers (PSS) is followed by a summary of some of the recent design proposals for adaptive PSS. A novel PSS based on the effort of cancelling the negative damping torque produced by the automatic voltage regulator (AVR) is presented along with some recent studies on a multimachine system using a frequency identification technique.     

An all-pair quantum SVM approach for big data multiclass classification

In this paper, we discuss a quantum approach for the all-pair multiclass classification problem. In an all-pair approach, there is one binary classification problem for each pair of classes, and so there are k(k???1)/2 classifiers for a k-class classification problem. As compared to the classical multiclass support vector machine that can be implemented with polynomial run time complexity, our approach exhibits exponential speedup due to quantum computing. The quantum all-pair algorithm can also be used with other classification algorithms, and a speedup gain can be achieved as compared to their classical counterparts.