“… The present work intro­duces new heuris­tic seg­men­ta­tion algo­rithms for the fol­low­ing opti­miza­tion prob­lem: Find a seg­men­ta­tion of an approx­i­mated matrix using only allowed fields and min­i­mize the approx­i­ma­tion error. Finally, the decom­po­si­tion algo­rithms were imple­mented into an opti­miza­tion pro­gramme in order to exam­ine the assump­tions of the algo­rithms for a clin­i­cal exam­ple. As a result, iden­ti­cal dose dis­tri­b­u­tions with much fewer seg­ments and a sig­nif­i­cantly smaller num­ber of mon­i­tor units could be achieved using dosi­met­ric con­straints. Con­se­quently, the dose deliv­ery is more effi­cient and less time consuming.”

“Unless we get bet­ter leg­is­la­tors, and a cou­ple of hun­dred years of his­tory says not to count on that, enhanc­ing the auto­matic sta­bi­liz­ers may be our best bet going for­ward. There’s con­sid­er­able empir­i­cal evi­dence show­ing that they work, includ­ing this new evi­dence that auto­matic sta­bi­liz­ers work “always and everywhere”…”

“We report on exper­i­ments of many small motors — cell phone vibra­tors — glued to and inter­act­ing through a res­o­nant plate. We find that indi­vid­ual motors inter­act­ing with the plate demon­strate hys­tere­sis in their steady-​​state fre­quency due to inter­ac­tions with plate res­o­nances. For mul­ti­ple motors run­ning simul­ta­ne­ously, the degree of syn­chro­niza­tion between motors increases when the motors’ fre­quen­cies are near a res­o­nance of the plate, and the fre­quency at which the motors syn­chro­nize shows a his­tory dependence.”

“We describe von Neumann’s ele­gant idea for sam­pling from the expo­nen­tial dis­tri­b­u­tion, Forsythe’s gen­er­al­iza­tion for sam­pling from a prob­a­bil­ity dis­tri­b­u­tion whose den­sity has the form exp(-G(x)), where G(x) is easy to com­pute (e.g. a poly­no­mial), and my refine­ment of these ideas to give an effi­cient algo­rithm for gen­er­at­ing pseudo-​​random num­bers with a nor­mal dis­tri­b­u­tion. Later devel­op­ments are also mentioned.”

“It clearly looks as if the NASDAQ was try­ing to pro­tect the inter­ests of some insti­tu­tions which lost large sums of money dur­ing Thursday’s col­lapse. These insti­tu­tions likely did not have safe­guards in place to deal with a lack of bid­ders. When their auto­mated mar­ket orders hit the mar­ket in the absence of bid­ders, trans­ac­tions were done at ridicu­lously low prices. But as NASDAQ has noted, there was no fail­ure of their systems.”

“So when France and Ger­many make sure that Greece can pay its debt, they are also res­cu­ing, well, France and Ger­many. Also makes it clear exactly how con­ta­gion could work in practice.”

“The aim of this chap­ter is twofold. In the first part we will pro­vide a brief overview of the math­e­mat­i­cal and sta­tis­ti­cal foun­da­tions of graph­i­cal mod­els, along with their fun­da­men­tal prop­er­ties, esti­ma­tion and basic infer­ence pro­ce­dures. In par­tic­u­lar we will develop Markov net­works (also known as Markov ran­dom fields) and Bayesian net­works, which com­prise most past and cur­rent lit­er­a­ture on graph­i­cal mod­els. In the sec­ond part we will review some appli­ca­tions of graph­i­cal mod­els in sys­tems biology.”

“Project Euler is a series of chal­leng­ing mathematical/​computer pro­gram­ming prob­lems that will require more than just math­e­mat­i­cal insights to solve. Although math­e­mat­ics will help you arrive at ele­gant and effi­cient meth­ods, the use of a com­puter and pro­gram­ming skills will be required to solve most problems.

The moti­va­tion for start­ing Project Euler, and its con­tin­u­a­tion, is to pro­vide a plat­form for the inquir­ing mind to delve into unfa­mil­iar areas and learn new con­cepts in a fun and recre­ational context.”

“It is a mantra of the deficit hawks that they are work­ing to ensure their chil­dren and grand­chil­dren will one day have the same oppor­tu­ni­ties that they have had. But right now, in real time, those same chil­dren and grand­chil­dren are hav­ing those oppor­tu­ni­ties taken away. …”

“Build­ing a repos­i­tory of proof-​​checked math­e­mat­i­cal knowl­edge is with­out any doubt a lot of work, and besides the actual for­mal­iza­tion process there also is the task of main­tain­ing the repos­i­tory. Thus it seems obvi­ous to keep a rep­soitory as small as pos­si­ble, in par­tic­u­lar each piece of math­e­mat­i­cal knowl­edge should be for­mal­ized only once. In this paper, how­ever, we claim that it might be rea­son­able or even nec­es­sary to dupli­cate knowl­edge in a math­e­mat­i­cal repos­i­tory. We ana­lyze dif­fer­ent sit­u­a­tions and rea­sons for doing so and pro­vide a num­ber of exam­ples sup­port­ing our thesis.”

“In this work, deci­sion tree learn­ing algo­rithms and fuzzy infer­enc­ing sys­tems are applied for galaxy mor­phol­ogy clas­si­fi­ca­tion. In par­tic­u­lar, the CART, the C4.5, the Ran­dom For­est and fuzzy logic algo­rithms are stud­ied and reli­able clas­si­fiers are devel­oped to dis­tin­guish between spi­ral galax­ies, ellip­ti­cal galax­ies or star/​unknown galac­tic objects. Mor­phol­ogy infor­ma­tion for the train­ing and test­ing datasets is obtained from the Galaxy Zoo project while the cor­re­spond­ing pho­to­met­ric and spec­tra para­me­ters are down­loaded from the SDSS DR7 catalogue.”

“The prop­er­ties of poly­mer liq­uids on hard and soft sub­strates are inves­ti­gated by mol­e­c­u­lar dynam­ics sim­u­la­tion of a coarse-​​grained bead-​​spring model and dynamic single-​​chain-​​in-​​mean-​​field (SCMF) sim­u­la­tions of a soft, coarse-​​grained poly­mer model. Hard, cor­ru­gated sub­strates are mod­elled by an FCC Lennard-​​Jones solid while poly­mer brushes are inves­ti­gated as a pro­to­typ­i­cal exam­ple of a soft, deformable sur­face. From the mol­e­c­u­lar sim­u­la­tion we extract the coarse-​​grained para­me­ters that char­ac­terise the equi­lib­rium and flow prop­er­ties of the liq­uid in con­tact with the sub­strate: the sur­face and inter­face ten­sions, and the para­me­ters of the hydro­dy­namic bound­ary con­di­tion. The so-​​determined para­me­ters enter a con­tin­uum descrip­tion like the Stokes equa­tion or the lubri­ca­tion approximation.”

“… Due to the tremen­dous growth of network-​​based ser­vices, intru­sion detec­tion has emerged as an impor­tant tech­nique for net­work secu­rity. Recently data min­ing algo­rithms are applied on network-​​based traf­fic data and host-​​based pro­gram behav­iors to detect intru­sions or mis­use pat­terns, but there exist some issues in cur­rent intru­sion detec­tion algo­rithms such as unbal­anced detec­tion rates, large num­bers of false pos­i­tives, and redun­dant attrib­utes that will lead to the com­plex­ity of detec­tion model and degra­da­tion of detec­tion accu­racy. The pur­pose of this study is to iden­tify impor­tant input attrib­utes for build­ing an intru­sion detec­tion sys­tem (IDS) that is com­pu­ta­tion­ally effi­cient and effective.…”

“We present an effi­cient solu­tion to the fol­low­ing prob­lem, of rel­e­vance in a numer­i­cal opti­miza­tion scheme: cal­cu­la­tion of inte­grals of the type \[\iint_{T \cap \{f\ge0\}} \phi_1\phi_2 \, dx\,dy\] for qua­dratic poly­no­mi­als $f,\phi_1,\phi_2$ on a plane tri­an­gle $T$. The naive approach would involve con­sid­er­a­tion of the many pos­si­ble shapes of $T\cap\{f\geq0\}$ (pos­si­bly after a con­ve­nient trans­for­ma­tion) and para­me­ter­iz­ing its bor­der, in order to inte­grate the vari­ables sep­a­rately. Our solu­tion involves par­ti­tion­ing the tri­an­gle into smaller tri­an­gles on which inte­gra­tion is much simpler.”

“This paper presents an effi­cient human recog­ni­tion sys­tem based on vein pat­tern from the palma dorsa. A new absorp­tion based tech­nique has been pro­posed to col­lect good qual­ity images with the help of a low cost cam­era and light source. The sys­tem auto­mat­i­cally detects the region of inter­est from the image and does the nec­es­sary pre­pro­cess­ing to extract fea­tures. A Euclid­ean Dis­tance based match­ing tech­nique has been used for mak­ing the deci­sion. It has been tested on a data set of 1750 image sam­ples col­lected from 341 indi­vid­u­als. The accu­racy of the ver­i­fi­ca­tion sys­tem is found to be 99.26% with false rejec­tion rate (FRR) of 0.03%.”

‘It turns out that people’s stated pref­er­ences have a weak rela­tion­ship to who they actu­ally like. Stated pref­er­ences, one study found, “seemed to van­ish when it came time to choose a part­ner in phys­i­cal space.”’

“Every algo­rithm which can be exe­cuted on a com­puter can at least in prin­ci­ple be real­ized in hard­ware, i.e. by a dis­crete phys­i­cal sys­tem. The prob­lem is that up to now there is no pro­gram­ming lan­guage by which phys­i­cal sys­tems can con­struc­tively be described. Such tool, how­ever, is essen­tial for the com­pact descrip­tion and auto­matic pro­duc­tion of com­plex sys­tems. This paper intro­duces a pro­gram­ming lan­guage, called Akton-​​Algebra, which pro­vides the foun­da­tion for the com­plete descrip­tion of dis­crete phys­i­cal systems.…”

“Per­for­mance tun­ing of Data­base Man­age­ment Systems(DBMS) is both com­plex and chal­leng­ing as it involves iden­ti­fy­ing and alter­ing sev­eral key per­for­mance tun­ing para­me­ters. The qual­ity of tun­ing and the extent of per­for­mance enhance­ment achieved greatly depends on the skill and expe­ri­ence of the Data­base Admin­is­tra­tor (DBA). As neural net­works have the abil­ity to adapt to dynam­i­cally chang­ing inputs and also their abil­ity to learn makes them ideal can­di­dates for employ­ing them for tun­ing pur­pose. In this paper, a novel tun­ing algo­rithm based on neural net­work esti­mated tun­ing para­me­ters is pre­sented. The key per­for­mance indi­ca­tors are proac­tively monitored….The tuner alters these tun­ing para­me­ters using the esti­mated val­ues using a rate change com­put­ing algo­rithm. The pre­lim­i­nary results show that the pro­posed method is effec­tive in improv­ing the query response time for a vari­ety of work­load types.”

“In this case, the com­put­ers kicked in their sell pro­grams and there were no buy pro­grams engaged — and so there was no mar­ket — and stocks wound up sell­ing for a penny a share.

Now already known as the “flash crash,” this remark­able event will almost surely put a whole gen­er­a­tion of young math whizzes out of busi­ness as Con­gress and the SEC crawl all over these oper­a­tions and limit this kind of insane action.

I’m all for it because, and you can call me old fash­ioned, I don’t think the global equi­ties mar­kets should be an online gam­bling casino which is what they’ve become with the rise of the “quants” and their hyper­ac­tive supercomputers.”

“… It avoids most of the short­com­ings encoun­tered with the now clas­sic Black-​​Scholes-​​Merton frame­work. Sev­eral con­vinc­ing com­puter sim­u­la­tions are pre­sented. Some of them are deal­ing with abrupt changes, i.e., jumps.”

“Dur­ing the last decade, incre­men­tal sampling-​​based motion plan­ning algo­rithms, such as the Rapidly-​​exploring Ran­dom Trees (RRTs) have been shown to work well in prac­tice and to pos­sess the­o­ret­i­cal guar­an­tees such as prob­a­bilis­tic com­plete­ness. How­ever, no the­o­ret­i­cal bounds on the qual­ity of the solu­tion obtained by these algo­rithms have been estab­lished so far. The first con­tri­bu­tion of this paper is a neg­a­tive result: it is proven that, under mild tech­ni­cal con­di­tions, the cost of the best path in the RRT con­verges almost surely to a non-​​optimal value. Sec­ond, a new algo­rithm is con­sid­ered, called the Rapidly-​​exploring Ran­dom Graph (RRG), and it is shown that the cost of the best path in the RRG con­verges to the opti­mum almost surely. Third, a tree ver­sion of RRG is intro­duced, called the RRT$^*$ algo­rithm, which pre­serves the asymp­totic opti­mal­ity of RRG while main­tain­ing a tree struc­ture like RRT.…”

“Recent research on mul­ti­ple ker­nel learn­ing has lead to a num­ber of approaches for com­bin­ing ker­nels in reg­u­lar­ized risk min­i­miza­tion. The pro­posed approaches include dif­fer­ent for­mu­la­tions of objec­tives and vary­ing reg­u­lar­iza­tion strate­gies. In this paper we present a uni­fy­ing gen­eral opti­miza­tion cri­te­rion for mul­ti­ple ker­nel learn­ing and show how exist­ing for­mu­la­tions are sub­sumed as spe­cial cases. We also derive the criterion’s dual rep­re­sen­ta­tion, which is suit­able for gen­eral smooth opti­miza­tion algo­rithms. Finally, we eval­u­ate mul­ti­ple ker­nel learn­ing in this frame­work ana­lyt­i­cally using a Rademacher com­plex­ity bound on the gen­er­al­iza­tion error and empir­i­cally in a set of experiments.”

“If a fund/​institution/​High Fre­quency Trader gen­er­ates 100mm shares or con­tracts a day/​week/​month, mar­ket observers will tell you that’s a great thing because it cre­ates a liq­uid­ity pre­mium. In other words, because there is always some­one on the other side of a trade, it is eas­ier to match buy­ers and sell­ers and that ease cre­ates smaller spreads and often lower pric­ing. On the sur­face that’s a great thing.

It is a great thing until the mar­ket becomes com­pletely depen­dent on that liq­uid­ity. If every model expects X vol­ume, what hap­pens when that vol­ume falls?”

“One of the most effec­tive tech­niques of exper­i­men­tal math­e­mat­ics is to com­pute math­e­mat­i­cal enti­ties such as inte­grals, series or lim­its to high pre­ci­sion, then attempt to rec­og­nize the result­ing numer­i­cal val­ues. Recently these tech­niques have been applied with great suc­cess to prob­lems in math– emat­i­cal physics. Notable among these appli­ca­tions are the iden­ti­fi­ca­tion of some key multi-​​dimensional inte­grals that arise in Ising the­ory, quan­tum field the­ory and in mag­netic spin theory.”

“…A ver­sion of the method inde­pen­dent of the con­trast of the image is con­sid­ered and is found to be use­ful for find­ing the most unusual part (and the most sim­i­lar part) of the image con­di­tioned on given image. The results can be used to scan large image data­bases, as for exam­ple med­ical databases.…”

“…A ver­sion of the method inde­pen­dent of the con­trast of the image is con­sid­ered and is found to be use­ful for find­ing the most unusual part (and the most sim­i­lar part) of the image con­di­tioned on given image. The results can be used to scan large image data­bases, as for exam­ple med­ical databases.”

“ECG Fea­ture Extrac­tion plays a sig­nif­i­cant role in diag­nos­ing most of the car­diac dis­eases. One car­diac cycle in an ECG sig­nal con­sists of the P-​​QRS-​​T waves. This fea­ture extrac­tion scheme deter­mines the ampli­tudes and inter­vals in the ECG sig­nal for sub­se­quent analy­sis. The ampli­tudes and inter­vals value of P-​​QRS-​​T seg­ment deter­mines the func­tion­ing of heart of every human. Recently, numer­ous research and tech­niques have been devel­oped for ana­lyz­ing the ECG sig­nal. The pro­posed schemes were mostly based on Fuzzy Logic Meth­ods, Arti­fi­cial Neural Net­works (ANN), Genetic Algo­rithm (GA), Sup­port Vec­tor Machines (SVM), and other Sig­nal Analy­sis tech­niques. All these tech­niques and algo­rithms have their advan­tages and limitations.…