Items of some interest:

These are my recent Pin​board​.in links:

Items of some interest:

These are my recent Pin​board​.in links:

  • [1206.6866] Sto­chas­tic Opti­mal Con­trol in Con­tin­u­ous Space-​​Time Multi-​​Agent Systems

    “Recently, a the­ory for sto­chas­tic opti­mal con­trol in non-​​linear dynam­i­cal sys­tems in con­tin­u­ous space-​​time has been devel­oped (Kap­pen, 2005). We apply this the­ory to col­lab­o­ra­tive multi-​​agent sys­tems. The agents evolve accord­ing to a given non-​​linear dynam­ics with addi­tive Wiener noise. Each agent can con­trol its own dynam­ics. The goal is to min­i­mize the accu­mu­lated joint cost, which con­sists of a state depen­dent term and a term that is qua­dratic in the con­trol. We focus on sys­tems of non-​​interacting agents that have to dis­trib­ute them­selves opti­mally over a num­ber of tar­gets, given a set of end-​​costs for the dif­fer­ent pos­si­ble agent-​​target com­bi­na­tions. We show that opti­mal con­trol is the com­bi­na­to­r­ial sum of inde­pen­dent single-​​agent single-​​target opti­mal con­trols weighted by a fac­tor pro­por­tional to the end-​​costs of the dif­fer­ent com­bi­na­tions. Thus, multi-​​agent con­trol is related to a stan­dard graph­i­cal model infer­ence prob­lem. The addi­tional com­pu­ta­tional cost com­pared to single-​​agent con­trol is expo­nen­tial in the tree-​​width of the graph spec­i­fy­ing the com­bi­na­to­r­ial sum times the num­ber of tar­gets. We illus­trate the result by sim­u­la­tions of sys­tems with up to 42 agents.”

    coor­di­na­tion agent-​​based emergent-​​design nudge-​​targets control-​​theory
  • [1205.6917] Robust self-​​triggered coor­di­na­tion with ternary controllers

    “This paper regards coor­di­na­tion of net­worked sys­tems, which is stud­ied in the frame­work of hybrid dynam­i­cal sys­tems. We design a coor­di­na­tion scheme which com­bines the use of ternary con­trollers with a self-​​triggered com­mu­ni­ca­tion pol­icy. The com­mu­ni­ca­tion pol­icy requires the agents to col­lect, at each sam­pling time, rel­a­tive mea­sure­ments of their neigh­bors’ states: the col­lected infor­ma­tion is then used to update the con­trol and deter­mine the fol­low­ing sam­pling time. We prove that the pro­posed scheme ensures finite-​​time con­ver­gence to a neigh­bor­hood of a con­sen­sus state. We then study the robust­ness of the pro­posed self-​​triggered coor­di­na­tion sys­tem with respect to skews in the agents’ local clocks, to delays, and to lim­ited pre­ci­sion in com­mu­ni­ca­tion. Fur­ther­more, we present two sig­nif­i­cant vari­a­tions of our scheme. First, we design a time-​​varying con­troller which asymp­tot­i­cally dri­ves the sys­tem to con­sen­sus. Sec­ond, we adapt our frame­work to a com­mu­ni­ca­tion model in which an agent does not poll all its neigh­bors simul­ta­ne­ously, but sin­gle neigh­bors instead. This com­mu­ni­ca­tion pol­icy actu­ally leads to a self-​​triggered “gos­sip” coor­di­na­tion system.”

    network-​​theory multi-​​agent-​​systems mechanism-​​design emergent-​​design consensus-​​systems nudge-​​targets algo­rithms
  • [1205.2607] Simulation-​​Based Game The­o­retic Analy­sis of Key­word Auc­tions with Low-​​Dimensional Bid­ding Strategies

    “We per­form a simulation-​​based analy­sis of key­word auc­tions mod­eled as one-​​shot games of incom­plete infor­ma­tion to study a series of mech­a­nism design ques­tions. Our first ques­tion addresses the degree to which incen­tive com­pat­i­bil­ity fails in gen­er­al­ized second-​​price (GSP) auc­tions. Our results sug­gest that sin­cere bid­ding in GSP auc­tions is a strik­ingly poor strat­egy and a poor pre­dic­tor of equi­lib­rium out­comes. We next show that the rank-​​by-​​revenue mech­a­nism is wel­fare opti­mal, cor­rob­o­rat­ing past results. Finally, we ana­lyze profit as a func­tion of auc­tion mech­a­nism under a series of alter­na­tive set­tings. Our con­clu­sions coin­cide with those of Lahaie and Pen­nock [2007] when val­ues and qual­ity scores are strongly pos­i­tively cor­re­lated: in such a case, rank-​​by-​​bid rules are clearly supe­rior. We diverge, how­ever, in show­ing that auc­tions that put lit­tle weight on qual­ity scores almost uni­ver­sally dom­i­nate the pure rank-​​by-​​revenue scheme.”

    game-​​theory agent-​​based sim­u­la­tion eco­nom­ics auc­tion nudge-​​targets
  • Graph­ing the his­tory of phi­los­o­phy « Drunks&Lampposts

    “Each philoso­pher is a node in the net­work and the lines between them (or edges in the ter­mi­nol­ogy of graph the­ory) rep­re­sents lines of influ­ence. The node and text are sized accord­ing to the num­ber of con­nec­tions (both in and out). The algo­rithm that visu­alises the graph also tends to put the bet­ter con­nected nodes in the cen­tre of the dia­gram so we see the most influ­en­tial philoso­phers, in large text, clus­tered in the cen­tre. It all seems about right with the major fig­ures in the west­ern philo­soph­i­cal tra­di­tion tak­ing the cen­tre stage. (I need to also add the direc­tion of influ­ence with a arrow head – some­thing I’ve not got round to yet.) A short­com­ing how­ever is that this eval­u­a­tion only takes into account direct lines of influ­ence. Indi­rect influ­ence via another per­son in the net­work does not enter into it. This prob­a­bly explains why Descartes is smaller than you’d think. It would also be bet­ter if the nodes were sized only by the num­ber of out­ward con­nec­tions although I think over­all the dif­fer­ences would be slight. I’ll get round to that.”

    visu­al­iza­tion philosopher-​​mining
  • [1206.3666] Unsu­per­vised adap­ta­tion of brain machine inter­face decoders

    “The per­for­mance of neural decoders can degrade over time due to non­sta­tion­ar­i­ties in the rela­tion­ship between neu­ronal activ­ity and behav­ior. In this case, brain-​​machine inter­faces (BMI) require adap­ta­tion of their decoders to main­tain high per­for­mance across time. One way to achieve this is by use of peri­od­i­cal cal­i­bra­tion phases, dur­ing which the BMI sys­tem (or an exter­nal human demon­stra­tor) instructs the user to per­form cer­tain move­ments or behav­iors. This approach has two dis­ad­van­tages: (i) cal­i­bra­tion phases inter­rupt the autonomous oper­a­tion of the BMI and (ii) between two cal­i­bra­tion phases the BMI per­for­mance might not be sta­ble but con­tin­u­ously decrease. A bet­ter alter­na­tive would be that the BMI decoder is able to con­tin­u­ously adapt in an unsu­per­vised man­ner dur­ing autonomous BMI oper­a­tion, i.e. with­out know­ing the move­ment inten­tions of the user. In the present arti­cle, we present an effi­cient method for such unsu­per­vised train­ing of BMI sys­tems for con­tin­u­ous move­ment con­trol. The pro­posed method uti­lizes a cost func­tion derived from neu­ronal record­ings, which guides a learn­ing algo­rithm to eval­u­ate the decod­ing para­me­ters. We ver­ify the per­for­mance of our adap­tive method by sim­u­lat­ing a BMI user with an opti­mal feed­back con­trol model and its inter­ac­tion with our adap­tive BMI decoder. The sim­u­la­tion results show that the cost func­tion and the algo­rithm yield fast and pre­cise tra­jec­to­ries towards tar­gets at ran­dom ori­en­ta­tions on a 2-​​dimensional com­puter screen. For ini­tially unknown and non-​​stationary tun­ing para­me­ters, our unsu­per­vised method is still able to gen­er­ate pre­cise tra­jec­to­ries and to keep its per­for­mance sta­ble in the long term. The algo­rithm can option­ally work also with neu­ronal error sig­nals instead or in con­junc­tion with the pro­posed unsu­per­vised adaptation.”

    FOR-​​SCIENCE brain-​​machine-​​interface user-​​interface adaptive-​​control emergent-​​design
  • [1206.3980] Visu­al­iz­ing Stream­ing Text Data with Dynamic Maps

    “The many end­less rivers of text now avail­able present a seri­ous chal­lenge in the task of glean­ing, ana­lyz­ing and dis­cov­er­ing use­ful infor­ma­tion. In this paper, we describe a method­ol­ogy for visu­al­iz­ing text streams in real time. The approach auto­mat­i­cally groups sim­i­lar mes­sages into “coun­tries,” with key­word sum­maries, using seman­tic analy­sis, graph clus­ter­ing and map gen­er­a­tion tech­niques. It han­dles the need for visual sta­bil­ity across time by dynamic graph lay­out and Pro­crustes pro­jec­tion tech­niques, enhanced with a novel sta­ble com­po­nent pack­ing algo­rithm. The result pro­vides a con­tin­u­ous, suc­cinct view of evolv­ing top­ics of inter­est. It can be used in pas­sive mode for overviews and sit­u­a­tional aware­ness, or as an inter­ac­tive data explo­ration tool. To make these ideas con­crete, we describe their appli­ca­tion to an online ser­vice called TwitterScope.”

    visu­al­iza­tion data algo­rithms user-​​experience
  • [1205.3352] Revis­it­ing the effect of exter­nal fields in Axelrod’s model of social dynamics

    “The study of the effects of spa­tially uni­form fields on the steady-​​state prop­er­ties of Axelrod’s model has yielded plenty of con­tro­ver­sial results. Here we re-​​examine the impact of this type of field for a selec­tion of para­me­ters such that the field-​​free steady state of the model is het­ero­ge­neous or mul­ti­cul­tural. Analy­ses of both one and two-​​dimensional ver­sions of Axelrod’s model indi­cate that, con­trary to pre­vi­ous claims in the lit­er­a­ture, the steady state remains het­ero­ge­neous regard­less of the value of the field strength. Turn­ing on the field leads to a dis­con­tin­u­ous decrease on the num­ber of cul­tural domains, which we argue is due to the insta­bil­ity of zero-​​field het­ero­ge­neous absorb­ing con­fig­u­ra­tions. We find, how­ever, that spa­tially nonuni­form fields that imple­ment a con­sen­sus rule among the neigh­bor­hood of the agents enforces homog­e­niza­tion. Although the over­all effects of the fields are essen­tially the same irre­spec­tive of the dimen­sion­al­ity of the model, we argue that the dimen­sion­al­ity has a sig­nif­i­cant impact on the sta­bil­ity of the field-​​free homo­ge­neous steady state.”

    com­plex­ol­ogy agent-​​based exter­nal­i­ties sim­u­la­tion evolutionary-​​economics nudge-​​targets
  • [1206.3421] Lin­ear Latent Vari­able Mod­els: The lava-​​package

    “An R pack­age for spec­i­fy­ing and esti­mat­ing lin­ear latent vari­able mod­els is pre­sented. The phi­los­o­phy of the imple­men­ta­tion is to sep­a­rate the model spec­i­fi­ca­tion from the actual data, which leads to a dynamic and easy way of mod­el­ing com­plex hier­ar­chi­cal struc­tures. Sev­eral advanced fea­tures are imple­mented includ­ing robust stan­dard errors for clus­tered cor­re­lated data, multi­group analy­ses, non-​​linear para­me­ter con­straints, infer­ence with incom­plete data, max­i­mum like­li­hood esti­ma­tion with cen­sored and binary obser­va­tions, and instru­men­tal vari­able esti­ma­tors. In addi­tion an exten­sive sim­u­la­tion inter­face cov­er­ing a broad range of non-​​linear gen­er­al­ized struc­tural equa­tion mod­els is described. The model and soft­ware are demon­strated in data of mea­sure­ments of the sero­tonin trans­porter in the human brain.”

    ontol­ogy sta­tis­tics algo­rithms R-​​language linear-​​models model-​​view-​​controller design-​​patterns
  • [1205.6129] The Miss­ing Mem­ris­tor: Novel Nan­otech­nol­ogy or rather new Case Study for the Phi­los­o­phy and Soci­ol­ogy of Science?

    “In 2008, it was widely announced that the miss­ing mem­ris­tor, a basic two-​​terminal elec­tri­cal cir­cuit ele­ment, had finally been dis­cov­ered. The mem­ris­tor is the fourth and last such cir­cuit ele­ment and thus com­pletes cir­cuit the­ory. Pre­dicted already in 1971, the even­tual dis­cov­ery of some­thing seem­ingly so basic needed almost 40 years. How­ever, this dis­cov­ery is doubted. The pre­dicted mem­ris­tor has no mate­r­ial mem­ory and is based on mag­netic flux, but the dis­cov­ered devices con­sti­tute ana­logue mem­ory stor­age that do not involve mag­net­ism. The per­son who orig­i­nally pro­posed the mem­ris­tor did not reject the dis­cov­ery but instead changed his mind about what a mem­ris­tor is. We briefly intro­duce the his­tory and then care­fully mem­ris­tance and the mem­ris­tor as such. We dis­cuss its sta­tus as a model rather than a device. We dis­cuss the dis­cov­ered devices, their sta­bil­ity, and how sta­bil­ity relates to the con­sis­tency of the the­o­ret­i­cal enti­ties. A thought exper­i­ment assumes a world with­out mag­net­ism. Induc­tors can­not exist there, but mem­ory resis­tors could still be con­structed. On the same grounds as the mem­ris­tor was his­tor­i­cally pre­dicted, an “induc­tor” could then be pre­dicted. Likely, some­body would also ‘dis­cover’ one. A ten­ta­tive soci­o­log­i­cal analy­sis com­pares to the flawed detec­tion of grav­i­ta­tional waves but comes to very dif­fer­ent conclusions.”

    symmetry-​​is-​​the-​​way-​​things-​​ought-​​to-​​be history-​​of-​​science technical-​​assumptions engineering-​​philosophy
  • [1205.2776] Trans­port on cou­pled spa­tial networks

    “Trans­port processes on spa­tial net­works are rep­re­sen­ta­tive of a broad class of real world sys­tems which, rather than being inde­pen­dent, are typ­i­cally inter­de­pen­dent. We pro­pose a mea­sure of util­ity to cap­ture key fea­tures that arise when such sys­tems are cou­pled together. The cou­pling is defined in a way that is not solely topo­log­i­cal, rely­ing on both the dis­tri­b­u­tion of sources and sinks, and the method of route assign­ment. Using a toy model, we explore rel­e­vant cases by sim­u­la­tion. For cer­tain para­me­ter val­ues, a pic­ture emerges of two regimes. The first occurs when the flows go from many sources to a small num­ber of sinks. In this case, net­work util­ity is largest when the cou­pling is at its max­i­mum and the aver­age short­est path is min­i­mized. The sec­ond regime arises when many sources cor­re­spond to many sinks. Here, the opti­mal cou­pling no longer cor­re­sponds to the min­i­mum aver­age short­est path, as the con­ges­tion of traf­fic must also be taken into account. More gen­er­ally, results indi­cate that cou­pled spa­tial sys­tems can give rise to behav­ior that relies sub­tly on the inter­play between the cou­pling and ran­dom­ness in the source-​​sink distribution.”

    nudge-​​targets network-​​theory emergent-​​design epi­demi­ol­ogy sim­u­la­tion
  • [1206.5849] Under­stand­ing the com­plex­ity of the L’evy-walk nature of human mobil­ity with a multi-​​scale cost/​benefit model

    “Prob­a­bil­ity dis­tri­b­u­tions of human dis­place­ments has been fit with expo­nen­tially trun­cated L’evy flights or fat tailed Pareto inverse power law prob­a­bil­ity dis­tri­b­u­tions. Thus, peo­ple usu­ally stay within a given loca­tion (for exam­ple, the city of res­i­dence), but with a non-​​vanishing fre­quency they visit nearby or far loca­tions too. Herein, we show that an impor­tant empir­i­cal dis­tri­b­u­tion of human dis­place­ments (range: from 1 to 1000 km) can be well fit by three con­sec­u­tive Pareto dis­tri­b­u­tions with sim­ple inte­ger expo­nents equal to 1, 2 and ($gtrap­prox$) 3. These three expo­nents cor­re­spond to three dis­place­ment range zones of about 1 km $less­sim Delta r less­sim$ 10 km, 10 km $less­sim Delta r less­sim$ 300 km and 300 km $less­sim Delta r less­sim $ 1000 km, respec­tively. These three zones can be geo­graph­i­cally and phys­i­cally well deter­mined as dis­place­ments within a city, vis­its to nearby cities that may occur within just one-​​day trips, and visit to far loca­tions that may require multi-​​days trips. The incre­men­tal inte­ger val­ues of the three expo­nents can be eas­ily explained with a three-​​scale mobil­ity cost/​benefit model for human dis­place­ments based on sim­ple geo­met­ri­cal con­strains. Essen­tially, peo­ple would divide the space into three major regions (close, medium and far dis­tances) and would assume that the travel ben­e­fits are randomly/​uniformly dis­trib­uted mostly only within spe­cific urban-​​like areas.”

    Levy-​​flights city-​​planning multiscale-​​phenomena self-​​similarity
  • [1206.7011] Ultra­thin Ter­a­hertz Pla­nar Lenses

    “Con­ven­tional opti­cal com­po­nents shape the wave­front of prop­a­gat­ing light by adjust­ing the opti­cal path length, which requires the use of rather thick lenses, espe­cially for the adjust­ment of ter­a­hertz (THz) radi­a­tion due to its long wave­length. Two ultra­thin THz pla­nar lenses were designed and fab­ri­cated based on inter­face phase mod­u­la­tion of antenna res­o­nance. The lens thick­nesses were extremely reduced to 100 nm, which is only 14000th of the illu­mi­nat­ing light wave­length. The focus­ing and imag­ing func­tions of the lenses were exper­i­men­tally demon­strated. The ultra­thin opti­cal com­po­nents described herein are a sig­nif­i­cant step toward the devel­op­ment of a micro-​​integrated THz system.”

    optics engineering-​​design nudge-​​targets
  • [1206.3789] Tree decom­po­si­tion and para­me­ter­ized algo­rithms for RNA structure-​​sequence align­ment includ­ing ter­tiary inter­ac­tions and pseudoknots

    “We present a gen­eral set­ting for structure-​​sequence com­par­i­son in a large class of RNA struc­tures that uni­fies and gen­er­al­izes a num­ber of recent works on spe­cific fam­i­lies on struc­tures. Our approach is based on tree decom­po­si­tion of struc­tures and gives rises to a gen­eral para­me­ter­ized algo­rithm, where the expo­nen­tial part of the com­plex­ity depends on the fam­ily of struc­tures. For each of the pre­vi­ously stud­ied fam­i­lies, our algo­rithm has the same com­plex­ity as the spe­cific algo­rithm that had been given before.”

    RNA structural-​​biology fold­ing bioin­for­mat­ics algo­rithms nudge-​​targets mod­el­ing
  • [1206.3403] Topo­log­i­cal Mea­sure Locat­ing the Effec­tive Crossover between Seg­re­ga­tion and Inte­gra­tion in a Mod­u­lar Network

    “Com­pu­ta­tional analy­sis of time-​​course data with an under­ly­ing causal struc­ture is needed in a vari­ety of domains, includ­ing neural spike trains, stock price move­ments, and gene expres­sion lev­els. How­ever, it can be chal­leng­ing to deter­mine from just the numer­i­cal time course data alone what is coor­di­nat­ing the vis­i­ble processes, to sep­a­rate the under­ly­ing prima facie causes into gen­uine and spu­ri­ous causes and to do so with a fea­si­ble com­pu­ta­tional com­plex­ity. For this pur­pose, we have been devel­op­ing a novel algo­rithm based on a frame­work that com­bines notions of causal­ity in phi­los­o­phy with algo­rith­mic approaches built on model check­ing and sta­tis­ti­cal tech­niques for mul­ti­ple hypothe­ses test­ing. The causal rela­tion­ships are described in terms of tem­po­ral logic for­mu­lae, refram­ing the infer­ence prob­lem in terms of model check­ing. The logic used, PCTL, allows descrip­tion of both the time between cause and effect and the prob­a­bil­ity of this rela­tion­ship being observed. We show that equipped with these causal for­mu­lae with their asso­ci­ated prob­a­bil­i­ties we may com­pute the aver­age impact a cause makes to its effect and then dis­cover sta­tis­ti­cally sig­nif­i­cant causes through the con­cepts of mul­ti­ple hypoth­e­sis test­ing (treat­ing each causal rela­tion­ship as a hypoth­e­sis), and false dis­cov­ery con­trol. By explor­ing a well-​​chosen fam­ily of poten­tially all sig­nif­i­cant hypothe­ses with rea­son­ably min­i­mal descrip­tion length, it is pos­si­ble to tame the algorithm’s com­pu­ta­tional com­plex­ity while explor­ing the nearly com­plete search-​​space of all prima facie causes. We have tested these ideas in a num­ber of domains and illus­trate them here with two examples.”

    time-​​series sta­tis­tics frame­work dynamical-​​systems to-​​read
  • [1206.5959] The Online Replace­ment Path Problem

    “We study a nat­ural online vari­ant of the replace­ment path prob­lem. The textit{replacement path prob­lem} asks to find for a given graph $G = (V,E)$, two des­ig­nated ver­tices $s,tin V$ and a short­est $s$-$t$ path $P$ in $G$, a textit{replacement path} $P_​e$ for every edge $e$ on the path $P$. The replace­ment path $P_​e$ is sim­ply a short­est $s$-$t$ path in the graph, which avoids the textit{failed} edge $e$. We adapt this prob­lem to deal with the nat­ural sce­nario, that the edge which failed is not known at the time of solu­tion imple­men­ta­tion. Instead, our prob­lem assumes that the iden­tity of the failed edge only becomes avail­able when the rout­ing mech­a­nism tries to cross the edge. This sit­u­a­tion is moti­vated by appli­ca­tions in dis­trib­uted net­works, where infor­ma­tion about recent changes in the net­work is only stored locally, and fault-​​tolerant opti­miza­tion, where an adver­sary tries to delay the dis­cov­ery of the mate­ri­al­ized sce­nario as much as pos­si­ble. Con­se­quently, we define the textit{online replace­ment path prob­lem}, which asks to find a nom­i­nal $s$-$t$ path $Q$ and detours $Q_​e$ for every edge on the path $Q$, such that the worst-​​case arrival time at the des­ti­na­tion is min­i­mized. Our main con­tri­bu­tion is a label set­ting algo­rithm, which solves the prob­lem in undi­rected graphs in time $O(m log n)$ and lin­ear space for all sources and a sin­gle des­ti­na­tion. We also present algo­rithms for exten­sions of the model to any bounded num­ber of failed edges.”

    operations-​​research plan­ning algo­rithms online-​​algorithms nudge-​​targets
  • [1206.6195] Par­rondo games with spa­tial depen­dence, II

    “…Again this sug­gests that the Par­rondo region (mu_​B is non­pos­i­tive and mu_[r,s] is pos­i­tive) has nonzero vol­ume in the limit. …”

    Parrondo-​​games game-​​theory counterintuitive-​​results sim­u­la­tion coop­er­a­tion
  • [1206.4648] Two-​​Manifold Prob­lems with Appli­ca­tions to Non­lin­ear Sys­tem Identification

    “Recently, there has been much inter­est in spec­tral approaches to learn­ing manifolds—so-called ker­nel eigen­map meth­ods. These meth­ods have had some suc­cesses, but their applic­a­bil­ity is lim­ited because they are not robust to noise. To address this lim­i­ta­tion, we look at two-​​manifold prob­lems, in which we simul­ta­ne­ously recon­struct two related man­i­folds, each rep­re­sent­ing a dif­fer­ent view of the same data. By solv­ing these inter­con­nected learn­ing prob­lems together, two-​​manifold algo­rithms are able to suc­ceed where a non-​​integrated approach would fail: each view allows us to sup­press noise in the other, reduc­ing bias. We pro­pose a class of algo­rithms for two-​​manifold prob­lems, based on spec­tral decom­po­si­tion of cross-​​covariance oper­a­tors in Hilbert space, and dis­cuss when two-​​manifold prob­lems are use­ful. Finally, we demon­strate that solv­ing a two-​​manifold prob­lem can aid in learn­ing a non­lin­ear dynam­i­cal sys­tem from lim­ited data.”

    sta­tis­tics inverse-​​problems system-​​identification algo­rithms nudge-​​targets bench­mark­ing
  • [1203.3367] Sto­chas­tic dif­fer­en­tial equa­tions for evo­lu­tion­ary dynam­ics with demo­graphic noise and mutations

    “We present a gen­eral frame­work to describe the evo­lu­tion­ary dynam­ics of an arbi­trary num­ber of types in finite pop­u­la­tions based on sto­chas­tic dif­fer­en­tial equa­tions (SDE). For large, but finite pop­u­la­tions this allows to include demo­graphic noise with­out requir­ing explicit sim­u­la­tions. Instead, the pop­u­la­tion size only rescales the ampli­tude of the noise. More­over, this frame­work admits the inclu­sion of muta­tions between dif­fer­ent types, pro­vided that muta­tion rates, $mu$, are not too small com­pared to the inverse pop­u­la­tion size 1/​N. This ensures that all types are almost always rep­re­sented in the pop­u­la­tion and that the occa­sional extinc­tion of one type does not result in an extended absence of that type. For $mu Nll1$ this lim­its the use of SDE’s, but in this case there are well estab­lished alter­na­tive approx­i­ma­tions based on time scale sep­a­ra­tion. We illus­trate our approach by a Rock-​​Scissors-​​Paper game with muta­tions, where we demon­strate excel­lent agree­ment with sim­u­la­tion based results for suf­fi­ciently large pop­u­la­tions. In the absence of muta­tions the excel­lent agree­ment extends to small pop­u­la­tion sizes.”

    finite-​​size-​​effects population-​​biology noise-​​in-​​design nudge-​​targets
  • [1206.5352] The Sub­word Com­plex­ity of k-​​Automatic Sequences is k-​​Synchronized

    “We show that the sub­word com­plex­ity func­tion p_​x (n), which counts the num­ber of dis­tinct fac­tors of length n of a k-​​automatic sequence x, is k-​​synchronized in the sense of Carpi. As an appli­ca­tion, we gen­er­al­ize recent results of Gold­stein. In con­trast, we show that func­tion which counts the num­ber of unbor­dered fac­tors of length n is not k-​​synchronized.”

    automata strings com­plex­ity nudge-​​targets
  • Con­vex hull algo­rithms — Wikipedia, the free encyclopedia

    “Com­put­ing the con­vex hull means that a non-​​ambiguous and effi­cient rep­re­sen­ta­tion of the required con­vex shape is con­structed. The com­plex­ity of the cor­re­spond­ing algo­rithms is usu­ally esti­mated in terms of n, the num­ber of input points, and h, the num­ber of points on the con­vex hull.”

    computational-​​geometry algo­rithms nudge-​​targets
  • [1205.6363] What Should Devel­op­ers Be Aware Of? An Empir­i­cal Study on the Direc­tives of API Documentation

    “Appli­ca­tion Pro­gram­ming Inter­faces (API) are exposed to devel­op­ers in order to reuse soft­ware libraries. API direc­tives are natural-​​language state­ments in API doc­u­men­ta­tion that make devel­op­ers aware of con­straints and guide­lines related to the usage of an API. This paper presents the design and the results of an empir­i­cal study on the direc­tives of API doc­u­men­ta­tion of object-​​oriented libraries. Its main con­tri­bu­tion is to pro­pose and exten­sively dis­cuss a tax­on­omy of 23 kinds of API directives.”

    digital-​​humanities doc­u­men­ta­tion text-​​mining software-​​development cultural-​​norms
  • Pub­li­ca­tions Avail­able Electronically

    God­fried T. Tou­s­saint, “The Erdos-​​Nagy the­o­rem and its ram­i­fi­ca­tions,” Proc. 11th Cana­dian Con­fer­ence on Com­pu­ta­tional Geom­e­try, Van­cou­ver, Canada, August 16–18 1999, pp. 9–12. Long ver­sion avail­able at: http://​www​.cs​.ubc​.ca/​c​o​n​f​e​r​e​n​c​e​s​/​C​CCG. Also: Tech­ni­cal Report No. SOCS-99.2, School of Com­puter Sci­ence, McGill Uni­ver­sity, June 18, 1999.

    nudge-​​targets geom­e­try genetic-​​programming-​​target low-​​hanging-​​fruit
  • Intro­duc­tion to Lam­ina Emer­gent Mech­a­nisms (LEMS) | Com­pli­ant Mechanisms

    “The fact that LEMs can be fab­ri­cated from pla­nar lay­ers influ­ences both what processes can be used for their man­u­fac­ture and what mate­ri­als may be used in their con­struc­tion. At the micro level, LEMS can be fab­ri­cated using single-​​layer MEMS fab­ri­ca­tion meth­ods and mate­ri­als, which offers sig­nif­i­cant cost and reli­a­bil­ity advan­tages. It also pro­vides oppor­tu­ni­ties for com­plex out-​​of-​​plane motion with only a sin­gle layer. At the macro level, man­u­fac­tur­ing processes used to make sta­tic struc­tures or com­po­nents for assem­bly can be used to cre­ate mech­a­nisms capa­ble of sophis­ti­cated motions and com­plex tasks. Example processes include stamp­ing, fine blank­ing, laser cut­ting, water jet cut­ting, plasma cut­ting, and wire elec­tri­cal dis­charge machin­ing (EDM). Some of these processes, such as stamp­ing, offer sig­nif­i­cant cost advan­tages for high-​​volume pro­duc­tion. Pla­nar fab­ri­ca­tion also allows the use of sheet goods to directly cre­ate mech­a­nisms. The use of low-​​cost, high-​​quality sheet goods has the poten­tial to dra­mat­i­cally reduce cost for high-​​volume pro­duc­tion. It also makes pos­si­ble the next char­ac­ter­is­tic: flat ini­tial state.”

    man­u­fac­tur­ing engineering-​​design fab­ri­ca­tion nudge-​​targets

Items of some interest:

These are my recent Pin​board​.in links:

  • Cin­derella Doc­u­men­ta­tion : The­o­ret­i­cal Background

    “At first sight it is not clear whether this require­ment is sat­is­fi­able in gen­eral. Turn on your favorite sys­tem for doing inter­ac­tive geom­e­try or para­met­ric CAD and make the fol­low­ing exper­i­ment: Draw a hor­i­zon­tal line and con­struct two cir­cles of equal radius whose cen­ters are con­strained to slide along the line. Move the cir­cles to a posi­tion in which they inter­sect, and con­struct the upper point of inter­sec­tion of the two cir­cles. Now move one cir­cle so that its cen­ter passes through the cen­ter of the other cir­cle. Most prob­a­bly you will see that the point of inter­sec­tion sud­denly jumps from the upper inter­sec­tion to the lower one. This is what has hap­pened in all the sys­tems we have tried so far. Such behav­ior runs counter to our require­ment of con­ti­nu­ity: You make a small move, and a depen­dent point sud­denly jumps.”

    sim­u­la­tion geom­e­try algo­rithms rep­re­sen­ta­tion nudge-​​targets
  • Reli­gion and the City | New​geog​ra​phy​.com

    “He talks about items rang­ing from mul­ti­cul­tural sen­si­tiv­i­ties to tak­ing the arts seri­ous to “being famous for help­ing the poor.” The lat­ter was an item that jumped out at me because, as I’ve noted before, too many urban­ist argu­ments are basi­cally argu­ments for what I call “Star­bucks urban­ism.” If called on this, peo­ple will say, “But of course tran­sit will ben­e­fit the poor too.” But that’s not how it’s sold. Urban­ists ought to be famous for the way they design, imple­ment, and talk about their poli­cies as instru­ments for help­ing the poor and facil­i­tat­ing upward eco­nomic and social mobil­ity. There’s a lot of other good stuff in the video that’s rel­e­vant to urban­ism. For those who pre­fer read­ing, Keller also wrote a paper called “Our New Global Cul­ture: Min­istry in Cities, which says of itself: “This paper sur­veys the rise of global cities, the cul­ture and dom­i­nant world­views within these cities, and a frame­work for min­is­ter­ing in them.”?

    city-​​planning orga­ni­za­tion mar­ket­ing Workan­tile Coscience out­reach diver­sity man­age­ment
  • Study Hacks » Blog Archive » What You Know Mat­ters More Than What You Do

    “Accord­ing to my col­leagues, this star researcher tends to begin with tech­niques, not prob­lems. He first mas­ters a tech­nique that seems promis­ing (and when I say “mas­ter,” I mean it — he really goes deep in build­ing his under­stand­ing). He then uses this new tech­nique to seek out prob­lems that were once hard but now yield eas­ily. He’s rest­less in this quest, often mas­ter­ing sev­eral new tech­niques each year.”

    heuris­tics work­life inno­va­tion pro­duc­tiv­ity problem-​​seeing problem-​​solving
  • Jour­natic worker takes ‘This Amer­i­can Life’ inside out­sourced jour­nal­ism | Poynter.

    “If you’ve never heard of Jour­natic, that’s kind of the idea. The com­pany, which was founded in 2006, has a web­site that doesn’t appear on at least the first five pages of Google search results. Job open­ings, often posted on Craigslist or Jour​nal​is​mJobs​.com, once men­tioned the company’s name, but no longer. Jour­natic cur­rently works with “dozens” of media com­pa­nies, Tim­pone said, though he declined to name them. He’s spo­ken before of the real estate sec­tion Jour­natic pro­duces for the San Fran­cisco Chron­i­cle. He said more are sign­ing up all the time.”

    disintermediation-​​in-​​action jour­nal­ism work­life out­sourc­ing exposé
  • A Step-​​by-​​Step Guide to Tribal Lead­er­ship: Part 1: The Five Stages of Tribal Cul­ture « emer­gent by design

    “Tribal Lead­ers are the peo­ple who focus their efforts on upgrad­ing the tribal cul­ture. (upgrad­ing the words we use to describe our real­ity and the behav­iors we prac­tice that shape the direc­tion of our lives) They set the stan­dard of per­for­mance in their indus­tries, from pro­duc­tiv­ity and prof­itabil­ity to employee reten­tion, and attract tal­ent. Most of all, they help bring groups to unity by rec­og­niz­ing their ‘trib­al­ness’ – get­ting peo­ple to talk about the things they really care about, com­ing together around these com­mon causes, and form­ing mis­sions to make some­thing great hap­pen, and to live in great­ness. The goal of Tribal Lead­er­ship is to learn how to get peo­ple ‘unstuck’ – from unhelp­ful lan­guage and behav­iors, so we can level up and tran­si­tion into higher-​​performance, less stress­ful, and more fun states of Being.”

    i-​​hate-​​the-​​word-​​tribes col­lab­o­ra­tion lead­er­ship cultural-​​dynamics advice
  • [1206.6532] Esti­mat­ing Nui­sance Para­me­ters in Inverse Problems

    “Many inverse prob­lems include nui­sance para­me­ters which, while not of direct inter­est, are required to recover pri­mary para­me­ters. Struc­ture present in these prob­lems allows effi­cient opti­miza­tion strate­gies — a well known exam­ple is vari­able pro­jec­tion, where non­lin­ear least squares prob­lems which are lin­ear in some para­me­ters can be very effi­ciently opti­mized. In this paper, we extend the idea of pro­ject­ing out a sub­set over the vari­ables to a broad class of max­i­mum like­li­hood (ML) and max­i­mum a pos­te­ri­ori like­li­hood (MAP) prob­lems with nui­sance para­me­ters, such as vari­ance or degrees of free­dom. As a result, we are able to incor­po­rate nui­sance para­me­ter esti­ma­tion into large-​​scale con­strained and uncon­strained inverse prob­lem for­mu­la­tions. We apply the approach to a vari­ety of prob­lems, includ­ing esti­ma­tion of unknown vari­ance para­me­ters in the Gauss­ian model, degree of free­dom (d.o.f.) para­me­ter esti­ma­tion in the con­text of robust inverse prob­lems, auto­matic cal­i­bra­tion, and opti­mal exper­i­men­tal design. Using numer­i­cal exam­ples, we demon­strate improve­ment in recov­ery of pri­mary para­me­ters for sev­eral large– scale inverse prob­lems. The pro­posed approach is com­pat­i­ble with a wide vari­ety of algo­rithms and for­mu­la­tions, and its imple­men­ta­tion requires only minor mod­i­fi­ca­tions to exist­ing algorithms.”

    reinventing-​​the-​​wheel feature-​​extraction opti­miza­tion modeling-​​is-​​not-​​mathematics nudge-​​targets
  • [1206.4608] A Hybrid Algo­rithm for Con­vex Semi­def­i­nite Optimization

    “We present a hybrid algo­rithm for opti­miz­ing a con­vex, smooth func­tion over the cone of pos­i­tive semi­def­i­nite matri­ces. Our algo­rithm con­verges to the global opti­mal solu­tion and can be used to solve gen­eral large-​​scale semi­def­i­nite pro­grams and hence can be read­ily applied to a vari­ety of machine learn­ing prob­lems. We show exper­i­men­tal results on three machine learn­ing prob­lems (matrix com­ple­tion, met­ric learn­ing, and sparse PCA) . Our approach out­per­forms state-​​of-​​the-​​art algorithms.”

    algo­rithms opti­miza­tion computational-​​complexity spe­cial­iza­tion nudge-​​targets
  • [1206.6690] Gen­er­a­tion and Prop­er­ties of Snarks

    “For many of the unsolved prob­lems con­cern­ing cycles and match­ings in graphs it is known that it is suf­fi­cient to prove them for emph{snarks}, the class of non­triv­ial 3-​​regular graphs which can­not be 3-​​edge coloured. In the first part of this paper we present a new algo­rithm for gen­er­at­ing all non-​​isomorphic snarks of a given order. Our imple­men­ta­tion of the new algo­rithm is 14 times faster than pre­vi­ous pro­grams for gen­er­at­ing snarks, and 29 times faster for gen­er­at­ing weak snarks. Using this pro­gram we have gen­er­ated all non-​​isomorphic snarks on $nleq 36$ ver­tices. Pre­vi­ously lists up to $n=28$ ver­tices have been pub­lished. In the sec­ond part of the paper we ana­lyze the sets of gen­er­ated snarks with respect to a num­ber of prop­er­ties and con­jec­tures. We find that some of the strongest ver­sions of the cycle dou­ble cover con­jec­ture hold for all snarks of these orders, as does Jaeger’s Petersen colour­ing con­jec­ture, which in turn implies that Fulkerson’s con­jec­ture has no small coun­terex­am­ples. In con­trast to these pos­i­tive results we also find coun­terex­am­ples to eight pre­vi­ously pub­lished con­jec­tures con­cern­ing cycle cov­er­ings and the gen­eral cycle struc­ture of cubic graphs.”

    graph-​​theory com­bi­na­torics algo­rithms nudge-​​targets
  • [1206.6238] Entrain­abil­ity enhance­ment by period mis­match in biloop genetic oscillators

    “Effects of the period mis­match on entrain­ment prop­er­ties in two cou­pled genetic oscil­la­tors are stud­ied. The entrain­ment is cal­cu­lated with a phase reduc­tion approach and a Flo­quet mul­ti­plier analy­sis, and their depen­den­cies on cou­pling strength and the period ratio are inves­ti­gated in two genetic oscil­la­tor mod­els (smooth and relax­ation oscil­la­tors). We find that the exis­tence of the period mis­match induces an enhance­ment of entrain­ment in both smooth and relax­ation oscil­la­tors. By cal­cu­lat­ing Flo­quet mul­ti­pli­ers, we show that the enhance­ment mech­a­nism is based on the cou­pled oscil­la­tors which are in the vicin­ity of bifur­ca­tion on limit cycle.”

    biological-​​engineering emergent-​​design reaction-​​networks oscil­la­tors control-​​theory
  • [1206.4672] Effi­cient Active Algo­rithms for Hier­ar­chi­cal Clustering

    “Advances in sens­ing tech­nolo­gies and the growth of the inter­net have resulted in an explo­sion in the size of mod­ern datasets, while stor­age and pro­cess­ing power con­tinue to lag behind. This moti­vates the need for algo­rithms that are effi­cient, both in terms of the num­ber of mea­sure­ments needed and run­ning time. To com­bat the chal­lenges asso­ci­ated with large datasets, we pro­pose a gen­eral frame­work for active hier­ar­chi­cal clus­ter­ing that repeat­edly runs an off-​​the-​​shelf clus­ter­ing algo­rithm on small sub­sets of the data and comes with guar­an­tees on per­for­mance, mea­sure­ment com­plex­ity and run­time com­plex­ity. We instan­ti­ate this frame­work with a sim­ple spec­tral clus­ter­ing algo­rithm and pro­vide con­crete results on its per­for­mance, show­ing that, under some assump­tions, this algo­rithm recov­ers all clus­ters of size ?(log n) using O(n log^2 n) sim­i­lar­i­ties and runs in O(n log^3 n) time for a dataset of n objects. Through exten­sive exper­i­men­ta­tion we also demon­strate that this frame­work is prac­ti­cally alluring.”

    clus­ter­ing algo­rithms nudge-​​targets practically-​​alluring
  • Most Bla­tant Pro-​​ACTA Cam­paign So Far Is A Copy­right Monop­oly Vio­la­tion — Falkvinge on Infopolicy

    “This episode shows clearer than ever that the copy­right and patent monop­o­lies are not intended to be pro­tec­tive of inno­va­tion or pro­tec­tive of the econ­omy. They’re obvi­ously too com­plex even for their strongest sup­port­ers and lob­by­ists to under­stand and adhere to. Rather, they are intended as legal clubs to be used by the now-​​rich incum­bents against resource-​​strapped upstarts. The copy­right and patent monop­o­lies are only pro­tec­tive of the past, pro­tec­tive against the present and future of inno­va­tion, cre­ativ­ity, and economy.”

    copy­right intellectual-​​property cor­po­ratism public-​​policy
  • [1112.5218] Pat­terns of neu­tral diver­sity under gen­eral mod­els of selec­tive sweeps

    “Two major sources of sto­chas­tic­ity in the dynam­ics of neu­tral alle­les result from resam­pling of finite pop­u­la­tions (genetic drift) and the ran­dom genetic back­ground of nearby selected alle­les on which the neu­tral alle­les are found (linked selec­tion). There is now good evi­dence that linked selec­tion plays an impor­tant role in shap­ing poly­mor­phism lev­els in a num­ber of species. One of the best inves­ti­gated mod­els of linked selec­tion is the recur­rent full sweep model, in which newly arisen selected alle­les fix rapidly. How­ever, the bulk of selected alle­les that sweep into the pop­u­la­tion may not be des­tined for rapid fix­a­tion. Here we develop a gen­eral model of recur­rent selec­tive sweeps in a coa­les­cent frame­work, one that gen­er­al­izes the recur­rent full sweep model to the case where selected alle­les do not sweep to fix­a­tion. We show that in a large pop­u­la­tion, only the ini­tial rapid increase of a selected allele affects the geneal­ogy at par­tially linked sites, which under fairly gen­eral assump­tions are unaf­fected by the sub­se­quent fate of the selected allele. We also apply the the­ory to a sim­ple model to inves­ti­gate the impact of recur­rent par­tial sweeps on lev­els of neu­tral diver­sity, and find that for a given reduc­tion in diver­sity, the impact of recur­rent par­tial sweeps on the fre­quency spec­trum at neu­tral sites is deter­mined pri­mar­ily by the fre­quen­cies achieved by the selected alle­les. Con­se­quently, recur­rent sweeps of selected alle­les to low fre­quen­cies can have a pro­found effect on lev­els of diver­sity but can leave the fre­quency spec­trum rel­a­tively unper­turbed. In fact, the lim­it­ing coa­les­cent model under a high rate of sweeps to low fre­quency is iden­ti­cal to the stan­dard neu­tral model. The gen­eral model of selec­tive sweeps we describe goes some way towards pro­vid­ing a more flex­i­ble frame­work to describe genomic pat­terns of diver­sity than is cur­rently available.”

    neutral-​​networks evolutionary-​​dynamics fitness-​​landscapes diver­sity theoretical-​​biology
  • [1206.3520] Recov­er­ing the tree-​​like trend of evo­lu­tion despite exten­sive lat­eral genetic trans­fer: A prob­a­bilis­tic analysis

    “In the pres­ence of high­ways, deal­ing with more gen­eral net­work set­tings would be desir­able. Also our def­i­n­i­tion of high­ways as con­nect­ing two edges is some­what restric­tive. In gen­eral, one is also inter­ested in pref­er­en­tial genetic trans­fers between clades.”

    algo­rithms lateral-​​gene-​​transfer cladis­tics phy­lo­ge­net­ics inverse-​​problems ontol­ogy modeling-​​is-​​not-​​mathematics nudge-​​targets
  • [1206.3279] The Phy­lo­ge­netic Indian Buf­fet Process: A Non-​​Exchangeable Non­para­met­ric Prior for Latent Features

    “Non­para­met­ric Bayesian mod­els are often based on the assump­tion that the objects being mod­eled are exchange­able. While appro­pri­ate in some appli­ca­tions (e.g., bag-​​of-​​words mod­els for doc­u­ments), exchange­abil­ity is some­times assumed sim­ply for com­pu­ta­tional rea­sons; non-​​exchangeable mod­els might be a bet­ter choice for appli­ca­tions based on sub­ject mat­ter. Draw­ing on ideas from graph­i­cal mod­els and phy­lo­ge­net­ics, we describe a non-​​exchangeable prior for a class of non­para­met­ric latent fea­ture mod­els that is nearly as effi­cient com­pu­ta­tion­ally as its exchange­able coun­ter­part. Our model is applic­a­ble to the gen­eral set­ting in which the depen­den­cies between objects can be expressed using a tree, where edge lengths indi­cate the strength of rela­tion­ships. We demon­strate an appli­ca­tion to mod­el­ing prob­a­bilis­tic choice.”

    sta­tis­tics algo­rithms ontol­ogy col­li­ga­tion feature-​​extraction philosophy-​​of-​​science nudge-​​targets
  • “The Eurozone’s Strat­egy is a Dis­as­ter” « naked capitalism

    “Why should Ger­man and other tax­pay­ers, mostly from the north, pay for the oth­ers, mostly from the south? Because their gov­ern­ments are respon­si­ble for the dis­as­trous sit­u­a­tion we are in.”

    financial-​​crisis public-​​policy eco­nom­ics cultural-​​dynamics fair-​​weather-​​bosses
  • [1206.6504] An Abstract Approach to Strat­i­fi­ca­tion in Lin­ear Logic

    “We study the notion of strat­i­fi­ca­tion, as used in sub­sys­tems of lin­ear logic with low com­plex­ity bounds on the cut-​​elimination pro­ce­dure (the so-​​called light log­ics), from an abstract point of view, intro­duc­ing a log­i­cal sys­tem in which strat­i­fi­ca­tion is han­dled by a sep­a­rate modal­ity. This modal­ity, which is a gen­er­al­iza­tion of the para­graph modal­ity of Girard’s light lin­ear logic, arises from a gen­eral cat­e­gor­i­cal con­struc­tion applic­a­ble to all mod­els of lin­ear logic. We thus learn that strat­i­fi­ca­tion may be for­mu­lated inde­pen­dently of expo­nen­tial modal­i­ties; when it is forced to be con­nected to expo­nen­tial modal­i­ties, it yields inter­est­ing com­plex­ity prop­er­ties. In par­tic­u­lar, from our analy­sis stem three alter­na­tive refor­mu­la­tions of Bail­lot and Mazza’s lin­ear logic by lev­els: one geo­met­ric, one inter­ac­tive, and one semantic.”

    linear-​​logic logic-​​programming for­mal­iza­tion nudge-​​targets rep­re­sen­ta­tion
  • [1205.0802] Win-​​stay-​​lose-​​learn pro­motes coop­er­a­tion in the spa­tial prisoner’s dilemma game

    “Hold­ing on to one’s strat­egy is nat­ural and com­mon if the later war­rants suc­cess and sat­is­fac­tion. This goes against wide­spread sim­u­la­tion prac­tices of evo­lu­tion­ary games, where play­ers fre­quently con­sider chang­ing their strat­egy even though their pay­offs may be mar­gin­ally dif­fer­ent than those of the other play­ers. Inspired by this obser­va­tion, we intro­duce an aspiration-​​based win-​​stay-​​lose-​​learn strat­egy updat­ing rule into the spa­tial prisoner’s dilemma game. The rule is sim­ple and intu­itive, fore­see­ing strat­egy changes only by dis­sat­is­fied play­ers, who then attempt to adopt the strat­egy of one of their near­est neigh­bors, while the strate­gies of sat­is­fied play­ers are not sub­ject to change. We find that the pro­posed win-​​stay-​​lose-​​learn rule pro­motes the evo­lu­tion of coop­er­a­tion, and it does so very robustly and inde­pen­dently of the ini­tial con­di­tions. In fact, we show that even a minute ini­tial frac­tion of coop­er­a­tors may be suf­fi­cient to even­tu­ally secure a highly coop­er­a­tive final state. In addi­tion to exten­sive sim­u­la­tion results that sup­port our con­clu­sions, we also present results obtained by means of the pair approx­i­ma­tion of the stud­ied game. Our find­ings con­tinue the suc­cess story of related win-​​stay strat­egy updat­ing rules, and by doing so reveal new ways of resolv­ing the prisoner’s dilemma.”

    game-​​theory agent-​​based com­plex­ol­ogy
  • The Rude Pundit

    “And there’s every­thing you need to know about the Repub­li­can Party. “Shit hap­pened, but so what? Peo­ple were vic­tim­ized, but why should we care? That was nearly forty years ago.” The demen­tia in refus­ing to look back­ward, refus­ing to make up for the mis­takes of the past, whether it’s the Bush tax cuts or the lies that got us into war or the lies that got us into this finan­cial cri­sis, makes us damned to repeat. ”

    sum­mary pol­i­tics Repub­li­cans

Items of some interest:

These are my recent Pin​board​.in links:

  • Tar­get Expres­sion Exam­ples — Eureqa Formulize

    ‘The “Tar­get Expres­sion” in the field at the top of the Set Tar­get tab tells For­mulize what type of model to search for. By default, the tar­get expres­sion is an equa­tion where y (or, if there’s no y, what­ever vari­able is in col­umn A) is mod­eled as a func­tion of all other vari­ables. To edit the tar­get expres­sion, click on it, then make the desired alter­ations. Use the spe­cial func­tion f(…) to spec­ify the part of the equa­tion that For­mulize will attempt to fill in; For­mulize will search for the for­mula f(…) using the vari­ables you put inside the parentheses.’

    for­mulize eureqa genetic-​​programming symbolic-​​regression mod­el­ing doc­u­men­ta­tion
  • A New Solu­tion to the Puz­zle of Sim­plic­ity — PhilSci-​​Archive

    “Explain­ing the con­nec­tion, if any, between sim­plic­ity and truth is among the deep­est prob­lems fac­ing the phi­los­o­phy of sci­ence, sta­tis­tics, and machine learn­ing. Say that an effi­cient truth-​​finding method min­i­mizes worst-​​case costs en route to con­verg­ing to the true answer to a the­ory choice prob­lem. Let the costs con­sid­ered include the num­ber of times a false answer is selected, the num­ber of times opin­ion is reversed, and the times at which the rever­sals occur. It is demon­strated that (1)always choos­ing the sim­plest the­ory com­pat­i­ble with expe­ri­ence and (2) hang­ing onto it while it remains sim­plest is both nec­es­sary and suf­fi­cient for efficiency.”

    via:cshalizi occam’s-razor sim­plic­ity model-​​discovery expla­na­tion philosophy-​​of-​​science
  • [1206.4599] A Uni­fied Robust Clas­si­fi­ca­tion Model

    “A wide vari­ety of machine learn­ing algo­rithms such as sup­port vec­tor machine (SVM), min­i­max prob­a­bil­ity machine (MPM), and Fisher dis­crim­i­nant analy­sis (FDA), exist for binary clas­si­fi­ca­tion. The pur­pose of this paper is to pro­vide a uni­fied clas­si­fi­ca­tion model that includes the above mod­els through a robust opti­miza­tion approach. This uni­fied model has sev­eral ben­e­fits. One is that the exten­sions and improve­ments intended for SVM become applic­a­ble to MPM and FDA, and vice versa. Another ben­e­fit is to pro­vide the­o­ret­i­cal results to above learn­ing meth­ods at once by deal­ing with the uni­fied model. We give a sta­tis­ti­cal inter­pre­ta­tion of the uni­fied clas­si­fi­ca­tion model and pro­pose a non-​​convex opti­miza­tion algo­rithm that can be applied to non-​​convex vari­ants of exist­ing learn­ing methods.”

    clas­si­fi­ca­tion algo­rithms lumpers-​​and-​​spliters-​​sittin-​​in-​​a-​​tree
  • CUDA Down­loads | NVIDIA Devel­oper Zone

    This release of the CUDA Toolkit  enables devel­op­ment using GPUs using the Kepler archi­tec­ture, such as the GeForce GTX680. Fea­ture and func­tion­al­ity builds on the foun­da­tion of the CUDA 4.1 release which intro­duced: A new  LLVM-​​based CUDA com­piler 1000+ new image pro­cess­ing func­tions Redesigned Visual Pro­filer with auto­mated per­for­mance analy­sis and inte­grated expert guidance

    CUDA GPU pro­gram­ming library MacOS
  • [1206.2057] Fin­ish­ing Flows Quickly with Pre­emp­tive Scheduling

    “Today’s data cen­ters face extreme chal­lenges in pro­vid­ing low latency. How­ever, fair shar­ing, a prin­ci­ple com­monly adopted in cur­rent con­ges­tion con­trol pro­to­cols, is far from opti­mal for sat­is­fy­ing latency require­ments. We pro­pose Pre­emp­tive Dis­trib­uted Quick (PDQ) flow sched­ul­ing, a pro­to­col designed to com­plete flows quickly and meet flow dead­lines. PDQ enables flow pre­emp­tion to approx­i­mate a range of sched­ul­ing dis­ci­plines. For exam­ple, PDQ can emu­late a short­est job first algo­rithm to give pri­or­ity to the short flows by paus­ing the con­tend­ing flows. PDQ bor­rows ideas from cen­tral­ized sched­ul­ing dis­ci­plines and imple­ments them in a fully dis­trib­uted man­ner, mak­ing it scal­able to today’s data cen­ters. Fur­ther, we develop a mul­ti­path ver­sion of PDQ to exploit path diver­sity. Through exten­sive packet-​​level and flow-​​level sim­u­la­tion, we demon­strate that PDQ sig­nif­i­cantly out­per­forms TCP, RCP and D3 in data cen­ter envi­ron­ments. We fur­ther show that PDQ is sta­ble, resilient to packet loss, and pre­serves nearly all its per­for­mance gains even given inac­cu­rate flow information.”

    queuing-​​models engineering-​​design algo­rithms performance-​​measure nudge-​​targets
  • [1206.2216] Com­plex Sys­tems Sci­ence: Dreams of Uni­ver­sal­ity, Real­ity of Interdisciplinarity

    “Using a large data­base (~ 215 000 records) of rel­e­vant arti­cles, we empir­i­cally study the “com­plex sys­tems” field and its claims to find uni­ver­sal prin­ci­ples apply­ing to sys­tems in gen­eral. The study of ref­er­ences shared by the papers allows us to obtain a global point of view on the struc­ture of this highly inter­dis­ci­pli­nary field. We show that its over­all coher­ence does not arise from a uni­ver­sal the­ory but instead from com­pu­ta­tional tech­niques and fruit­ful adap­ta­tions of the idea of self-​​organization to spe­cific sys­tems. We also find that com­mu­ni­ca­tion between dif­fer­ent dis­ci­plines goes through spe­cific “trad­ing zones”, ie sub-​​communities that cre­ate an inter­face around spe­cific tools (a DNA microchip) or con­cepts (a network).”

    via:cshalizi com­plex­ol­ogy pro­fes­sion­al­iza­tion network-​​theory disappointed-​​by-​​lack-​​of-​​Abbott-​​ref citation-​​networks

Items of some interest:

These are my recent Pin​board​.in links:

  • [1006.5366] “Not only defended but also applied”: The per­ceived absur­dity of Bayesian inference

    “The mis­sion­ary zeal of many Bayesians of old has been matched, in the other direc­tion, by a view among some the­o­reti­cians that Bayesian meth­ods are absurd-​​not merely mis­guided but obvi­ously wrong in prin­ci­ple. We con­sider sev­eral exam­ples, begin­ning with Feller’s clas­sic text on prob­a­bil­ity the­ory and con­tin­u­ing with more recent cases such as the per­ceived Bayesian nature of the so-​​called dooms­day argu­ment. We ana­lyze in this note the intel­lec­tual back­ground behind var­i­ous mis­con­cep­tions about Bayesian sta­tis­tics, with­out aim­ing at a com­plete his­tor­i­cal cov­er­age of the rea­sons for this dismissal.”

    social-​​dynamics sta­tis­tics martial-​​arts-​​schools
  • [1206.3268] Fea­ture Selec­tion via Block-​​Regularized Regression

    “In this paper, we con­sid­ered the prob­lem of find­ing a sub­set of covari­ates in a high-​​dimensional space that affect the out­put vari­able when there is a block struc– ture in the covari­ates. In the con­text of asso­ci­a­tion map­ping, we pro­posed a regression-​​based model with a Markov chain prior that encodes the infor­ma­tion in the cor­re­la­tion struc­ture such as dis­tance and re– com­bi­na­tion rate between adja­cent SNP mark­ers. We demon­strated on the sim­u­lated and mouse data that our pro­posed algo­rithm can be used to iden­tify groups of SNP mark­ers as a rel­e­vant block of causal SNPs. The idea of rep­re­sent­ing the cor­re­la­tion struc­ture as a Markov chain in a vari­able selec­tion method to learn grouped rel­e­vant vari­ables can be gen­er­al­ized to use a graph­i­cal model as a prior in a vari­able selec­tion prob– lem to rep­re­sent an arbi­trary cor­re­la­tion struc­ture in vari­ables in a high-​​dimensional space. Another inter– est­ing exten­sion of the model is to model a struc­ture in out­put vari­ables as well when mea­sure­ments of mul– tiple out­put vari­ables are available.”

    sta­tis­tics bioin­for­mat­ics algo­rithms data-​​mining feature-​​extraction
  • Fil­ipe Kiss : A bet­ter git log

    “So, are you tired of this old and bored git log screen?”

    yes software-​​development git tricks-​​n-​​tips bash
  • Neu­roskep­tic: Brains are Dif­fer­ent on Macs

    “The paper goes into lots more detail, but the les­son for researchers is extremely sim­ple: don’t cross the streams of data-​​analysis. Set up your analy­sis stream and then use it on all of your data. Same hard­ware, same soft­ware, same set­tings. Imag­ine you’re doing a study com­par­ing brain struc­ture in two groups. Halfway through ana­lyz­ing your data, you upgrade your MacOS. All of the brains you ana­lyze after that will be, say, 5% “big­ger”. That’ll cer­tainly make your data much nois­ier, and if you hap­pen to ana­lyze most of Group A before Group B, it’ll give you a false pos­i­tive find­ing. Some­times you just can’t avoid changes in hard­ware or soft­ware — IT techs have a habit of upgrad­ing things with­out ask­ing — but in these cases, you should run the same data under the old and the new regime to see if it’s mak­ing a dif­fer­ence. Finally, it would be wrong to blame FreeSurfer for this. I’d be sur­prised if they were any worse than the other soft­ware pack­ages. Mix­ing and match­ing ver­sions is some­thing that the FreeSurfer devel­op­ers specif­i­cally warn against. This paper shows why.”

    data-​​analysis repro­ducibil­ity technical-​​assumptions anomalies-​​are-​​where-​​you-​​find-​​them
  • Plug: What is infer­en­tial­ism? « Odontomachus’s Blog

    “I’ve been crit­i­cal of objects and the idea of ref­er­ence for a while now. To me sen­tences and propo­si­tions, by virtue of their role as “moves” in social inter­ac­tions, are likely to have pri­or­ity in a prop­erly objec­tive account of mean­ing. Many puta­tive objects (e.g. cor­po­ra­tions or muta­ble dig­i­tal doc­u­ments) bor­der on being fic­tional, gain­ing their object­hood only through what we say about them; and many refer­ring phrases seem to refer to dif­fer­ent things, depend­ing on what is being pred­i­cated. I think this opin­ion would make me what Pere­grin calls a “strong infer­en­tial­ist”. Even­tu­ally I hope that think­ing clearly about seman­tics ought to (among other things) help bring calm to the cur­rent mass hys­te­ria which is the Seman­tic Web and Linked Data, and help steer all of that energy expen­di­ture to improve its consequence.”

    prag­ma­tism indirect-​​links phi­los­o­phy talking-​​about-​​thinking-​​and-​​the-​​reverse
  • [1206.3552] A Clas­si­fi­ca­tion for Com­mu­nity Dis­cov­ery Meth­ods in Com­plex Networks

    “In the last few years many real-​​world net­works have been found to show a so-​​called com­mu­nity struc­ture orga­ni­za­tion. Much effort has been devoted in the lit­er­a­ture to develop meth­ods and algo­rithms that can effi­ciently high­light this hid­den struc­ture of the net­work, tra­di­tion­ally by par­ti­tion­ing the graph. Since net­work rep­re­sen­ta­tion can be very com­plex and can con­tain dif­fer­ent vari­ants in the tra­di­tional graph model, each algo­rithm in the lit­er­a­ture focuses on some of these prop­er­ties and estab­lishes, explic­itly or implic­itly, its own def­i­n­i­tion of com­mu­nity. Accord­ing to this def­i­n­i­tion it then extracts the com­mu­ni­ties that are able to reflect only some of the fea­tures of real com­mu­ni­ties. The aim of this sur­vey is to pro­vide a man­ual for the com­mu­nity dis­cov­ery prob­lem. Given a meta def­i­n­i­tion of what a com­mu­nity in a social net­work is, our aim is to orga­nize the main cat­e­gories of com­mu­nity dis­cov­ery based on their own def­i­n­i­tion of com­mu­nity. Given a desired def­i­n­i­tion of com­mu­nity and the fea­tures of a prob­lem (size of net­work, direc­tion of edges, mul­ti­di­men­sion­al­ity, and so on) this review paper is designed to pro­vide a set of approaches that researchers could focus on.”

    via:cshalizi graph-​​theory com­mu­nity clas­si­fi­ca­tion algo­rithms nudge
  • [1205.0792] Exact Wavelets on the Ball

    “We develop an exact wavelet trans­form on the three-​​dimensional ball (i.e. on the solid sphere), which we name the fla­glet trans­form. For this pur­pose we first con­struct an exact har­monic trans­form on the radial line using damped Laguerre poly­no­mi­als and develop a cor­re­spond­ing quad­ra­ture rule. Com­bined with the spher­i­cal har­monic trans­form, this approach leads to a sam­pling the­o­rem on the ball and a novel three-​​dimensional decom­po­si­tion which we call the Fourier-​​Laguerre trans­form. We relate this new trans­form to the well-​​known Fourier-​​Bessel decom­po­si­tion and show that band-​​limitness in the Fourier-​​Laguerre basis is a suf­fi­cient con­di­tion to com­pute the Fourier-​​Bessel decom­po­si­tion exactly. We then con­struct the fla­glet trans­form on the ball through a har­monic tiling, which is exact thanks to the exact­ness of the Fourier-​​Laguerre trans­form (from which the name fla­glets is coined). The cor­re­spond­ing wavelet ker­nels have com­pact local­i­sa­tion prop­er­ties in real and har­monic space and their angu­lar aper­ture is invari­ant under radial trans­la­tion. We intro­duce a mul­tires­o­lu­tion algo­rithm to per­form the fla­glet trans­form rapidly, while cap­tur­ing all infor­ma­tion at each wavelet scale in the min­i­mal num­ber of sam­ples on the ball. Our imple­men­ta­tion of these new tools achieves float­ing point pre­ci­sion and is made pub­licly avail­able. We per­form numer­i­cal exper­i­ments demon­strat­ing the speed and accu­racy of these libraries and illus­trate their capa­bil­i­ties on a sim­ple denois­ing example.”

    wavelets geom­e­try representation-​​theory signal-​​processing answer-​​languages
  • [1205.3077] Efficiency-​​Revenue Trade-​​offs in Auctions

    “When agents with inde­pen­dent pri­ors bid for a sin­gle item, Myerson’s opti­mal auc­tion max­i­mizes expected rev­enue, whereas Vickrey’s second-​​price auc­tion opti­mizes social wel­fare. We address the nat­ural ques­tion of trade-​​offs between the two cri­te­ria, that is, auc­tions that opti­mize, say, rev­enue under the con­straint that the wel­fare is above a given level. If one allows for ran­dom­ized mech­a­nisms, it is easy to see that there are polynomial-​​time mech­a­nisms that achieve any point in the trade-​​off (the Pareto curve) between rev­enue and wel­fare. We inves­ti­gate whether one can achieve the same guar­an­tees using deter­min­is­tic mech­a­nisms. We pro­vide a neg­a­tive answer to this ques­tion by show­ing that this is a (weakly) NP-​​hard prob­lem. On the pos­i­tive side, we pro­vide polynomial-​​time deter­min­is­tic mech­a­nisms that approx­i­mate with arbi­trary pre­ci­sion any point of the trade-​​off between these two fun­da­men­tal objec­tives for the case of two bid­ders, even when the val­u­a­tions are cor­re­lated arbi­trar­ily. The major prob­lem left open by our work is whether there is such an algo­rithm for three or more bid­ders with inde­pen­dent val­u­a­tion distributions.”

    algo­rithms Pareto-​​front performance-​​measure multiobjective-​​optimization
  • Sym­bol­set

    “Sym­bol­sets are seman­tic sym­bol fonts. They work in mod­ern browsers and any­where Open­Type fea­tures are supported.”

    typog­ra­phy uni­code
  • [1204.6653] Elim­i­na­tion of Glass Arti­facts and Object Segmentation

    “Many images nowa­days are cap­tured from behind the glasses and may have cer­tain stains dis­crep­ancy because of glass and must be processed to make dif­fer­en­ti­a­tion between the glass and objects behind it. This research paper pro­poses an algo­rithm to remove the dam­aged or cor­rupted part of the image and make it con­sis­tent with other part of the image and to seg­ment objects behind the glass. The dam­aged part is removed using total vari­a­tion inpaint­ing method and seg­men­ta­tion is done using kmeans clus­ter­ing, anisotropic dif­fu­sion and water­shed trans­for­ma­tion. The final out­put is obtained by inter­po­la­tion. This algo­rithm can be use­ful to appli­ca­tions in which some part of the images are cor­rupted due to data trans­mis­sion or needs to seg­ment objects from an image for fur­ther processing.”

    image-​​segmentation image-​​processing nudge-​​targets algo­rithms
  • The whole of the law — Things from your life

    “But it’ll be your deci­sion, not iner­tia or fate. The ongo­ing cadence of ask­ing these ques­tions (and, maybe, the con­tent of any answers you come up with) will con­vene an open space for you to live in. A world where what­ever you do is right.”

    this
  • The Pirate Uni­ver­sity | Pirate university

    “The Pirate Uni­ver­sity is an on-​​line bul­letin board on which stu­dents post requests for aca­d­e­mic pub­li­ca­tions. You can com­pare it to an aca­d­e­mic wish list. Oth­ers, who know where to find these pub­li­ca­tions, reply and if pos­si­ble, pro­vide links to the resources searched. The Pirate Uni­ver­sity is not pro­vid­ing, stor­ing or shar­ing copy­righted mate­r­ial. An impor­tant ques­tion is if the upload­ing of arti­cles, pub­li­ca­tions is legal. If you are the copy­right holder of the arti­cle requested, there should be no prob­lem. Also in cer­tain cases, if you or your insti­tute have acquired the rights of the pub­li­ca­tion, or if it is free of rights, there shouldn’t be a prob­lem. It is prob­a­bly best to con­sult with your librar­ian to see which kind of pub­li­ca­tion is okay to share on the Internet.”

    academic-​​culture pub­lish­ing col­lab­o­ra­tion crowd­sourc­ing librar­i­ans open-​​access schol­ar­ship
  • [1206.3793] A dis­trib­uted classification/​estimation algo­rithm for sen­sor networks

    “…We pro­pose a novel coop­er­a­tive iter­a­tive algo­rithm which copes with the com­mu­ni­ca­tion con­straints imposed by the net­work and shows remark­able per­for­mance. Our main result is a rig­or­ous proof of the con­ver­gence of the algo­rithm and a char­ac­ter­i­za­tion of the limit behav­ior. We also show that, in the limit when the num­ber of sen­sors goes to infin­ity, the com­mon unknown para­me­ter is esti­mated with arbi­trary small error, while the clas­si­fi­ca­tion error con­verges to that of the opti­mal cen­tral­ized max­i­mum like­li­hood esti­ma­tor. We also show numer­i­cal results that val­i­date the the­o­ret­i­cal analy­sis and sup­port their pos­si­ble gen­er­al­iza­tion. We com­pare our strat­egy with the Expectation-​​Maximization algo­rithm and we dis­cuss trade-​​offs in terms of robust­ness, speed of con­ver­gence and imple­men­ta­tion simplicity.”

    distributed-​​processing collective-​​behavior sensor-​​networks algo­rithms nudge-​​targets
  • [1204.6391] Extend­ing par­tial rep­re­sen­ta­tions of func­tion graphs and per­mu­ta­tion graphs

    “Func­tion graphs are graphs rep­re­sentable by inter­sec­tions of con­tin­u­ous real-​​valued func­tions on the inter­val [0,1] and are known to be exactly the com­ple­ments of com­pa­ra­bil­ity graphs. As such they are rec­og­niz­able in poly­no­mial time. Func­tion graphs gen­er­al­ize per­mu­ta­tion graphs, which arise when all func­tions con­sid­ered are lin­ear. We focus on the prob­lem of extend­ing par­tial rep­re­sen­ta­tions, which gen­er­al­izes the recog­ni­tion prob­lem. We observe that for per­mu­ta­tion graphs an easy exten­sion of Golumbic’s com­pa­ra­bil­ity graph recog­ni­tion algo­rithm can be exploited. This approach fails for func­tion graphs. Nev­er­the­less, we present a polynomial-​​time algo­rithm for extend­ing a par­tial rep­re­sen­ta­tion of a graph by func­tions defined on the entire inter­val [0,1] pro­vided for some of the ver­tices. On the other hand, we show that if a par­tial rep­re­sen­ta­tion con­sists of func­tions defined on subin­ter­vals of [0,1], then the prob­lem of extend­ing this rep­re­sen­ta­tion to func­tions on the entire inter­val [0,1] becomes NP-​​complete.”

    graph-​​theory math-i-didn’t-know representation-​​theory ontol­ogy inter­est­ing
  • [1206.3294] Flex­i­ble Pri­ors for Exemplar-​​based Clustering

    “Exemplar-​​based clus­ter­ing meth­ods have been shown to pro­duce state-​​of-​​the-​​art results on a num­ber of syn­thetic and real-​​world clus­ter­ing prob­lems. They are appeal­ing because they offer com­pu­ta­tional ben­e­fits over latent-​​mean mod­els and can han­dle arbi­trary pair­wise sim­i­lar­ity mea­sures between data points. How­ever, when try­ing to recover under­ly­ing struc­ture in clus­ter­ing prob­lems, tai­lored sim­i­lar­ity mea­sures are often not enough; we also desire con­trol over the dis­tri­b­u­tion of clus­ter sizes. Pri­ors such as Dirich­let process pri­ors allow the num­ber of clus­ters to be unspec­i­fied while express­ing pri­ors over data par­ti­tions. To our knowl­edge, they have not been applied to exemplar-​​based mod­els. We show how to incor­po­rate pri­ors, includ­ing Dirich­let process pri­ors, into the recently intro­duced affin­ity prop­a­ga­tion algo­rithm. We develop an effi­cient max­prod­uct belief prop­a­ga­tion algo­rithm for our new model and demon­strate exper­i­men­tally how the expanded range of clus­ter­ing pri­ors allows us to bet­ter recover true clus­ter­ings in sit­u­a­tions where we have some infor­ma­tion about the gen­er­at­ing process.”

    clus­ter­ing algo­rithms
  • Mag­a­zine — The Case Against Cre­den­tial­ism — The Atlantic

    ’”ALL OF OUR WORK HAS GIVEN ME A VERY STRONG view,” Richard Boy­atzis told me one after­noon. The con­sult­ing firm Boy­atzis heads, McBer and Com­pany, was founded by David McClel­land in 1963. Its spe­cialty has been ana­lyz­ing what peo­ple actu­ally do in busi­ness jobs—not what their job descrip­tions say, but how they spend their time and which skills seem most impor­tant to their suc­cess. “I’ve come to see that when­ever a group insti­tutes a cre­den­tial­ing process, whether by licens­ing or insist­ing on advanced degrees, the espoused rhetoric is to enforce the stan­dards of pro­fes­sion­al­ism. This is true whether it’s among accoun­tants or plumbers or physi­cians. But the observed con­se­quences always seem to be these two: the exclu­sion of cer­tain groups, whether by inten­tion or not, and the estab­lish­ment of mediocre per­for­mance standards.“‘

    pro­fes­sion­al­iza­tion cre­den­tial­ing Andrew-​​Abbott-​​smiles-​​in-​​Chicago author­ity exper­tise cultural-​​assumptions disintermediation-​​targets
  • [1205.2483] Edge-​​clique graphs of cock­tail par­ties have unbounded rankwidth

    “In an attempt to find a polynomial-​​time algo­rithm for the edge-​​clique cover prob­lem on cographs we tried to prove that the edge-​​clique graphs of cographs have bounded rankwidth. How­ever, this is not the case. In this note we show that the edge-​​clique graphs of cock­tail party graphs have unbounded rank width.”

    open-​​questions nudge-​​targets graph-​​theory algo­rithms
  • [1206.3235] Iden­ti­fy­ing rea­son­ing pat­terns in games

    “We present an algo­rithm that iden­ti­fies the rea­son­ing pat­terns of agents in a game, by iter­a­tively exam­in­ing the graph struc­ture of its Multi-​​Agent Influ­ence Dia­gram (MAID) rep­re­sen­ta­tion. If the deci­sion of an agent par­tic­i­pates in no rea­son­ing pat­terns, then we can effec­tively ignore that deci­sion for the pur­pose of cal­cu­lat­ing a Nash equi­lib­rium for the game. In some cases, this can lead to expo­nen­tial time sav­ings in the process of equi­lib­rium cal­cu­la­tion. More­over, our algo­rithm can be used to enu­mer­ate the rea­son­ing pat­terns in a game, which can be use­ful for con­struct­ing more effec­tive com­put­er­ized agents inter­act­ing with humans.”

    game-​​theory infer­ence strat­egy nudge-​​targets learning-​​by-​​watching