“… As a depar­ture from such tra­di­tional trust mod­els, we pro­pose a generic, machine learn­ing approach based trust frame­work where an agent uses its own pre­vi­ous trans­ac­tions (with other agents) to build a knowl­edge base, and uti­lize this to assess the trust­wor­thi­ness of a trans­ac­tion based on asso­ci­ated fea­tures, which are capa­ble of dis­tin­guish­ing suc­cess­ful trans­ac­tions from unsuc­cess­ful ones. These fea­tures are har­nessed using appro­pri­ate machine learn­ing algo­rithms to extract rela­tion­ships between the poten­tial trans­ac­tion and pre­vi­ous transactions.…”

“We study the 2-​​dimensional vec­tor pack­ing prob­lem, which is a gen­er­al­iza­tion of the clas­si­cal bin pack­ing prob­lem where each item has 2 dis­tinct weights and each bin has 2 cor­re­spond­ing capac­i­ties. The goal is to group items into min­i­mum num­ber of bins, with­out vio­lat­ing the bin capac­ity constraints.…”

“Thin­ning of char­ac­ter images is a big chal­lenge. Removal of strokes or defor­mi­ties in thin­ning is a dif­fi­cult problem.…”

“… We exam­ine here how the algo­rithm of phyl­lotaxis could con­tribute to the analy­sis of the struc­ture of col­la­gen fib­rils. Such an algo­rithm indeed leads to orga­ni­za­tions giv­ing to each ele­ment of the assem­bly the most homo­ge­neous and isotropic dense envi­ron­ment in a sit­u­a­tion of cylin­dri­cal sym­me­try. The scat­tered inten­sity expected from a phyl­lotac­tic dis­tri­b­u­tion of triple helices in col­la­gen fib­rils well agrees with the major fea­tures observed along the equa­to­r­ial direc­tion of their X ray pat­terns. Fol­low­ing this approach, the aggre­ga­tion of triple helices in fib­rils should be con­sid­ered within the frame of soft con­densed mat­ter stud­ies rather than that of mol­e­c­u­lar crys­tal studies.”

“Even when the num­bers of non-​​zero entries per column/​row in the mea­sure­ment matri­ces are lim­ited to $O(1)$, numer­i­cal exper­i­ments indi­cate that the algo­rithm can still typ­i­cally recover the orig­i­nal sig­nal per­fectly with an $O(N)$ com­pu­ta­tional cost per update as well if the den­sity $\rho$ of non-​​zero entries of the sig­nal is lower than a cer­tain crit­i­cal value $\rho_{\rm th}(\alpha)$ as $N,M \to \infty$.”

“We inves­ti­gate the evo­lu­tion­ary dynam­ics of an ide­alised model for the robust self-​​assembly of two-​​dimensional struc­tures called poly­omi­noes. The model includes rules that encode inter­ac­tions between sets of square tiles that drive the self-​​assembly process. The rela­tion­ship between the model’s rule set and its result­ing self-​​assembled struc­ture can be viewed as a genotype-​​phenotype map and incor­po­rated into a genetic algorithm.”