Count-Min Sketches for Estimating Password Frequency within Hamming Distance Two

Size: px
Start display at page:

Download "Count-Min Sketches for Estimating Password Frequency within Hamming Distance Two"

Transcription

1 Count-Min Sketches for Estimating Password Frequency within Hamming Distance Two Leah South and Dougas Stebia Schoo of Mathematica Sciences, Queensand University of Technoogy, Brisbane, Queensand, Austraia Abstract. The count-min sketch is a usefu data structure for recording and estimating the frequency of string occurrences, such as passwords, in sub-inear space with high accuracy. However, it cannot be used to draw concusions on groups of strings that are simiar, for exampe cose in Hamming distance. This paper introduces a variant of the count-min sketch which aows for estimating counts within a specified Hamming distance of the queried string. This variant can be used to prevent users from choosing popuar passwords, ike the origina sketch, but it aso aows for a more efficient method of anaysing password statistics. Keywords: count-min sketch, Boom fiter, password frequency, approximate string matching 1 Introduction The use of passwords for identity verification is widespread. There is a ong ine of research on anayzing the security and guessabiity of password [8, 3]. In arge onine systems that we see on the Internet today, an important characteristic that affects the overa security of the system is that passwords within the system shoud not be too popuar. In an idea setting, of course, users woud create a unique password that is hard to guess, and not popuar, so that ony that user and no one ese coud access their account. However, users tend to choose passwords that are easy to remember and famiiar to them, such as dictionary words, or perhaps strings associated with the system in question. This tendency means that certain passwords are used with higher frequency, making them popuar. If an attacker knew the distribution of passwords, they coud use its statistics and guess the most popuar passwords first. This is known as the statistica guessing technique [9]. When there is a high percentage of popuar passwords, the attacker can compromise a high percentage of accounts. For exampe, the 009 breach of RockYou.com s 3 miion account password database showed that the most popuar password (13456) was used by 0.9% of a accounts, and the next 4 most popuar passwords (1345, , password, and ioveyou) were used by another 0.8% of a accounts. Ceary, the system was not screening passwords for popuarity. As a resut, a statistica guessing attack woud ead to miions of accounts being compromised. In order to prevent a successfu statistica guessing

2 Leah South and Dougas Stebia attack such as this, it is common to imit the number of guesses; more recenty, it has been proposed [9] to imit the popuarity of passwords: when users try to set their password to a string that is used in a percentage of accounts above some threshod, it is rejected and the user is required to choose another password. In order to keep track of password popuarity, some sort of system which counts passwords must be used. Onine sites with a arge number of users are best suited for systems which restrict popuar passwords, as such sites are at high risk of trawing attacks, in which attackers aim to guess the passwords to many accounts without targeting any singe account. How can we store password information in a way that aows us to cacuate frequency when users attempt to register a password? The simpest technique for cacuating frequency during password registration woud be to store a separate tabe of passwords aong with their frequency. This is undesirabe both for efficiency reasons (since the size of the tabe grows ineary in the number of distinct passwords) and for security reasons (since it immediatey provides an attacker with the fu distribution of passwords). Best-practice recommendations for storing passwords for ogin invove storing sated password hashes for each account; the set of such passwords does not admit statistica anaysis since, by design, the hash of the same password under different sats yieds different, seemingy independent, outputs, thus yieding no information about the frequency with which a password is used. 1.1 Boom Fiters and Count-Min Sketches The Boom fiter [] can be used [10] to store in sub-inear space a tabe representing a dictionary of prohibited passwords. The system is setup as foows. A w h tabe T of bits is used, aong with h independent hash functions hash 1,..., hash h : {A Z, a z, 0 9,... } {1,..., w}. Each word x in the dictionary of prohibited passwords is hashed under each hash function hash k, and the entry T hashk (x),k is set to 1. When a proposed password y is to be tested for membership in the ist of prohibited passwords, if a of the vaues T hashk (y),k are equa to 1, then y is deemed to be prohibited, but if at east one of those tabe entries is zero, then y is not prohibited. There are no fase negatives, meaning that it is impossibe for a password that is prohibited to not be recognized as such, but there may be fase positives, meaning that some passwords that are not prohibited may, due to coisions on a rows, sti be identified by the tabe as prohibited. Assuming the hash functions are independent random functions, the fase positive rate (1 (1 h w )N ) h, where N is the number of prohibited passwords originay added to the tabe [10]. The count-min sketch [5] enhances the Boom fiter by storing a tabe of integers, not bits. The update(x, c) function records that string x has been used c more times by adding c to each tabe entry T hashk (x),k. The estimate(x) function returns an estimate on the number of times that string x has been used by computing min{t hashk (x),k : 1 k h}. The use of count-min sketches for recording password frequency was proposed by Schechter, Herey, and Mitzenmacher [9], who propose preventing users from registering with passwords whose

3 Count-Min Sketches for Estimating Password Frequency within Distance 3 current popuarity is above a certain threshod. As with the Boom fiter, fase negatives cannot occur, meaning that for any password x it is impossibe for the estimate(x) to return a vaue ower than the sum of the c over a cas of update(x, c). However, fase positives may sti occur, meaning that estimate(x) may over-estimate the frequency of password x, due to coisions across a hash functions. For a singe hash, the expected error due to coisions is N/w, where N is the tota sum of the counts of a passwords; this is because the tota count of N wi spread approximatey eveny across a w coumns in the tabe. By the Markov inequaity, the error of the count min-sketch (the minimum across a hash functions) is at most N/w with probabiity at east 1 ( 1 )h. 1. Contributions When used for passwords, the Boom fiter and the count-min sketch can be usefu in prohibiting certain passwords or imiting the frequency with which any password is registered. However, they cannot accommodate any reation between simiar passwords. For exampe, a user who tries to register the password password and finds that it is prohibited may try again with p@ssword or passw rd. An attacker, knowing this prevaence for eetspeak, may aso make use of these simiarities in an attack strategy that targets amost popuar passwords. Since the Boom fiter and count-min sketch use independent random passwords, they ose any semantic connection between such simiar strings. In this work, we expore a variant of the count-min sketch that aows one to compute estimates not ony for the given string but for a strings within a fixed Hamming distance of the given string. This technique aows a system to detect frequenty used passwords at the registration phase. Our technica approach is to introduce widcard characters and then record and estimate based on a strings that can be constructed with widcard characters within the required Hamming distance. This technique imposes a computationa overhead of ( d), where is the ength of the password in question and d is the maximum Hamming distance. This compares favouraby with the naive technique which woud have a computationa overhead of ( α d ), where α is the size of the aphabet. We provide fase positive error rates as we, and compared with the naive method our technique provides better accuracy for the same size tabe for a wide range of password engths. Our technique can naturay be extended to higher Hamming distances. Reated Work Whie the Boom fiter and count-min sketch are widey used, there are severa other methods to store estimate counts, some of which may be usefu in obtaining statistics on groups of simiar passwords. There have been some systems in the past which have stored ceartext passwords aongside their counts. This method is mosty outdated due to the ack of security it provides. If attackers gain access to password databases such as these, a highy successfu statistica guessing attack can be carried out because the

4 4 Leah South and Dougas Stebia actua passwords are visibe and there are no fase positives - the exact counts are known. Using hash functions is preferabe to this method. Decision trees, as suggested by Bergadano, Crispo and Ruffo, can be used to determine password membership. These decision trees consist of nodes for particuar attributes, arcs for vaues of these attributes and eaves for cassification of whether or not the password has been used before in that database [1]. When using decision trees, there is a training phase in which the nodes, arcs and eaves are chosen to produce the best resuts. For each update to the database, the training phase must be repeated. Due to this, decision trees can be used for determining membership but are not as practica when frequent updates may be invoved. There has aso been some work on the topic of determining popuarity of simiar words. This incudes work which aows for checking membership of words within Levenshtein distance 1, that is words which have ony one insertion, deetion or substitution. Manber and Wu [7] suggested an approach simiar to the origina Boom fiter, with the main difference being that the membership of words within Levenshtein distance 1 are checked in the estimation stage. It is proposed that if any password within distance 1 appears to be positive, the password cannot be used [7]. As with a Boom fiter-based techniques, this ony records the binary data of whether two simiar passwords have been used, not counts on how frequenty simiar passwords are used. 3 Construction The adaptation of the origina count-min sketch that is introduced in this paper aows for the popuarity estimations of words within Hamming distance zero to two, that is words that differ by up to two substitutions. Like the count-min sketch, this adaptation consists of three main parts: hashing the passwords, updating the tabe and estimating the counts. In Appendix A, we provide a worked exampe of each of the three operations hashing, updating, and estimating. 3.1 Hash We represent passwords x as integers X. Any canonica representation of a word as an integer wi suffice. For expository convenience, we can imagine a mapping of an -etter word x = x 1... x where each character x i is coded as a two-digit integer X i {00,..., 98}; the integer 99 is reserved to represent a widcard character *. The vector X 1,..., X is then viewed as an integer X = i=1 100i 1 X i. This suffices to encode for exampe a passwords that can be typed using characters on a standard US keyboard. As in [6], each hash function hash k is a Carter Wegman -universa hash function [4] of the form hash k (x) = (((a k X + b k ) mod p k ) mod w) + 1,

5 Count-Min Sketches for Estimating Password Frequency within Distance 5 where X is the canonica integer representation of the word x, p k is a prime number much arger than w (reca w is the width of the tabe), say p k = 31 1 or p i = 61 1, and a k and b k are chosen uniformy at random from {0,..., p k 1}. We can simpify to using the same prime p = p 1 = = p h across a hash functions, but a a k and b k need to be chosen independenty to ensure negigibe probabiity of coisions; otherwise the benefits of using mutipe hash functions wi be eiminated. We define the vector-wise hash function hash(x) where the kth entry of hash(x) consists of hash k (x) with a k and b k. 3. Update The function update(t, x, a, b, c) T updates the sketch based on the previous tabe T, the password x to be updated, the vaues or vectors of a and b for the hash functions and the amount of times c that the password is being added. The update agorithm in this work differs greaty to that in the traditiona count-min sketch. Firsty, the count for the actua password is updated. Like the origina countmin sketch, this is done by finding the hash of the password for each hash function then updating the count at these positions in the tabe. Next, the updates are done for words within distance 1. To do this, a widcard character is introduced. This widcard character, which for our canonica encoding above is denoted by 99, is used to group simiar passwords together: for exampe if the word abcd, or a, b, c, d, is represented in integer form as 1,, 3, 4, then 99,, 3, 4 represents *bcd, a four etter words ending in bcd. A widcard character indicates that any character coud go in its pace. After the count of the actua password is updated, a widcard passwords within Hamming distance 1 are updated by creating = ength(x) variations of the password, each with a different character from the initia word repaced with 99, then hashing these variations and updating their positions in the tabe. Next, the widcard words within Hamming distance of the password x are hashed and their positions in the tabe are updated. In Agorithm 1, update, using the widcard character means that additiona updates are done for adding widcard words within distance 1. The aternative, naive technique, woud be to search a α passwords within distance 1 during the estimate agorithm, where α is the size of the aphabet. When the password ength α, the proposed technique is much more efficient than the naive technique. 3.3 Estimate The function estimate(t, x, a, b) is used to obtain estimates of the count of passwords at exacty Hamming distances 0, 1, and of x and is specified in fu in Agorithm. 1 1 Note that we have estimate return estimates for counts of passwords at exacty, rather than within, the specified Hamming distance for carity of exposition; estimates

6 6 Leah South and Dougas Stebia Agorithm 1 update(t, x, a, b, c) 1: T T : H hash(x, a, b) 3: for k = 1 to h do 4: T H k,k T H k,k + c 5: for i = 1 to = ength(x) do 6: x x 7: x i 99 8: H hash(x, a, b) 9: for k = 1 to h do 10: T H k,k T H k,k + c 11: for i = 1 to 1 do 1: for j = i + 1 to do 13: x x 14: x i 99 15: x j 99 16: H hash(x, a, b) 17: for k = 1 to h do 18: T H k,k H k,k 19: return T At distance 0. When the specified Hamming distance d is zero, that is, when the count of the actua word is desired, the estimate process is very simiar to the origina count-min sketch. The h different hashes of the password and the counts at a h positions in the tabe are found. These counts can then be put into a vector est 0 of ength h, where each eement represents the count at a different hash function. The resuting estimate est 0 is the minimum of these counts. At distance 1. At a maximum Hamming distance of one, the process is sighty more compex. First, estimates est 1 i, i = 1,...,, for the frequency of words that may differ from x in the ith character; each of these estimates can be found using the technique estimating the (distance 0) occurrences of the widcard word. In other words, est 1 i is the (distance 0) estimate for the widcard word x where the ith character of x has been repaced with the specia widcard character. Once estimates est 1 i for the frequency of words that may differ from x in the ith character have been found, we can find an estimate est 1 for the number of occurrences of words at distance 1 from x by summing est 1 i for a i, then subtract est 0. The reason for the subtraction is as foows. As expained in the update section, when a singe password x is added into the database, the counts for the exact password and the counts for a = ength(x) passwords within distance 1 are a increased. In the estimation stage, the counts for a passwords with one widcard character, or within distance 1, are summed together. This means that the count within the specified distance can be found by summing the estimates exacty at a distances ess than or equa to.

7 Count-Min Sketches for Estimating Password Frequency within Distance 7 for the exact password has been incuded times in the cacuation, when it shoud ony have been incuded once. To overcome this probem, the count of that exact password mutipied by the ength of the password is subtracted. At distance. The frequency estimate est for passwords at distance is found in a simiar manner to that of distance 1. For each of the ( widcard passwords of x that may differ in the ith and jth characters, we compute (distance 0) estimates est i,j of the widcard word x where the ith and jth characters of x have been repaced with the specia widcard character. We then sum these estimates of words within distance. However, again we have overcounted. The words at precisey distance 1 have been counted 1 times. Simiary, the word at precisey distance 0 has been counted ( times. Agorithm estimate(t, x, a, b) 1: H hash(x, a, b) : for k = 1 to h do 3: est 0 k T Hk,k 4: est 0 min k {est 0 k} 5: for i = 1 to = ength(x) do 6: x x 7: x i 99 8: H hash(x, a, b) 9: for k = 1 to h do 10: est 1 i,k T H k,k 11: est 1 i min k {est 1 i,k} 1: est 1 ( ) i est1 0 i est 13: for i = 1 to 1 do 14: for j = i + 1 to do 15: x x 16: x i 99 17: x j 99 18: H hash(x, a, b) 19: for k = 1 to h do 0: 1: est i,j,k T H k,k est i,j min k {est i,j,k} ( : est i,j est i,j ( 1) est 1 ( 3: return est 0, est 1, est ) ) est 0

8 8 Leah South and Dougas Stebia 4 Anaysis of Construction 4.1 Error Like the origina count-min sketch, coisions can occur in this adaptation. These coisions resut in some error, causing either sight overestimations of counts or fase positives. The error in estimating exact words can be expressed fairy simpy. In determining this error reative to the origina sketch, it is important to note that the tota count for this method is not the same as what it woud be in the origina count-min sketch. In this adaptation of the sketch, the tota count across the sketch is (1 + + ( ) times arger than the actua tota of passwords used. This additiona count is the resut of the update stage, in which the password itsef, a passwords within distance 1 and a ( combinations of passwords within distance are added into the sketch. Let ˆN = (1 + + ( )N denote the tota count on the new sketch, where N is the tota number of (non-distinct) passwords entered. In genera, hash outputs are expected to spread approximatey eveny across a coumns of the tabe, making the average count per hash the quotient of the tota sum of counts and the width w of the tabe: ˆN/w. The expected maximum error is therefore ˆN/w, which may occur, for exampe, if an estimate is done on a password which is not present but happens to have the same hash across a h hash functions as other passwords which have a count of ˆN/w. The expected maximum error for estimating counts of passwords at distance 1 can be derived from the formua for distance 1 in Agorithm : ( ) error(est 1 ) error est 1 i + error( est 0 ). i Since each term of i est1 i is cacuated as distance-0 estimate, the error of each term is the same as the error in a distance-0 estimate, and thus error(est 1 ) error(est 0 ) + error( est 0 ) error(est 0 ) = ˆN w. For passwords at distance, there is another increase in error. Based on the estimate for words at distance in Agorithm, the expected maximum error is error(est ) error i,j est i,j + error(( 1) est 1 ) + error (( ) ) est 0 Since each term of i,j est i,j is cacuated as distance-0 estimate, the error of each term is the same as the error in a distance-0 estimate, and thus ( ) (( ) ) error(est ) error(est 0 ) + error(( 1) est 1 ) + error est 0..

9 Count-Min Sketches for Estimating Password Frequency within Distance 9 Simpifying and substituting, we get ( ) ( ) error(est ) error(est 0 ) + ( 1) error(est 1 ) + error(est 0 ) (( ) ( )) + ( 1)() + error(est 0 ) ( ( ) ) = + ( 1) error(est 0 ) ( ( ) ( )) = + 4 error(est 0 ) ( ) ( ) = 6 error(est 0 ˆN ) = 6 w. 4. Comparison with Naive Method The purpose of the adaptation is to provide a more effective method of estimating counts of passwords within specified Hamming distances. Whie the count-min sketch can be used to do this, this naive method using the origina sketch incrementing the entry for every one of the α ( d) words within distance d in an aphabet of size α woud be ess efficient and have a higher error rate. In this section, a comparison of efficiency and error is carried out. A of the foowing graphs have the same response variabe: the expected average error as a mutipe of N w where N is the tota number of passwords entered (aso the tota count in the origina sketch) and w is the width of the tabe. Estimates at distance 0. When estimating how many times a specific word appears in a database (i.e. estimating passwords at distance zero), the origina count-min sketch is more effective. The expected maximum error in estimating counts is N ˆN w in the origina method and w, or (1 + + ( )N/w, in this variation. Therefore the expected maximum error for estimating counts at distance zero wi aways be ( times arger in this variation. However, this increase in error is not a major probem since this sketch was not designed for estimating exact word counts. Estimates at distance 1. The benefits of using this modified sketch are more evident when estimating counts at distance 1 of a password. In order to estimate counts at distance 1 of a specified password using the origina sketch, mutipe estimates woud have to be done. More specificay, the number of estimates woud be α, where α is the size of the aphabet and is the ength of the password. Since the error in estimating each individua password is N w, the expected maximum error in estimating α passwords is αn/w. With the new method, the expected maximum error in estimating words at distance 1 is ˆN/w = (1 + + ( )N/w. A comparison of these errors can be seen in Figure 1, where the size of the aphabet has been set to α = 7 (6 uppercase and 6 owercase etters, 10

10 10 Leah South and Dougas Stebia numbers, and 10 specia characters). From this graph, it can be seen that the error for the proposed method is ower unti the ength of the password reaches nine characters. Error factor 1, naive scheme proposed scheme Password ength Fig. 1. Error as a mutipe of N/w for passwords at distance 1, using a count-min sketch supporting maximum distance of, compared with naive scheme. Note that this cacuation above assumes that the update stage is as described in Section 3., aowing for estimates within distance, even though we are currenty estimating distance 1. If the sketch is not going to be used to estimate counts at distance at a, then it is possibe to remove the distance section. Updates for passwords at distance woud then not be incuded, making the expected maximum error for words at distance 1 equa to (1 + )N/w. A graph comparing the expected maximum error using the traditiona method and this variation in which ony distance 1 is incuded can be seen beow, in Figure. For this graph, the size of the aphabet is again α = 7. The proposed method has a ower error than the naive method unti the ength of the password reaches 36 characters. Estimates at distance. For passwords at distance, ( α estimates woud have to be done using the origina sketch. This makes the error for the origina sketch α ( ( N/w whereas our proposed technique has an error of 6 ( (1 + + )N/w. When the size of the aphabet is α = 7, the resuting differences in error can be seen in Figure 3: the error when using our proposed scheme is better than the naive method when the ength of the word is ess than 4 characters. 5 Exampe Parameter Instantiation By using the error estimates found in previous sections, it is possibe to estimate the size of the tabe required in order to obtain certain eves of accuracy. To do this, the Markov inequaity can be appied to a expected maximum errors. The resuts are as foows:

11 Count-Min Sketches for Estimating Password Frequency within Distance 11 Error factor,000 1,000 naive scheme proposed scheme Password ength Fig.. Error as a mutipe of N/w for passwords at distance 1, using a count-min sketch supporting maximum distance of 1, compared with naive scheme naive scheme proposed scheme Error factor Password ength Fig. 3. Error as a mutipe of N/w for passwords at distance, compared with naive scheme.

12 1 Leah South and Dougas Stebia When estimating the number of times a specific word appears, there is an error of at most (1 + + ( ) )N/w with probabiity of at east 1 ( 1 )h, where h is the number of hash functions. There is an error of at most 4(1 + + ( )N/w with probabiity of at east 1 ( 1 )h when estimating words within distance 1. There is an error of at most 1 ( ( (1 + + )N/w with probabiity of at east 1 ( 1 )h when estimating words within distance. The foowing exact parameter cacuations are done for distances one and two, but not for Hamming distance zero because, if the user desires a certain error rate for estimating exact words, it is preferabe to use the origina count-min sketch. If the purpose of this sketch was to estimate passwords within distance 1 with an error of at most 1% with probabiity of at east 99.9%, then the width w of the tabe and the number h of hash functions woud have to be as foows: 4(1 + + ( ) = 1 w ( ) h 1 = = w = = h 6.64 ( The number of hash functions woud have to be 7 or more and the width of the tabe woud depend on the ength of the password. If the ength of the passwords is 6 characters, then the width of the tabe woud have to be at east 48,400. This size woud be smaer if the tabe ony incuded passwords within distance zero to one, as suggested previousy. Simiary, if the sketch was needed to estimate passwords within distance with an error of at most 1% with probabiity of at east 99.9%, then the number of hash functions woud sti be at east 7 but the width of the tabe woud be: 1 ( ( (1 + + ) = 1 w 100 ( ( ) ( = w = For 6-character passwords, this woud make the width 338,800. Whie this width seems arge, it is possibe that the fase positive (or error) rate coud be higher when estimating words within specified distances. )) ( )) 6 Concusion When no restrictions are paced on passwords choices, users tend to choose popuar passwords. This eaves systems vunerabe to statistica guessing attacks. By imiting the percentage of popuar passwords, these attacks are not as successfu. In order to do this, efficient toos must be avaiabe to track password usage. The count-min sketch can be used to estimate the counts of password usage within a system. However, the count-min sketch is not as effective when estimating the

13 Count-Min Sketches for Estimating Password Frequency within Distance 13 counts of passwords within specified Hamming distances. In this paper, we have proposed a variant of the count-min sketch using widcard characters that can be used to cacuate estimates of words that are cose in Hamming distance. As with the origina count-min sketch, there wi never be fase negatives where the estimate agorithm under-reports usage of the password but there may be fase positives meaning the estimate agorithm may over-report usage of the password due to coisions. We have cacuated the error rate for estimation as the Hamming distance increases, which aows for cacuation of sketch size for a given error rate. For a reasonabe aphabet size, the error rates in our proposed method are ower than they woud be using the naive approach on a standard count-min sketch for a wide range of password engths, up to 36- and 4-character passwords when estimating passwords within distances 1 and respective. Our technique is most suited when a passwords in the database have the same ength. In future work, it may be desirabe to deveop another variation in which edit distance, such as Levenshtein distance, is used, rather than Hamming distance, to take into account passwords that are cose due to deetions or insertions of characters. Other future work may be to consider the impact of using fractiona contributions for passwords within a certain distance, where the fraction is either a function of the Hamming distance divided by the password ength, or based on some mode of textua simiarity: for exampe, in eetspeak, 5 is a common substitution for s. References 1. F. Bergadano, B. Crispo, G. Ruffo. Proactive password checking with decision trees. Computer and Communications Security, 4(1):67-77, B. H. Boom. Space/time trade-offs in hash coding with aowabe errors. Communications of the ACM, 13(7):4 46, Juy J. Bonneau. The science of guessing: anayzing an anonymized corpus of 70 miion passwords. In Proc. 01 IEEE Symposium on Security and Privacy (S&P), J. Lawrence Carter and M.N. Wegman. Universa casses of hash functions. Journa of Computer and System Sciences, 18(: , G. Cormode, S. Muthukrishnan. An improved data stream summary: the count-min sketch and its appications. Journa of Agorithms, 55(1):58-75, Apri G. Cormode, S. Muthukrishnan. Approximating data with the count-min sketch. IEEE Software, 9(1):64 69, January/February U. Manber, S. Wu. An agorithm for approximate membership checking with appication to password security. Information Processing Letters, 50(1): , J.O. Piam. On the incomparabiity of entropy and margina guesswork in brute-force attacks. In Proc. INDOCRYPT 000, LNCS, vo. 1977, pp S. Schechter, C. Herey, M. Mitzenmacher. Popuarity is everything: A new approach to protecting passwords from statistica guessing attacks. In Proc. 5th USENIX Conference on Hot Topics in Security (HotSec), E. Spafford. Preventing weak password choices. Purdue University Computer Science Technica Reports, paper 875, report number 91-08,

14 14 Leah South and Dougas Stebia A Exampe Count-Min Sketch Cacuation Fix the tabe T to be of width w = 101 and height h =. A.1 Hash First we show how the hash vaues can be cacuated for a singe password, say abcd, under two hash functions. We encode each character as two-digit integer, say abcd x = 1,, 3, 4, then find the integer representation, X = Set a common prime p = 3571, and for each of the two hash functions hash k, choose the parameters a k and b k at random moduo p; for exampe, a = [1151, 111] and b = [941, 1433]. The hashes of abcd are as foows: hash 1 (X) = (( ) mod 3571) mod 101 = 0 hash (X) = (( ) mod 3571) mod 101 = 83. The vector-wise hash is thus hash(abcd, a, b) = 0, 83. A. Update We now show how to update the tabe T to record the use of a password, first updating just the entries for the password itsef, then the entries for passwords within Hamming distance. Suppose the tabe T is currenty as foows: coumn row 1 row Update at distance 0. Now, suppose we ca update(t, x = abcd, a, b, c = 1) T, which is meant to increment (since c = 1) the use of the password abcd. Assume a and b are as in the previous subsection. Then we have that hash(abcd) = 0, 83. Thus, we increment the 0th entry of row 1 and the 83rd entry of row, to obtain T : coumn row 1 row

15 Count-Min Sketches for Estimating Password Frequency within Distance 15 Update at distance 1. Next, we increment a four widcard passwords within distance 1 of abcd, namey *bcd, a*cd, ab*d, and abc*. This is done by computing the corresponding hashes hash(*bcd) = 38, 17 hash(a*cd) = 37, 47 hash(ab*d) = 37, 63 hash(abc*) = 78, 1 and updating a the corresponding entries of the tabe accordingy. Notice that a few partia coisions occur, for exampe, both ab*d and a*cd coide under hash 1, but fortunatey do not coide under hash. In our tabe extract, we ony see a few of the 8 updates since not a coumns are shown (though we imagine a the updates are appied): coumn row 1 row Update at distance. Finay, we increment a ( 4 = 6 widcard passwords within distance of abcd; namey hash(**cd) = 55, 8 hash(*b*d) = 55, 33 hash(*bc*) = 31, 36 hash(a**d) = 18, 7 hash(a*c*) = 95, 66 hash(ab**) = 95, 8 In our tabe extract, we again ony see a few of the 1 updates: coumn row 1 row A.3 Estimate Suppose we now estimate(t, x = bbcd, a, b) to obtain the estimate est 1 of the frequency of passwords at distance 1 of bbcd. Estimate at distance 0. First, we need to compute an estimate est 0 for the number of times bbcd itsef has been used. We do this by computing hash k (bbcd) and retrieving the corresponding ce from row i, then taking the minimum. In this case, hash(bbcd) = 76, 53. The 76th entry in the first row and the 53rd entry in the second row are both 0, so this yieds a (correct) estimate that the exact password bbcd has been seen 0 times before.

16 16 Leah South and Dougas Stebia Estimate at distance 1. To compute an estimate est 1 for the number of times any password at distance 1 of bbcd has been used, we first need to compute estimates for the number of times each of the four widcard passwords *bcd, b*cd, bb*d, and bbc* has been used before. hash(*bcd) = 38, 17 = est 1 1 = min{1, 1} = 1 hash(b*cd) = 100, 8 = est 1 = min{0, 0} = 0 hash(bb*d) = 100, 60 = est 1 3 = min{0, 0} = 0 hash(bbc*) = 76, 63 = est 1 4 = min{0, 1} = 0 Then, we sum these vaues and subtract the estimate for the number of times the string bbcd itsef was used: ( ) est 1 = est 1 i est 0 = 1. i

MARKOV CHAINS AND MARKOV DECISION THEORY. Contents

MARKOV CHAINS AND MARKOV DECISION THEORY. Contents MARKOV CHAINS AND MARKOV DECISION THEORY ARINDRIMA DATTA Abstract. In this paper, we begin with a forma introduction to probabiity and expain the concept of random variabes and stochastic processes. After

More information

A Brief Introduction to Markov Chains and Hidden Markov Models

A Brief Introduction to Markov Chains and Hidden Markov Models A Brief Introduction to Markov Chains and Hidden Markov Modes Aen B MacKenzie Notes for December 1, 3, &8, 2015 Discrete-Time Markov Chains You may reca that when we first introduced random processes,

More information

CS229 Lecture notes. Andrew Ng

CS229 Lecture notes. Andrew Ng CS229 Lecture notes Andrew Ng Part IX The EM agorithm In the previous set of notes, we taked about the EM agorithm as appied to fitting a mixture of Gaussians. In this set of notes, we give a broader view

More information

Cryptanalysis of PKP: A New Approach

Cryptanalysis of PKP: A New Approach Cryptanaysis of PKP: A New Approach Éiane Jaumes and Antoine Joux DCSSI 18, rue du Dr. Zamenhoff F-92131 Issy-es-Mx Cedex France eiane.jaumes@wanadoo.fr Antoine.Joux@ens.fr Abstract. Quite recenty, in

More information

Problem set 6 The Perron Frobenius theorem.

Problem set 6 The Perron Frobenius theorem. Probem set 6 The Perron Frobenius theorem. Math 22a4 Oct 2 204, Due Oct.28 In a future probem set I want to discuss some criteria which aow us to concude that that the ground state of a sef-adjoint operator

More information

A. Distribution of the test statistic

A. Distribution of the test statistic A. Distribution of the test statistic In the sequentia test, we first compute the test statistic from a mini-batch of size m. If a decision cannot be made with this statistic, we keep increasing the mini-batch

More information

BDD-Based Analysis of Gapped q-gram Filters

BDD-Based Analysis of Gapped q-gram Filters BDD-Based Anaysis of Gapped q-gram Fiters Marc Fontaine, Stefan Burkhardt 2 and Juha Kärkkäinen 2 Max-Panck-Institut für Informatik Stuhsatzenhausweg 85, 6623 Saarbrücken, Germany e-mai: stburk@mpi-sb.mpg.de

More information

Approximated MLC shape matrix decomposition with interleaf collision constraint

Approximated MLC shape matrix decomposition with interleaf collision constraint Approximated MLC shape matrix decomposition with intereaf coision constraint Thomas Kainowski Antje Kiese Abstract Shape matrix decomposition is a subprobem in radiation therapy panning. A given fuence

More information

Combining reaction kinetics to the multi-phase Gibbs energy calculation

Combining reaction kinetics to the multi-phase Gibbs energy calculation 7 th European Symposium on Computer Aided Process Engineering ESCAPE7 V. Pesu and P.S. Agachi (Editors) 2007 Esevier B.V. A rights reserved. Combining reaction inetics to the muti-phase Gibbs energy cacuation

More information

Efficiently Generating Random Bits from Finite State Markov Chains

Efficiently Generating Random Bits from Finite State Markov Chains 1 Efficienty Generating Random Bits from Finite State Markov Chains Hongchao Zhou and Jehoshua Bruck, Feow, IEEE Abstract The probem of random number generation from an uncorreated random source (of unknown

More information

Algorithms to solve massively under-defined systems of multivariate quadratic equations

Algorithms to solve massively under-defined systems of multivariate quadratic equations Agorithms to sove massivey under-defined systems of mutivariate quadratic equations Yasufumi Hashimoto Abstract It is we known that the probem to sove a set of randomy chosen mutivariate quadratic equations

More information

XSAT of linear CNF formulas

XSAT of linear CNF formulas XSAT of inear CN formuas Bernd R. Schuh Dr. Bernd Schuh, D-50968 Kön, Germany; bernd.schuh@netcoogne.de eywords: compexity, XSAT, exact inear formua, -reguarity, -uniformity, NPcompeteness Abstract. Open

More information

Stochastic Complement Analysis of Multi-Server Threshold Queues. with Hysteresis. Abstract

Stochastic Complement Analysis of Multi-Server Threshold Queues. with Hysteresis. Abstract Stochastic Compement Anaysis of Muti-Server Threshod Queues with Hysteresis John C.S. Lui The Dept. of Computer Science & Engineering The Chinese University of Hong Kong Leana Goubchik Dept. of Computer

More information

Partial permutation decoding for MacDonald codes

Partial permutation decoding for MacDonald codes Partia permutation decoding for MacDonad codes J.D. Key Department of Mathematics and Appied Mathematics University of the Western Cape 7535 Bevie, South Africa P. Seneviratne Department of Mathematics

More information

Bayesian Learning. You hear a which which could equally be Thanks or Tanks, which would you go with?

Bayesian Learning. You hear a which which could equally be Thanks or Tanks, which would you go with? Bayesian Learning A powerfu and growing approach in machine earning We use it in our own decision making a the time You hear a which which coud equay be Thanks or Tanks, which woud you go with? Combine

More information

Chemical Kinetics Part 2

Chemical Kinetics Part 2 Integrated Rate Laws Chemica Kinetics Part 2 The rate aw we have discussed thus far is the differentia rate aw. Let us consider the very simpe reaction: a A à products The differentia rate reates the rate

More information

Approximated MLC shape matrix decomposition with interleaf collision constraint

Approximated MLC shape matrix decomposition with interleaf collision constraint Agorithmic Operations Research Vo.4 (29) 49 57 Approximated MLC shape matrix decomposition with intereaf coision constraint Antje Kiese and Thomas Kainowski Institut für Mathematik, Universität Rostock,

More information

On the Goal Value of a Boolean Function

On the Goal Value of a Boolean Function On the Goa Vaue of a Booean Function Eric Bach Dept. of CS University of Wisconsin 1210 W. Dayton St. Madison, WI 53706 Lisa Heerstein Dept of CSE NYU Schoo of Engineering 2 Metrotech Center, 10th Foor

More information

Separation of Variables and a Spherical Shell with Surface Charge

Separation of Variables and a Spherical Shell with Surface Charge Separation of Variabes and a Spherica She with Surface Charge In cass we worked out the eectrostatic potentia due to a spherica she of radius R with a surface charge density σθ = σ cos θ. This cacuation

More information

Discrete Techniques. Chapter Introduction

Discrete Techniques. Chapter Introduction Chapter 3 Discrete Techniques 3. Introduction In the previous two chapters we introduced Fourier transforms of continuous functions of the periodic and non-periodic (finite energy) type, as we as various

More information

STA 216 Project: Spline Approach to Discrete Survival Analysis

STA 216 Project: Spline Approach to Discrete Survival Analysis : Spine Approach to Discrete Surviva Anaysis November 4, 005 1 Introduction Athough continuous surviva anaysis differs much from the discrete surviva anaysis, there is certain ink between the two modeing

More information

Statistical Learning Theory: A Primer

Statistical Learning Theory: A Primer Internationa Journa of Computer Vision 38(), 9 3, 2000 c 2000 uwer Academic Pubishers. Manufactured in The Netherands. Statistica Learning Theory: A Primer THEODOROS EVGENIOU, MASSIMILIANO PONTIL AND TOMASO

More information

An Algorithm for Pruning Redundant Modules in Min-Max Modular Network

An Algorithm for Pruning Redundant Modules in Min-Max Modular Network An Agorithm for Pruning Redundant Modues in Min-Max Moduar Network Hui-Cheng Lian and Bao-Liang Lu Department of Computer Science and Engineering, Shanghai Jiao Tong University 1954 Hua Shan Rd., Shanghai

More information

SUPPLEMENTARY MATERIAL TO INNOVATED SCALABLE EFFICIENT ESTIMATION IN ULTRA-LARGE GAUSSIAN GRAPHICAL MODELS

SUPPLEMENTARY MATERIAL TO INNOVATED SCALABLE EFFICIENT ESTIMATION IN ULTRA-LARGE GAUSSIAN GRAPHICAL MODELS ISEE 1 SUPPLEMENTARY MATERIAL TO INNOVATED SCALABLE EFFICIENT ESTIMATION IN ULTRA-LARGE GAUSSIAN GRAPHICAL MODELS By Yingying Fan and Jinchi Lv University of Southern Caifornia This Suppementary Materia

More information

Discrete Techniques. Chapter Introduction

Discrete Techniques. Chapter Introduction Chapter 3 Discrete Techniques 3. Introduction In the previous two chapters we introduced Fourier transforms of continuous functions of the periodic and non-periodic (finite energy) type, we as various

More information

Integrating Factor Methods as Exponential Integrators

Integrating Factor Methods as Exponential Integrators Integrating Factor Methods as Exponentia Integrators Borisav V. Minchev Department of Mathematica Science, NTNU, 7491 Trondheim, Norway Borko.Minchev@ii.uib.no Abstract. Recenty a ot of effort has been

More information

General Certificate of Education Advanced Level Examination June 2010

General Certificate of Education Advanced Level Examination June 2010 Genera Certificate of Education Advanced Leve Examination June 2010 Human Bioogy HBI6T/Q10/task Unit 6T A2 Investigative Skis Assignment Task Sheet The effect of using one or two eyes on the perception

More information

Some Measures for Asymmetry of Distributions

Some Measures for Asymmetry of Distributions Some Measures for Asymmetry of Distributions Georgi N. Boshnakov First version: 31 January 2006 Research Report No. 5, 2006, Probabiity and Statistics Group Schoo of Mathematics, The University of Manchester

More information

Source and Relay Matrices Optimization for Multiuser Multi-Hop MIMO Relay Systems

Source and Relay Matrices Optimization for Multiuser Multi-Hop MIMO Relay Systems Source and Reay Matrices Optimization for Mutiuser Muti-Hop MIMO Reay Systems Yue Rong Department of Eectrica and Computer Engineering, Curtin University, Bentey, WA 6102, Austraia Abstract In this paper,

More information

Efficient Generation of Random Bits from Finite State Markov Chains

Efficient Generation of Random Bits from Finite State Markov Chains Efficient Generation of Random Bits from Finite State Markov Chains Hongchao Zhou and Jehoshua Bruck, Feow, IEEE Abstract The probem of random number generation from an uncorreated random source (of unknown

More information

8 Digifl'.11 Cth:uits and devices

8 Digifl'.11 Cth:uits and devices 8 Digif'. Cth:uits and devices 8. Introduction In anaog eectronics, votage is a continuous variabe. This is usefu because most physica quantities we encounter are continuous: sound eves, ight intensity,

More information

MATH 172: MOTIVATION FOR FOURIER SERIES: SEPARATION OF VARIABLES

MATH 172: MOTIVATION FOR FOURIER SERIES: SEPARATION OF VARIABLES MATH 172: MOTIVATION FOR FOURIER SERIES: SEPARATION OF VARIABLES Separation of variabes is a method to sove certain PDEs which have a warped product structure. First, on R n, a inear PDE of order m is

More information

BP neural network-based sports performance prediction model applied research

BP neural network-based sports performance prediction model applied research Avaiabe onine www.jocpr.com Journa of Chemica and Pharmaceutica Research, 204, 6(7:93-936 Research Artice ISSN : 0975-7384 CODEN(USA : JCPRC5 BP neura networ-based sports performance prediction mode appied

More information

C. Fourier Sine Series Overview

C. Fourier Sine Series Overview 12 PHILIP D. LOEWEN C. Fourier Sine Series Overview Let some constant > be given. The symboic form of the FSS Eigenvaue probem combines an ordinary differentia equation (ODE) on the interva (, ) with a

More information

Chemical Kinetics Part 2. Chapter 16

Chemical Kinetics Part 2. Chapter 16 Chemica Kinetics Part 2 Chapter 16 Integrated Rate Laws The rate aw we have discussed thus far is the differentia rate aw. Let us consider the very simpe reaction: a A à products The differentia rate reates

More information

c 2007 Society for Industrial and Applied Mathematics

c 2007 Society for Industrial and Applied Mathematics SIAM REVIEW Vo. 49,No. 1,pp. 111 1 c 7 Society for Industria and Appied Mathematics Domino Waves C. J. Efthimiou M. D. Johnson Abstract. Motivated by a proposa of Daykin [Probem 71-19*, SIAM Rev., 13 (1971),

More information

Paragraph Topic Classification

Paragraph Topic Classification Paragraph Topic Cassification Eugene Nho Graduate Schoo of Business Stanford University Stanford, CA 94305 enho@stanford.edu Edward Ng Department of Eectrica Engineering Stanford University Stanford, CA

More information

Turbo Codes. Coding and Communication Laboratory. Dept. of Electrical Engineering, National Chung Hsing University

Turbo Codes. Coding and Communication Laboratory. Dept. of Electrical Engineering, National Chung Hsing University Turbo Codes Coding and Communication Laboratory Dept. of Eectrica Engineering, Nationa Chung Hsing University Turbo codes 1 Chapter 12: Turbo Codes 1. Introduction 2. Turbo code encoder 3. Design of intereaver

More information

arxiv: v2 [cond-mat.stat-mech] 14 Nov 2008

arxiv: v2 [cond-mat.stat-mech] 14 Nov 2008 Random Booean Networks Barbara Drosse Institute of Condensed Matter Physics, Darmstadt University of Technoogy, Hochschustraße 6, 64289 Darmstadt, Germany (Dated: June 27) arxiv:76.335v2 [cond-mat.stat-mech]

More information

Sequential Decoding of Polar Codes with Arbitrary Binary Kernel

Sequential Decoding of Polar Codes with Arbitrary Binary Kernel Sequentia Decoding of Poar Codes with Arbitrary Binary Kerne Vera Miosavskaya, Peter Trifonov Saint-Petersburg State Poytechnic University Emai: veram,petert}@dcn.icc.spbstu.ru Abstract The probem of efficient

More information

Do Schools Matter for High Math Achievement? Evidence from the American Mathematics Competitions Glenn Ellison and Ashley Swanson Online Appendix

Do Schools Matter for High Math Achievement? Evidence from the American Mathematics Competitions Glenn Ellison and Ashley Swanson Online Appendix VOL. NO. DO SCHOOLS MATTER FOR HIGH MATH ACHIEVEMENT? 43 Do Schoos Matter for High Math Achievement? Evidence from the American Mathematics Competitions Genn Eison and Ashey Swanson Onine Appendix Appendix

More information

Analysis of Emerson s Multiple Model Interpolation Estimation Algorithms: The MIMO Case

Analysis of Emerson s Multiple Model Interpolation Estimation Algorithms: The MIMO Case Technica Report PC-04-00 Anaysis of Emerson s Mutipe Mode Interpoation Estimation Agorithms: The MIMO Case João P. Hespanha Dae E. Seborg University of Caifornia, Santa Barbara February 0, 004 Anaysis

More information

Recursive Constructions of Parallel FIFO and LIFO Queues with Switched Delay Lines

Recursive Constructions of Parallel FIFO and LIFO Queues with Switched Delay Lines Recursive Constructions of Parae FIFO and LIFO Queues with Switched Deay Lines Po-Kai Huang, Cheng-Shang Chang, Feow, IEEE, Jay Cheng, Member, IEEE, and Duan-Shin Lee, Senior Member, IEEE Abstract One

More information

(This is a sample cover image for this issue. The actual cover is not yet available at this time.)

(This is a sample cover image for this issue. The actual cover is not yet available at this time.) (This is a sampe cover image for this issue The actua cover is not yet avaiabe at this time) This artice appeared in a journa pubished by Esevier The attached copy is furnished to the author for interna

More information

Uniprocessor Feasibility of Sporadic Tasks with Constrained Deadlines is Strongly conp-complete

Uniprocessor Feasibility of Sporadic Tasks with Constrained Deadlines is Strongly conp-complete Uniprocessor Feasibiity of Sporadic Tasks with Constrained Deadines is Strongy conp-compete Pontus Ekberg and Wang Yi Uppsaa University, Sweden Emai: {pontus.ekberg yi}@it.uu.se Abstract Deciding the feasibiity

More information

Two view learning: SVM-2K, Theory and Practice

Two view learning: SVM-2K, Theory and Practice Two view earning: SVM-2K, Theory and Practice Jason D.R. Farquhar jdrf99r@ecs.soton.ac.uk Hongying Meng hongying@cs.york.ac.uk David R. Hardoon drh@ecs.soton.ac.uk John Shawe-Tayor jst@ecs.soton.ac.uk

More information

Higher dimensional PDEs and multidimensional eigenvalue problems

Higher dimensional PDEs and multidimensional eigenvalue problems Higher dimensiona PEs and mutidimensiona eigenvaue probems 1 Probems with three independent variabes Consider the prototypica equations u t = u (iffusion) u tt = u (W ave) u zz = u (Lapace) where u = u

More information

Stat 155 Game theory, Yuval Peres Fall Lectures 4,5,6

Stat 155 Game theory, Yuval Peres Fall Lectures 4,5,6 Stat 155 Game theory, Yuva Peres Fa 2004 Lectures 4,5,6 In the ast ecture, we defined N and P positions for a combinatoria game. We wi now show more formay that each starting position in a combinatoria

More information

arxiv:quant-ph/ v3 6 Jan 1995

arxiv:quant-ph/ v3 6 Jan 1995 arxiv:quant-ph/9501001v3 6 Jan 1995 Critique of proposed imit to space time measurement, based on Wigner s cocks and mirrors L. Diósi and B. Lukács KFKI Research Institute for Partice and Nucear Physics

More information

A Novel Learning Method for Elman Neural Network Using Local Search

A Novel Learning Method for Elman Neural Network Using Local Search Neura Information Processing Letters and Reviews Vo. 11, No. 8, August 2007 LETTER A Nove Learning Method for Eman Neura Networ Using Loca Search Facuty of Engineering, Toyama University, Gofuu 3190 Toyama

More information

The influence of temperature of photovoltaic modules on performance of solar power plant

The influence of temperature of photovoltaic modules on performance of solar power plant IOSR Journa of Engineering (IOSRJEN) ISSN (e): 2250-3021, ISSN (p): 2278-8719 Vo. 05, Issue 04 (Apri. 2015), V1 PP 09-15 www.iosrjen.org The infuence of temperature of photovotaic modues on performance

More information

VI.G Exact free energy of the Square Lattice Ising model

VI.G Exact free energy of the Square Lattice Ising model VI.G Exact free energy of the Square Lattice Ising mode As indicated in eq.(vi.35), the Ising partition function is reated to a sum S, over coections of paths on the attice. The aowed graphs for a square

More information

Lecture Note 3: Stationary Iterative Methods

Lecture Note 3: Stationary Iterative Methods MATH 5330: Computationa Methods of Linear Agebra Lecture Note 3: Stationary Iterative Methods Xianyi Zeng Department of Mathematica Sciences, UTEP Stationary Iterative Methods The Gaussian eimination (or

More information

A proposed nonparametric mixture density estimation using B-spline functions

A proposed nonparametric mixture density estimation using B-spline functions A proposed nonparametric mixture density estimation using B-spine functions Atizez Hadrich a,b, Mourad Zribi a, Afif Masmoudi b a Laboratoire d Informatique Signa et Image de a Côte d Opae (LISIC-EA 4491),

More information

4 1-D Boundary Value Problems Heat Equation

4 1-D Boundary Value Problems Heat Equation 4 -D Boundary Vaue Probems Heat Equation The main purpose of this chapter is to study boundary vaue probems for the heat equation on a finite rod a x b. u t (x, t = ku xx (x, t, a < x < b, t > u(x, = ϕ(x

More information

Coded Caching for Files with Distinct File Sizes

Coded Caching for Files with Distinct File Sizes Coded Caching for Fies with Distinct Fie Sizes Jinbei Zhang iaojun Lin Chih-Chun Wang inbing Wang Department of Eectronic Engineering Shanghai Jiao ong University China Schoo of Eectrica and Computer Engineering

More information

Componentwise Determination of the Interval Hull Solution for Linear Interval Parameter Systems

Componentwise Determination of the Interval Hull Solution for Linear Interval Parameter Systems Componentwise Determination of the Interva Hu Soution for Linear Interva Parameter Systems L. V. Koev Dept. of Theoretica Eectrotechnics, Facuty of Automatics, Technica University of Sofia, 1000 Sofia,

More information

A Statistical Framework for Real-time Event Detection in Power Systems

A Statistical Framework for Real-time Event Detection in Power Systems 1 A Statistica Framework for Rea-time Event Detection in Power Systems Noan Uhrich, Tim Christman, Phiip Swisher, and Xichen Jiang Abstract A quickest change detection (QCD) agorithm is appied to the probem

More information

arxiv: v1 [math.co] 12 May 2013

arxiv: v1 [math.co] 12 May 2013 EMBEDDING CYCLES IN FINITE PLANES FELIX LAZEBNIK, KEITH E. MELLINGER, AND SCAR VEGA arxiv:1305.2646v1 [math.c] 12 May 2013 Abstract. We define and study embeddings of cyces in finite affine and projective

More information

Asymptotic Properties of a Generalized Cross Entropy Optimization Algorithm

Asymptotic Properties of a Generalized Cross Entropy Optimization Algorithm 1 Asymptotic Properties of a Generaized Cross Entropy Optimization Agorithm Zijun Wu, Michae Koonko, Institute for Appied Stochastics and Operations Research, Caustha Technica University Abstract The discrete

More information

THE OUT-OF-PLANE BEHAVIOUR OF SPREAD-TOW FABRICS

THE OUT-OF-PLANE BEHAVIOUR OF SPREAD-TOW FABRICS ECCM6-6 TH EUROPEAN CONFERENCE ON COMPOSITE MATERIALS, Sevie, Spain, -6 June 04 THE OUT-OF-PLANE BEHAVIOUR OF SPREAD-TOW FABRICS M. Wysocki a,b*, M. Szpieg a, P. Heström a and F. Ohsson c a Swerea SICOMP

More information

Asynchronous Control for Coupled Markov Decision Systems

Asynchronous Control for Coupled Markov Decision Systems INFORMATION THEORY WORKSHOP (ITW) 22 Asynchronous Contro for Couped Marov Decision Systems Michae J. Neey University of Southern Caifornia Abstract This paper considers optima contro for a coection of

More information

First-Order Corrections to Gutzwiller s Trace Formula for Systems with Discrete Symmetries

First-Order Corrections to Gutzwiller s Trace Formula for Systems with Discrete Symmetries c 26 Noninear Phenomena in Compex Systems First-Order Corrections to Gutzwier s Trace Formua for Systems with Discrete Symmetries Hoger Cartarius, Jörg Main, and Günter Wunner Institut für Theoretische

More information

Explicit overall risk minimization transductive bound

Explicit overall risk minimization transductive bound 1 Expicit overa risk minimization transductive bound Sergio Decherchi, Paoo Gastado, Sandro Ridea, Rodofo Zunino Dept. of Biophysica and Eectronic Engineering (DIBE), Genoa University Via Opera Pia 11a,

More information

K a,k minors in graphs of bounded tree-width *

K a,k minors in graphs of bounded tree-width * K a,k minors in graphs of bounded tree-width * Thomas Böhme Institut für Mathematik Technische Universität Imenau Imenau, Germany E-mai: tboehme@theoinf.tu-imenau.de and John Maharry Department of Mathematics

More information

THE THREE POINT STEINER PROBLEM ON THE FLAT TORUS: THE MINIMAL LUNE CASE

THE THREE POINT STEINER PROBLEM ON THE FLAT TORUS: THE MINIMAL LUNE CASE THE THREE POINT STEINER PROBLEM ON THE FLAT TORUS: THE MINIMAL LUNE CASE KATIE L. MAY AND MELISSA A. MITCHELL Abstract. We show how to identify the minima path network connecting three fixed points on

More information

Statistical Inference, Econometric Analysis and Matrix Algebra

Statistical Inference, Econometric Analysis and Matrix Algebra Statistica Inference, Econometric Anaysis and Matrix Agebra Bernhard Schipp Water Krämer Editors Statistica Inference, Econometric Anaysis and Matrix Agebra Festschrift in Honour of Götz Trenker Physica-Verag

More information

High Efficiency Development of a Reciprocating Compressor by Clarification of Loss Generation in Bearings

High Efficiency Development of a Reciprocating Compressor by Clarification of Loss Generation in Bearings Purdue University Purdue e-pubs Internationa Compressor Engineering Conference Schoo of Mechanica Engineering 2010 High Efficiency Deveopment of a Reciprocating Compressor by Carification of Loss Generation

More information

Construction of Supersaturated Design with Large Number of Factors by the Complementary Design Method

Construction of Supersaturated Design with Large Number of Factors by the Complementary Design Method Acta Mathematicae Appicatae Sinica, Engish Series Vo. 29, No. 2 (2013) 253 262 DOI: 10.1007/s10255-013-0214-6 http://www.appmath.com.cn & www.springerlink.com Acta Mathema cae Appicatae Sinica, Engish

More information

Math 124B January 17, 2012

Math 124B January 17, 2012 Math 124B January 17, 212 Viktor Grigoryan 3 Fu Fourier series We saw in previous ectures how the Dirichet and Neumann boundary conditions ead to respectivey sine and cosine Fourier series of the initia

More information

International Journal of Mass Spectrometry

International Journal of Mass Spectrometry Internationa Journa of Mass Spectrometry 280 (2009) 179 183 Contents ists avaiabe at ScienceDirect Internationa Journa of Mass Spectrometry journa homepage: www.esevier.com/ocate/ijms Stark mixing by ion-rydberg

More information

Lecture 17 - The Secrets we have Swept Under the Rug

Lecture 17 - The Secrets we have Swept Under the Rug Lecture 17 - The Secrets we have Swept Under the Rug Today s ectures examines some of the uirky features of eectrostatics that we have negected up unti this point A Puzze... Let s go back to the basics

More information

FRST Multivariate Statistics. Multivariate Discriminant Analysis (MDA)

FRST Multivariate Statistics. Multivariate Discriminant Analysis (MDA) 1 FRST 531 -- Mutivariate Statistics Mutivariate Discriminant Anaysis (MDA) Purpose: 1. To predict which group (Y) an observation beongs to based on the characteristics of p predictor (X) variabes, using

More information

Biometrics Unit, 337 Warren Hall Cornell University, Ithaca, NY and. B. L. Raktoe

Biometrics Unit, 337 Warren Hall Cornell University, Ithaca, NY and. B. L. Raktoe NONISCMORPHIC CCMPLETE SETS OF ORTHOGONAL F-SQ.UARES, HADAMARD MATRICES, AND DECCMPOSITIONS OF A 2 4 DESIGN S. J. Schwager and w. T. Federer Biometrics Unit, 337 Warren Ha Corne University, Ithaca, NY

More information

Distributed average consensus: Beyond the realm of linearity

Distributed average consensus: Beyond the realm of linearity Distributed average consensus: Beyond the ream of inearity Usman A. Khan, Soummya Kar, and José M. F. Moura Department of Eectrica and Computer Engineering Carnegie Meon University 5 Forbes Ave, Pittsburgh,

More information

IE 361 Exam 1. b) Give *&% confidence limits for the bias of this viscometer. (No need to simplify.)

IE 361 Exam 1. b) Give *&% confidence limits for the bias of this viscometer. (No need to simplify.) October 9, 00 IE 6 Exam Prof. Vardeman. The viscosity of paint is measured with a "viscometer" in units of "Krebs." First, a standard iquid of "known" viscosity *# Krebs is tested with a company viscometer

More information

FRIEZE GROUPS IN R 2

FRIEZE GROUPS IN R 2 FRIEZE GROUPS IN R 2 MAXWELL STOLARSKI Abstract. Focusing on the Eucidean pane under the Pythagorean Metric, our goa is to cassify the frieze groups, discrete subgroups of the set of isometries of the

More information

Throughput Optimal Scheduling for Wireless Downlinks with Reconfiguration Delay

Throughput Optimal Scheduling for Wireless Downlinks with Reconfiguration Delay Throughput Optima Scheduing for Wireess Downinks with Reconfiguration Deay Vineeth Baa Sukumaran vineethbs@gmai.com Department of Avionics Indian Institute of Space Science and Technoogy. Abstract We consider

More information

Theory and implementation behind: Universal surface creation - smallest unitcell

Theory and implementation behind: Universal surface creation - smallest unitcell Teory and impementation beind: Universa surface creation - smaest unitce Bjare Brin Buus, Jaob Howat & Tomas Bigaard September 15, 218 1 Construction of surface sabs Te aim for tis part of te project is

More information

The EM Algorithm applied to determining new limit points of Mahler measures

The EM Algorithm applied to determining new limit points of Mahler measures Contro and Cybernetics vo. 39 (2010) No. 4 The EM Agorithm appied to determining new imit points of Maher measures by Souad E Otmani, Georges Rhin and Jean-Marc Sac-Épée Université Pau Veraine-Metz, LMAM,

More information

(f) is called a nearly holomorphic modular form of weight k + 2r as in [5].

(f) is called a nearly holomorphic modular form of weight k + 2r as in [5]. PRODUCTS OF NEARLY HOLOMORPHIC EIGENFORMS JEFFREY BEYERL, KEVIN JAMES, CATHERINE TRENTACOSTE, AND HUI XUE Abstract. We prove that the product of two neary hoomorphic Hece eigenforms is again a Hece eigenform

More information

General Certificate of Education Advanced Level Examination June 2010

General Certificate of Education Advanced Level Examination June 2010 Genera Certificate of Education Advanced Leve Examination June 2010 Human Bioogy HBI6T/P10/task Unit 6T A2 Investigative Skis Assignment Task Sheet The effect of temperature on the rate of photosynthesis

More information

Rate-Distortion Theory of Finite Point Processes

Rate-Distortion Theory of Finite Point Processes Rate-Distortion Theory of Finite Point Processes Günther Koiander, Dominic Schuhmacher, and Franz Hawatsch, Feow, IEEE Abstract We study the compression of data in the case where the usefu information

More information

Nonperturbative Shell Correction to the Bethe Bloch Formula for the Energy Losses of Fast Charged Particles

Nonperturbative Shell Correction to the Bethe Bloch Formula for the Energy Losses of Fast Charged Particles ISSN 002-3640, JETP Letters, 20, Vo. 94, No., pp. 5. Peiades Pubishing, Inc., 20. Origina Russian Text V.I. Matveev, D.N. Makarov, 20, pubished in Pis ma v Zhurna Eksperimenta noi i Teoreticheskoi Fiziki,

More information

Timing Attacks on Cognitive Authentication Schemes

Timing Attacks on Cognitive Authentication Schemes Timing Attacks on Cognitive Authentication Schemes Mario Čagaj, Toni Perković, Member, IEEE, Marin Bugarić Department of Eectrica Engineering, FESB, University of Spit, Croatia {mcagaj, toperkov, mbugaric}@fesb.hr

More information

Formulas for Angular-Momentum Barrier Factors Version II

Formulas for Angular-Momentum Barrier Factors Version II BNL PREPRINT BNL-QGS-06-101 brfactor1.tex Formuas for Anguar-Momentum Barrier Factors Version II S. U. Chung Physics Department, Brookhaven Nationa Laboratory, Upton, NY 11973 March 19, 2015 abstract A

More information

Gauss Law. 2. Gauss s Law: connects charge and field 3. Applications of Gauss s Law

Gauss Law. 2. Gauss s Law: connects charge and field 3. Applications of Gauss s Law Gauss Law 1. Review on 1) Couomb s Law (charge and force) 2) Eectric Fied (fied and force) 2. Gauss s Law: connects charge and fied 3. Appications of Gauss s Law Couomb s Law and Eectric Fied Couomb s

More information

Efficient Similarity Search across Top-k Lists under the Kendall s Tau Distance

Efficient Similarity Search across Top-k Lists under the Kendall s Tau Distance Efficient Simiarity Search across Top-k Lists under the Kenda s Tau Distance Koninika Pa TU Kaisersautern Kaisersautern, Germany pa@cs.uni-k.de Sebastian Miche TU Kaisersautern Kaisersautern, Germany smiche@cs.uni-k.de

More information

A simple reliability block diagram method for safety integrity verification

A simple reliability block diagram method for safety integrity verification Reiabiity Engineering and System Safety 92 (2007) 1267 1273 www.esevier.com/ocate/ress A simpe reiabiity bock diagram method for safety integrity verification Haitao Guo, Xianhui Yang epartment of Automation,

More information

Statistical Learning Theory: a Primer

Statistical Learning Theory: a Primer ??,??, 1 6 (??) c?? Kuwer Academic Pubishers, Boston. Manufactured in The Netherands. Statistica Learning Theory: a Primer THEODOROS EVGENIOU AND MASSIMILIANO PONTIL Center for Bioogica and Computationa

More information

LECTURE NOTES 9 TRACELESS SYMMETRIC TENSOR APPROACH TO LEGENDRE POLYNOMIALS AND SPHERICAL HARMONICS

LECTURE NOTES 9 TRACELESS SYMMETRIC TENSOR APPROACH TO LEGENDRE POLYNOMIALS AND SPHERICAL HARMONICS MASSACHUSETTS INSTITUTE OF TECHNOLOGY Physics Department Physics 8.07: Eectromagnetism II October 7, 202 Prof. Aan Guth LECTURE NOTES 9 TRACELESS SYMMETRIC TENSOR APPROACH TO LEGENDRE POLYNOMIALS AND SPHERICAL

More information

DIGITAL FILTER DESIGN OF IIR FILTERS USING REAL VALUED GENETIC ALGORITHM

DIGITAL FILTER DESIGN OF IIR FILTERS USING REAL VALUED GENETIC ALGORITHM DIGITAL FILTER DESIGN OF IIR FILTERS USING REAL VALUED GENETIC ALGORITHM MIKAEL NILSSON, MATTIAS DAHL AND INGVAR CLAESSON Bekinge Institute of Technoogy Department of Teecommunications and Signa Processing

More information

Lecture 6: Moderately Large Deflection Theory of Beams

Lecture 6: Moderately Large Deflection Theory of Beams Structura Mechanics 2.8 Lecture 6 Semester Yr Lecture 6: Moderatey Large Defection Theory of Beams 6.1 Genera Formuation Compare to the cassica theory of beams with infinitesima deformation, the moderatey

More information

NIKOS FRANTZIKINAKIS. N n N where (Φ N) N N is any Følner sequence

NIKOS FRANTZIKINAKIS. N n N where (Φ N) N N is any Følner sequence SOME OPE PROBLEMS O MULTIPLE ERGODIC AVERAGES IKOS FRATZIKIAKIS. Probems reated to poynomia sequences In this section we give a ist of probems reated to the study of mutipe ergodic averages invoving iterates

More information

In-plane shear stiffness of bare steel deck through shell finite element models. G. Bian, B.W. Schafer. June 2017

In-plane shear stiffness of bare steel deck through shell finite element models. G. Bian, B.W. Schafer. June 2017 In-pane shear stiffness of bare stee deck through she finite eement modes G. Bian, B.W. Schafer June 7 COLD-FORMED STEEL RESEARCH CONSORTIUM REPORT SERIES CFSRC R-7- SDII Stee Diaphragm Innovation Initiative

More information

II. PROBLEM. A. Description. For the space of audio signals

II. PROBLEM. A. Description. For the space of audio signals CS229 - Fina Report Speech Recording based Language Recognition (Natura Language) Leopod Cambier - cambier; Matan Leibovich - matane; Cindy Orozco Bohorquez - orozcocc ABSTRACT We construct a rea time

More information

SEMINAR 2. PENDULUMS. V = mgl cos θ. (2) L = T V = 1 2 ml2 θ2 + mgl cos θ, (3) d dt ml2 θ2 + mgl sin θ = 0, (4) θ + g l

SEMINAR 2. PENDULUMS. V = mgl cos θ. (2) L = T V = 1 2 ml2 θ2 + mgl cos θ, (3) d dt ml2 θ2 + mgl sin θ = 0, (4) θ + g l Probem 7. Simpe Penduum SEMINAR. PENDULUMS A simpe penduum means a mass m suspended by a string weightess rigid rod of ength so that it can swing in a pane. The y-axis is directed down, x-axis is directed

More information

THINKING IN PYRAMIDS

THINKING IN PYRAMIDS ECS 178 Course Notes THINKING IN PYRAMIDS Kenneth I. Joy Institute for Data Anaysis and Visuaization Department of Computer Science University of Caifornia, Davis Overview It is frequenty usefu to think

More information

Error Estimating Codes for Insertion and Deletion Channels

Error Estimating Codes for Insertion and Deletion Channels Error Estimating Codes for Insertion and Deetion Channes Jiwei Huang Department of Computer cience and Technoogy Tsinghua University Beijing, China huangjw05@gmai.com Justin Romberg choo of Eectrica and

More information

Fitting Algorithms for MMPP ATM Traffic Models

Fitting Algorithms for MMPP ATM Traffic Models Fitting Agorithms for PP AT Traffic odes A. Nogueira, P. Savador, R. Vaadas University of Aveiro / Institute of Teecommunications, 38-93 Aveiro, Portuga; e-mai: (nogueira, savador, rv)@av.it.pt ABSTRACT

More information