Dependency parsing based on dynamic local optimization

Dependency parsing in Myanmar face many challenges due to agglutinative morphology, flexible word order, and lack of explicit word boundaries. This paper addresses these challenges by investigating the performance of Myanmar dependency parsing based three pretrained-transformer models: XLM-RoBERTa-Large, XLM-RoBERTa-base-Longformer-4096, and language-specific model, MyanBERTa. The proposed methodology utilizes a parameter-efficient model with adapter layers and a biaffine parsing mechanism by optimizing multi-task learning objective across six linguistic prediction tasks. The experimental results demonstrated that XLM-RoBERTa-Large achieves the highest Labeled Attachment Score (LAS) and Unlabeled Attachment Score (UAS) on both development set and test set. Despite XLM-RoBERTa-base-Longformer-4096 model can extended its input capacity and MyanBERTa is pretrained on a language capacity, their performance drop compared to XLM-RoBERTa-Large. These findings suggest that the scale and architectural strength of models are more critical for high-performance dependency parsing than models with an extended input capacity or those trained on a language-specific corpus.

Finite state automata are of fundamental importance in theory as well as applications in computer science. This dissertation presents several parallel algorithms for processing of them, including minimizing the number of states for a given finite state automaton, finding homing sequences for a given finite state automaton, converting a regular expression into an $\\varepsilon$-free nondeterministic finite automaton. Implementation of some algorithms, as well as their experimental results, are also given. For minimization of deterministic finite automata, we present a parallel algorithm and discuss its implementation on a practical parallel machine using data parallel and message passing paradigms. Its time complexity on a p processor EREW PRAM $(p\\le n)$ for inputs of size n is $O({n\\log\\sp2n\\over p}+\\log n\\ \\log p)$ uniformly on almost all instances. The work done by our algorithm is thus within a factor of $O(\\log n)$ of the best known sequential algorithm. The space used by our algorithm is linear in the input size. The actual resource requirements of our implementations are consistent with these estimates. For finding homing sequences of deterministic finite automata, we present both deterministic and randomized algorithms. Two randomized algorithms of time complexity $O(log\\sp2n)$ are given. One of the randomized algorithms produces a homing sequence of length $O(n\\ log\\sp2n)$ for almost all DFA's with n states. The deterministic algorithm is based on a well known sequential algorithm. The implementation of the deterministic parallel algorithm shows that it can handle randomly generated instances with millions of states, and all DFA's with thousands of states using currently available parallel machines. Finally, we present an efficient parallel algorithm for converting a regular expression into an $\\varepsilon$-free nondeterministic finite automaton. The resulting automaton is natural representation of a regular expression and has some important and practical uses in the document processing area. The time complexity of the algorithm is $O(\\log n)$ and the number of processors used is $O(n\\sp2).$

We revisit a classical problem in computational geometry that has been studied since the 1980s: in the rectangle enclosure problem we want to report all k enclosing pairs of n input rectangles in 2D. We present the first deterministic algorithm that takes O(nlogn + k) worst-case time and O(n) space in the word-RAM model. This improves previous deterministic algorithms with O((nlogn + k)loglogn) running time. We achieve the result by derandomizing the algorithm of Chan, Larsen and Pătraşcu [SoCG’11] that attains the same time complexity but in expectation.The 2D rectangle enclosure problem is related to the offline dominance range reporting problem in 4D, and our result leads to the currently fastest deterministic algorithm for offline dominance reporting in any constant dimension d ≥ 4.A key tool behind Chan et al.’s previous randomized algorithm is shallow cuttings for 3D dominance ranges. Recently, Afshani and Tsakalidis [SODA’14] obtained a deterministic O(nlogn)-time algorithm to construct such cuttings. We first present an improved deterministic construction algorithm that runs in O(nloglogn) time in the word-RAM; this result is of independent interest. Many additional ideas are then incorporated, including a linear-time algorithm for merging shallow cuttings and an algorithm for an offline tree point location problem.KeywordsQuery PointDeterministic AlgorithmInput PointOuter CornerPlanar SeparatorThese keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

Gathering a group of mobile agents is a fundamental task in the field of distributed and mobile systems. This can be made drastically more difficult to achieve when some agents are subject to faults, especially the Byzantine ones that are known as being the worst faults to handle. In this paper we study, from a deterministic point of view, the task of Byzantine gathering in a network modeled as a graph. In other words, despite the presence of Byzantine agents, all the other (good) agents, starting from {possibly} different nodes and applying the same deterministic algorithm, have to meet at the same node in finite time and stop moving. An adversary chooses the initial nodes of the agents (the number of agents may be larger than the number of nodes) and assigns a different positive integer (called label) to each of them. Initially, each agent knows its label. The agents move in synchronous rounds and can communicate with each other only when located at the same node. Within the team, f of the agents are Byzantine. A Byzantine agent acts in an unpredictable and arbitrary way. For example, it can choose an arbitrary port when it moves, can convey arbitrary information to other agents and can change its label in every round, in particular by forging the label of another agent or by creating a completely new one. Besides its label, which corresponds to a local knowledge, an agent is assigned some global knowledge denoted by GK that is common to all agents. In literature, the Byzantine gathering problem has been analyzed in arbitrary n-node graphs by considering the scenario when GK=(n,f) and the scenario when GK=f. In the first (resp. second) scenario, it has been shown that the minimum number of good agents guaranteeing deterministic gathering of all of them is f+1 (resp. f+2). However, for both these scenarios, all the existing deterministic algorithms, whether or not they are optimal in terms of required number of good agents, have the major disadvantage of having a time complexity that is exponential in n and L, where L is the value of the largest label belonging to a good agent. In this paper, we seek to design a deterministic solution for Byzantine gathering that makes a concession on the proportion of Byzantine agents within the team, but that offers a significantly lower complexity. We also seek to use a global knowledge whose the length of the binary representation (that we also call size) is small. In this respect, assuming that the agents are in a strong team i.e., a team in which the number of good agents is at least some prescribed value that is quadratic in f, we give positive and negative results. On the positive side, we show an algorithm that solves Byzantine gathering with all strong teams in all graphs of size at most n, for any integers n and f, in a time polynomial in n and the length |l_{min}| of the binary representation of the smallest label of a good agent. The algorithm works using a global knowledge of size O(log log log n), which is of optimal order of magnitude in our context to reach a time complexity that is polynomial in n and |l_{min}|. Indeed, on the negative side, we show that there is no deterministic algorithm solving Byzantine gathering with all strong teams, in all graphs of size at most n, in a time polynomial in n and |l_{min}| and using a global knowledge of size o(log log log n).

The selection problem, where one wishes to locate the k th smallest element in an unsorted array of size n, is one of the basic problems studied in computer science. The main focus of this work is designing algorithms for solving the selection problem in the presence of memory faults.Specifically, the computational model assumed here is a faulty variant of the RAM model (abbreviated as FRAM), which was introduced by Finocchi and Italiano [FI04].. In this model, the content of memory cells might get corrupted adversarially during the execution, and the algorithm cannot distinguish between corrupted cells and uncorrupted cells. The model assumes a constant number of reliable memory cells that never become corrupted, and an upper bound δ on the number of corruptions that may occur, which is given as an auxiliary input to the algorithm.The main contribution of this work is a deterministic resilient selection algorithm with optimal O(n) worst-case running time. Interestingly, the running time does not depend on the number of faults, and the algorithm does not need to know δ. As part of the solution, several techniques that allow to sometimes use non-tail recursion algorithms in the FRAM model are developed. Notice that using recursive algorithms in this model is problematic, as the stack might be too large to fit in reliable memory.The aforementioned resilient selection algorithm can be used to improve the complexity bounds for resilient k-d trees developed by Gieseke, Moruz and Vahrenhold [GMV10]. Specifically, the time complexity for constructing a k-d tree is improved from O(nlog2 n + δ 2) to O(n logn).Besides the deterministic algorithm, a randomized resilient selection algorithm is developed, which is simpler than the deterministic one, and has O(n + α) expected time complexity and O(1) space complexity (i.e., is in-place). This algorithm is used to develop the first resilient sorting algorithm that is in-place and achieves optimal O(nlogn + αδ) expected running time.KeywordsSelection ProblemRecursive CallSoft ErrorType ChildProgram CounterThese keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

We consider the time and energy complexities of randomized leader election in a multiple-access channel, where the number of devices $n\geq 2$ is unknown. It is well-known that for polynomial-time randomized leader election algorithms with success probability $1-1/poly(n)$, the optimal energy complexity is $\Theta(\log\log^*n)$ if receivers can detect collisions, and $\Theta(\log^*n)$ otherwise. Without collision detection, all existing randomized leader election algorithms using $o(\log\log n)$ energy are Monte Carlo in that they may fail with some small probability, and they may consume unbounded energy and never halt when they fail. Though the optimal energy complexity of leader election appears to be settled, it is still an open question to attain the optimal $O(\log^*n)$ energy complexity by an efficient Las Vegas algorithm that never fails. In this paper we address this fundamental question. $\textbf{Separation between Monte Carlo and Las Vegas:}$ Without collision detection, we prove that any Las Vegas leader election algorithm with finite expected time complexity must use $\Omega(\log\log n)$ energy, establishing a large separation between Monte Carlo and Las Vegas algorithms. $\textbf{Exponential improvement with sender collision detection:}$ In the setting where senders can detect collisions, we design a new leader election algorithm that finishes in $O(\log^{1+\epsilon}n)$ time and uses $O(\epsilon^{-1}\log\log\log n)$ energy in expectation, showing that sender collision detection helps improve the energy complexity exponentially. $\textbf{Optimal deterministic leader election algorithm:}$ As a side result, via derandomization, we show a new deterministic algorithm that takes $O(n\log(N/n))$ time and $O(\log(N/n))$ energy to elect a leader from $n$ devices, where each device has a unique identifier in $[N]$. This algorithm is time-optimal and energy-optimal.

For the weighted rD-Matching problem, we present a deterministic parameterized algorithm with time complexity O *(4(r − 1)k ), improving the previous best upper bound O *(4 rk ). In particular, the algorithm can be applied to solve the unweighted 3D-Matching problem with time O *(16 k ), improving the previous best result O *(21.26 k ). For the weighted r-Set Packing problem, we present a deterministic parameterized algorithm with time complexity O *(2(2r − 1)k ), improving the previous best result O *(22rk ). The algorithm, when applied to the unweighted 3-Set Packing problem, has running time O *(32 k ), improving the previous best result O *(43.62 k ). Moreover, for the weighted rD-Matching and weighted r-Set Packing problems, we get a kernel of size O(k r ).

We present a deterministic near-linear time algorithm that computes the edge-connectivity and finds a minimum cut for a simple undirected unweighted graph G with n vertices and m edges. This is the first o(mn) time deterministic algorithm for the problem. In near-linear time we can also construct the classic cactus representation of all minimum cuts.The previous fastest deterministic algorithm by Gabow from STOC'91 took O(m+λ2 n), where λ is the edge connectivity, but λ could be Ω(n).At STOC'96 Karger presented a randomized near linear time Monte Carlo algorithm for the minimum cut problem. As he points out, there is no better way of certifying the minimality of the returned cut than to use Gabow's slower deterministic algorithm and compare sizes.Our main technical contribution is a near-linear time algorithm that contracts vertex sets of a simple input graph G with minimum degree δ, producing a multigraph G with ~O(m/δ) edges which preserves all minimum cuts of G with at least two vertices on each side.In our deterministic near-linear time algorithm, we will decompose the problem via low-conductance cuts found using PageRank a la Brin and Page (1998), as analyzed by Andersson, Chung, and Lang at FOCS'06. Normally such algorithms for low-conductance cuts are randomized Monte Carlo algorithms, because they rely on guessing a good start vertex. However, in our case, we have so much structure that no guessing is needed.

We explore a special case of rational identity testing and algorithmic versions of two theorems on noncommutative polynomials, namely, Amitsur's theorem [S.A Amitsur, 1966] and the Bresar-Klep theorem [Bresar and Klep, 2008] when the input polynomial is given by an algebraic branching program (ABP). Let f be a degree-d n-variate noncommutative polynomial in the free ring Q over rationals. 1) We consider the following special case of rational identity testing: Given a noncommutative ABP as white-box, whose edge labels are linear forms or inverses of linear forms, we show a deterministic polynomial-time algorithm to decide if the rational function computed by it is equivalent to zero in the free skew field Q . Given black-box access to the ABP, we give a deterministic quasi-polynomial time algorithm for this problem. 2) Amitsur's theorem implies that if a noncommutative polynomial f is nonzero on k x k matrices then, in fact, f(M_1,M_2,...,M_n) is invertible for some matrix tuple (M_1,M_2,...,M_n) in (M_k(ℚ))^n. While a randomized polynomial time algorithm to find such (M_1,M_2,...,M_n) given black-box access to f is simple, we obtain a deterministic s^{O(log d)} time algorithm for the problem with black-box access to f, where s is the minimum ABP size for f and d is the degree of f. 3) The Bresar-Klep Theorem states that the span of the range of any noncommutative polynomial f on k x k matrices over Q is one of the following: zero, scalar multiples of I_k, trace-zero matrices in M_k(Q), or all of M_k(Q). We obtain a deterministic polynomial-time algorithm to decide which case occurs, given white-box access to an ABP for f. We also give a deterministic s^{O(log d)} time algorithm given black-box access to an ABP of size s for f. Our algorithms work when k >= d. Our techniques are based on some automata theory combined with known techniques for noncommutative ABP identity testing [Ran Raz and Amir Shpilka, 2005; Michael A. Forbes and Amir Shpilka, 2013].

The existence of polynomial time algorithms for the solution of parity games is a major open problem. The fastest known algorithms for the problem are randomized algorithms that run in subexponential time. These algorithms are all ultimately based on the randomized subexponential simplex algorithms of Kalai and of Matousek, Sharir and Welzl. Randomness seems to play an essential role in these algorithms. We use a completely different, and elementary, approach to obtain a deterministic subexponential algorithm for the solution of parity games. Our deterministic algorithm is almost as fast as the randomized algorithms mentioned above.

In this paper, we proposed a medical image fusion algorithm based on sparse representation and guided filtering. One of attractive features in the algorithm is that it can preserve the structural information of structural image and color information of functional image. We use a sparse representation in low frequency, initialize the dictionary by using DCT transform, and train the dictionary with each input source image as a training example. It not only ensures time complexity, but also ensures that the low-frequency fusion rules are adaptive. At high frequencies, we use the method of guided filtering to extract structural information from high-frequency images, and use the injection method to fuse high-frequency sub-bands to ensure the validity and richness of structural information. Experimental results show that the proposed fusion algorithm is superior to comparative algorithms in terms of subjective and objective evaluation methods.

Real World is filled with various hard and complex problems. One such complex problem is an optimization problem. Optimization has been an active area of research for several decades. Optimized solutions are hard to find so there are no deterministic algorithms that can find exact solution in polynomial time. In large domain of applications of intelligence techniques we are interested in exploring the application of Biographical Based Optimization (BBO) and Grey Wolf Optimizer (GWO) meta-heuristic algorithm to the domain of software testing. The GWO mimics the leadership hierarchy and hunting mechanism of grey wolves in nature. In this paper we adapt BBO and GWO for test suite prioritization and minimization and evaluate their performance with other nature inspired meta-heuristics.

Maximum gap labelings of graphs

In August 2002, Agrawal, Kayal and Saxena announced the first deterministic and polynomial-time primality-testing algorithm. For an input n, the Agarwal-Kayal-Saxena (AKS) algorithm runs in time $\tilde{O}(\log^{7.5} n)$ (heuristic time $\tilde{O}(\log^6 n)$). Verification takes roughly the same amount of time. On the other hand, the Elliptic Curve Primality Proving algorithm (ECPP) runs in random heuristic time $\tilde{O}(\log^{6} n)$ (some variant has heuristic time complexity $\tilde{O}(\log^4 n)$) and generates certificates which can be easily verified. However, it is hard to analyze the provable time complexity of ECPP even for a small portion of primes. More recently, Berrizbeitia gave a variant of the AKS algorithm, in which some primes (of density $O({ 1 / {\log}^2\,n })$) cost much less time to prove than a general prime does. Building on these celebrated results, this paper explores the possibility of designing a randomized primality-proving algorithm based on the AKS algorithm. We first generalize Berrizbeitia's algorithm to one which has higher density ($\Omega( {1/{\log}\log n })$) of primes whose primality can be proved in time complexity $\tilde{O}(\log^{4} n)$. For a general prime, one round of ECPP is deployed to reduce its primality proof to the proof of a random easily proved prime, thus we achieve heuristic time complexity $\tilde{O}(\log^{4} n)$ for all primes.

Two mobile agents, starting from different nodes of a network at possibly different times, have to meet at the same node. This problem is known as rendezvous. Agents move in synchronous rounds. Each agent has a distinct integer label from the set $$\{1,\ldots ,L\}$${1,?,L}. Two main efficiency measures of rendezvous are its time (the number of rounds until the meeting) and its cost (the total number of edge traversals). We investigate tradeoffs between these two measures. A natural benchmark for both time and cost of rendezvous in a network is the number of edge traversals needed for visiting all nodes of the network, called the exploration time. Hence we express the time and cost of rendezvous as functions of an upper bound E on the time of exploration (where E and a corresponding exploration procedure are known to both agents) and of the size L of the label space. We present two natural rendezvous algorithms. Algorithm Cheap has cost O(E) (and, in fact, a version of this algorithm for the model where the agents start simultaneously has cost exactly E) and time O(EL). Algorithm Fast has both time and cost $$O(E\log L)$$O(ElogL). Our main contributions are lower bounds showing that, perhaps surprisingly, these two algorithms capture the tradeoffs between time and cost of rendezvous almost tightly. We show that any deterministic rendezvous algorithm of cost asymptotically E (i.e., of cost $$E+o(E)$$E+o(E)) must have time $$\varOmega (EL)$$Ω(EL). On the other hand, we show that any deterministic rendezvous algorithm with time complexity $$O(E\log L)$$O(ElogL) must have cost $$\varOmega (E\log L)$$Ω(ElogL).