Building defect prediction models based on online learning can enhance prediction accuracy. It continuously rebuilds a new prediction model while adding new data points. However, a module predicted as “non-defective” can result in fewer test cases for such modules. Thus, a defective module can be overlooked during testing. The erroneous test results are used as learning data by online learning, which could negatively affect prediction accuracy. To suppress the negative influence, we propose to apply a method that fixes the prediction as positive during the initial stage of online learning. Additionally, we improved the method to consider the probability of defect overlooking. In our experiment, we demonstrate this negative influence on prediction accuracy and the effectiveness of our approach. The results show that our approach did not negatively affect AUC but significantly improved recall.

Cross-project defect prediction (CPDP) aims to use data from external projects as historical data may not be available from the same project. In CPDP, deciding on a particular historical project to build a training model can be difficult. To help with this decision, a Bandit Algorithm (BA) based approach has been proposed in prior research to select the most suitable learning project. However, this BA method could lead to the selection of unsuitable data during the early iteration of BA (i.e., early stage of software testing). Selecting an unsuitable model can reduce the prediction accuracy, leading to potential defect overlooking. This study aims to improve the BA method to reduce defects overlooking, especially during the early testing stages. Once all modules have been tested, modules tested in the early stage are re-predicted, and some modules are retested based on the re-prediction. To assess the impact of re-prediction and retesting, we applied five kinds of BA methods, using 8, 16, and 32 OSS projects as learning data. The results show that the newly proposed approach steadily reduced the probability of defect overlooking without degradation of prediction accuracy.

A floorplan is a partition of an axis-aligned rectangle into a set of smaller rectangles. Given an axis-aligned rectangle R and a set P of points in R we wish to partition R into a set S of rectangles so that each point in P is on a boundary of a rectangle in S. We call such a partition of R a floorplan covering P. Intuitively P is the locations of columns and a floorplan covering P is a floorplan in which no column is in the proper inside of a room and each column is on a wall. In this paper we design an algorithm to generate all floorplans covering P when P is the set of given grid points. The algorithm generates each floorplan in O(|P|) time.

A floorplan is a partition of an axis-aligned rectangle into a set of smaller rectangles. Given an integer e we regard a floorplan is desirable for its safety if each room has an escape route to a room facing the outside, where the route passes through at most e walls, and simple, which is, directs to only either (1) south or east, (2) east or north, (3) north or west or (4) west or south. In this paper we design an algorithm to generate all floorplans with exactly n rooms and satisfying the property above. Our algorithm generates all such floorplans in O(n²) time for each.

Constructing a suitable schedule for sports competitions is a crucial issue in sports scheduling. The round-robin tournament is a competition adopted in many professional sports. For most round-robin tournaments, it is considered undesirable that a team plays consecutive away or home matches; such an occurrence is called a break. Accordingly, it is preferable to reduce the number of breaks in a tournament. A common approach is to first construct a schedule and then determine a home-away assignment based on the given schedule to minimize the number of breaks (first-schedule-then-break). In this study, we concentrate on the problem that arises at the second stage of the first-schedule-then-break approach, namely, the break minimization problem (BMP). We propose a novel integer linear programming formulation called the “bigram based formulation.” The computational experiments show its effectiveness over the well-known integer linear programming formulation. We also investigate its valid inequalities, which further enhances the computational performance.

Yajisan-Kazusan and Stained Glass are Nikoli’s pencil puzzles. We study the computational complexity of Yajisan-Kazusan and Stained Glass puzzles. It is shown that deciding whether a given instance of each puzzle has a solution is NP-complete.

The degeneracy of a graph G is defined as the smallest value k such that every subgraph of G has a vertex with a degree of at most k. Given a graph G, its degeneracy can be easily calculated provided sufficient memory is available. In this paper, we focus on scenarios where only o(n) bits of additional read-write memory are available, apart from the input stored in read-only memory. Within this context, we introduce two FPT algorithms for degeneracy, parameterized by neighborhood diversity and the cluster vertex deletion number.

The exact matching problem is a constrained variant of the maximum matching problem: given a graph with each edge having a weight 0 or 1 and an integer k, the goal is to find a perfect matching of weight exactly k. Mulmuley, Vazirani, and Vazirani (1987) proposed a randomized polynomial-time algorithm for this problem, and it is still open whether it can be derandomized. Very recently, El Maalouly, Steiner, and Wulf (2023) showed that for bipartite graphs there exists a deterministic FPT algorithm parameterized by the (bipartite) independence number. In this paper, by extending a part of their work, we propose a deterministic FPT algorithm in general parameterized by the minimum size of an odd cycle transversal in addition to the (bipartite) independence number. We also consider a relaxed problem called the correct parity matching problem, and show that a slight generalization of an equivalent problem is NP-hard.

Metarounding is an approach to convert an approximation algorithm for linear optimization over some combinatorial classes to an online linear optimization algorithm for the same class. We propose a new metarounding algorithm under a natural assumption that a relax-based approximation algorithm exists for the combinatorial class. Our algorithm is much more efficient in both theoretical and practical aspects.

Recently, we introduced and investigated a colored variant of finite automata, so-called “colored finite automata.” Its accepting states are able to be differently colored each and therefore a single automaton can classify and distinguish multiple languages at once. In this paper, we further extend the concept of colored accepting states and propose a new automaton, called bilaterally colored finite automaton (biCFA) which can possess as many differently colored initial states as possible, rather than a specified single initial state. We next introduce its regular expression counterpart, called bilaterally colored regular expression (biCRE), which exactly expresses the same tuple of languages as accepted by the corresponding biCFA. Notably, the mono-colored version of colored regular expression is a succinct and intuitive alternative of the existing ordinary regular expression. We also demonstrate the usefulness and feasibility of biCRE by applying the concept to extended (i.e., regular right part) context-free grammar and exhibit a super-short pure Python program which parses basic arithmetic expressions with addition and multiplication operators.

Game theory on graphs is a basic tool in computer science. In this paper, we propose a new game-theoretic framework for studying the privacy protection of a user who interactively uses a software service. Our framework is based on the idea that an objective of a user using software services should not be known to an adversary because the objective is often closely related to personal information of the user. We propose two new notions, O-indistinguishable strategy (O-IS) and objective-indistinguishability equilibrium (OIE). For a given game and a subset O of winning objectives (or objectives in short), a strategy of a player is O-indistinguishable if an adversary cannot shrink O by excluding any objective from O as an impossible objective. A strategy profile, which is a tuple of strategies of all players, is an OIE if the profile is locally optimal in the sense that no player can expand her set of objectives indistinguishable from her real objective from the viewpoint of an adversary. We analyze the complexities of deciding the existence of O-IS and prove the decidability of the existence of OIE under a weaker assumption on rationality.

A polygon obtained by cutting the surface of a polyhedron is called an unfolding. An unfolding obtained by cutting along only edges is called an edge unfolding. An unfolding may have overlapping, which are self-intersections on its boundary. It is a well-known open question in computational origami whether or not every convex polyhedron has a non-overlapping edge unfolding. On the other hand, Sharir and Schorr showed that any convex polyhedron could unfold without overlapping when allowed to cut its faces. Therefore, there is a gap between edge unfoldings and general unfoldings. Bridging this gap is necessary as a foothold on this open question of edge unfolding. Instead of cutting faces arbitrarily, there are studies considering whether specific cutting lines on the faces can result in unfoldings without overlaps. Lattice unfoldings of a cuboid made by unit cubes are one such example. A lattice unfolding of a cuboid is a polygon obtained by cutting the faces along the edges of unit squares. An unfolding may have overlapping, even in the case of small cuboids. In particular, Uno showed that a (1, 1, 3)-cuboid with dimensions (height, width, depth) has an overlapping lattice unfolding, while Mitani and Uehara showed the same for three faces of a (1, 2, 3)-cuboid. In contrast, it is known that some cuboids have no overlapping lattice unfolding. Hearn showed it for a (1, 1, 2)-cuboid, and Sugihara showed the same for a (2, 2, 2)-cuboid. In this study, we completely clarify the existence of overlapping lattice unfoldings, which also contains the case where the sizes are non-integers.

We address the problem of calculating the least core value of the routing game (the traveling salesman game with a fixed route) without the assumption of triangle inequalities. We propose a polynomial size LP formulation for finding a payoff vector in the least core.

We present a flip sequence of length ⎡(15/14)n+2⎤ for sorting the Heydari and Sudborough’s stack of n pancakes, which was introduced to prove the best-known lower bound of (15/14)n for the pancake number of n pancakes.

Stylized image captioning is the task of generating image captions that have a description style, such as positive or negative sentiments. Recently, deep learning models have reached high performance in this task, but they still lack description accuracy and diversity, and they often suffer from the small size and the low descriptiveness of existing datasets. In this paper, we introduce a new dataset, UTStyleCap4K, which contains 4,644 images with three positive and three negative captions for every image (27,864 captions in total), collected by a crowdsourcing service. Experimental results show that our dataset is accurate in meaning and sentiments, diverse in the ways to describe the styles, and less similar to the base dataset, the MSCOCO dataset, than existing stylized image captioning datasets. We train multiple models on our dataset to set a baseline. We also propose a new Bidirectional Encoder Representations from Transformers (BERT) based model, StyleCapBERT, that controls the length and style of the generated captions at the same time, by introducing length and style information into the embeddings of caption words. Experimental results show that our model is capable of generating captions of three sentimental styles, positive, factual, and negative, at the same time, and achieving the best performance on our dataset.

To address the limitation of annotated datasets confined to fixed relation domains, which hampers the effective extraction of triplets, especially for novel relation types, our work introduces an innovative approach. We propose a method for training large-scale language models using prompt templates designed for zero-shot learning in relation triplet extraction tasks. By utilizing these specially crafted prompt templates in combination with fine-grained matching scoring rules, we transform the structured prediction task into a cloze task. This transformation aligns the task more closely with the intrinsic capabilities of the language model, facilitating a more natural processing flow.Experimental evaluations on two public datasets show that our method achieves stable and enhanced performance compared to baseline models. This improvement underscores the efficiency and potential of our approach in facilitating zero-shot extraction of relation triplets, thus broadening the scope of applicable relation types without the need for domain-specific training data.

Negotiation is an essential form of social interaction; however, reaching a reasonable agreement in negotiations is difficult. This is especially due to the complexity of the issues and backgrounds of the negotiations. In addition, when the interests of the negotiators conflict, the negotiation may break down. Therefore, machine learning and natural language processing must be used to support negotiations in real life. However, few studies have reported the use of such approaches. In this study, we propose the task of anticipating final disagreements in negotiation mid-dialogues. By anticipating final disagreements in negotiations, we can support human negotiations by understanding the elements necessary for negotiations to reach an agreement and building a system that supports negotiations to reach an agreement. We aim to reduce the training time by fine-tuning a dialogue act-based GRU model and a text-based BERT pre-trained on a large-scale dataset with different datasets compared to training on the target dataset alone. The results show that fine-tuning improves the predictive performance of machine learning models that detect disagreements in the early stages of negotiations, when the input process dialogue is in the middle of the negotiation. A comparative experiment between a dialogue act-based GRU model and a text-based BERT shows that the dialogue act-based GRU model is more accurate than the text-based BERT model. We also visualize the final attention layer of the BERT model for further analysis.

Feature-based detection method has been widely used for hardware trojan detection where hardware trojan features are explicitly extracted and trained a classifier or build an algorithm that can differentiate hardware trojan signals from normal ones. This method shows a good performance in the available benchmark circuits. However, when tested on a randomly generated circuit, that contains different structural features of benchmark circuit, it started to fail detecting the stealthy signals, because of the constraints when the features are explicitly extracted. To overcome this situation, in this paper, the authors aim to explore the benefit of graph neural network (GNN) to enhance the hardware trojan detection performance in both benchmark circuits and randomly generated circuits. More specifically, finding the appropriate representation of the digital circuit that is suitable for the GNN model and that can learn hidden feature and relationship between the signals. Experiments show satisfactory result on benchmark circuit with an average accuracy of 95.34%. Further, tests done on randomly generated circuits gives positive result with an average accuracy of 90.82%.