arXiv subjects

Akihiro Nishi

Publications and source records attributed to Akihiro Nishi.

7 recordsLinked to original sources

arXiv2026

Temporal dependence in exposure and hazard-based infectious disease interventions

In randomized controlled trials (RCTs) of infectious disease interventions, it is well recognized that unmeasured individual heterogeneity at baseline can induce selection bias over time, thereby complicating the interpretation of the estimated hazard ratio. The present study examines a simplified setting: RCTs consisting of homogeneous participants, with no individual heterogeneity at baseline. However, even in such an apparently ideal setting, selection bias can emerge over time due to temporal dependence in exposure, a realistic feature of infectious disease transmission. In this study, we mathematically characterize the mechanism underlying this bias and quantitatively evaluate its magnitude. Our results show that this bias should be recognized as an issue in both the design and interpretation of RCTs of infectious disease interventions.

stat.ME↗

arXiv2025

Statistical Modeling of Networked Evolutionary Public Goods Games

Repeated small dynamic networks are integral to studies in evolutionary game theory, where networked public goods games offer novel insights into human behaviors. Building on these findings, it is necessary to develop a statistical model that effectively captures dependencies across multiple small dynamic networks. While Separable Temporal Exponential-family Random Graph Models (STERGMs) have demonstrated success in modeling a large single dynamic network, their application to multiple small dynamic networks with less than 10 actors, remains unexplored. In this study, we extend the STERGM framework to accommodate multiple small dynamic networks, offering an approach to analyzing such systems. Taking advantage of the small network sizes, our proposed approach improves accuracy in statistical inference through direct computation, unlike conventional approaches that rely on Markov Chain Monte Carlo methods. We demonstrate the validity of this framework through the analysis of a networked public goods experiment into individual decision-making about cooperation and defection. The resulting statistical inference uncovers novel insights into the dynamics of social dilemmas, showcasing the effectiveness and robustness of this modeling and approach.

stat.AP↗

arXiv2025

Making the Peers' Subjective Well-being Visible Impairs Cooperator-centered Experimental Social Networks

Past experiments show that reputation or the knowledge of peers' past cooperation can enhance cooperation in human social networks. On the other hand, the knowledge of peers' wealth undermines cooperativeness, and that of peers' interconnectedness and network structure does not affect it. However, it is unknown if making peers' subjective well-being (SWB) available or visible in social networks may enhance or undermine cooperation. Therefore, we implemented online network experiments (N = 662 in 50 networked groups with 15 rounds of interactions), in which study participants cooperated with or defected against connected peers through Public Goods Game, made and cut social ties with others, and rated their SWB. We manipulated the visibility of connected peers' SWB (25 visible vs. 25 invisible SWB networked groups) while keeping the connected peers' reputation and in-game wealth visible. Results show that making the peers/ SWB visible did not alter overall cooperativeness, wealth, inter-connectedness, or SWB. In contrast, the visible SWB networked groups exhibited a higher number of communities and lower transitivity (the proportion of the cases where a peer of a peer is also a peer) than the invisible SWB networked groups. These phenomena are explained by an altered decision-making pattern in the visible SWB networks: cooperators were less likely to connect with cooperators and more likely to connect with defectors, and consequently, cooperators could not maintain their popularity or stay in the center of the networks.

cs.SI↗

arXiv2022

On Measures of Biases and Harms in NLP

Recent studies show that Natural Language Processing (NLP) technologies propagate societal biases about demographic groups associated with attributes such as gender, race, and nationality. To create interventions and mitigate these biases and associated harms, it is vital to be able to detect and measure such biases. While existing works propose bias evaluation and mitigation methods for various tasks, there remains a need to cohesively understand the biases and the specific harms they measure, and how different measures compare with each other. To address this gap, this work presents a practical framework of harms and a series of questions that practitioners can answer to guide the development of bias measures. As a validation of our framework and documentation questions, we also present several case studies of how existing bias measures in NLP -- both intrinsic measures of bias in representations and extrinsic measures of bias of downstream applications -- can be aligned with different harms and how our proposed documentation questions facilitates more holistic understanding of what bias measures are measuring.

cs.CL↗

arXiv2022

Socially Aware Bias Measurements for Hindi Language Representations

Language representations are efficient tools used across NLP applications, but they are strife with encoded societal biases. These biases are studied extensively, but with a primary focus on English language representations and biases common in the context of Western society. In this work, we investigate biases present in Hindi language representations with focuses on caste and religion-associated biases. We demonstrate how biases are unique to specific language representations based on the history and culture of the region they are widely spoken in, and how the same societal bias (such as binary gender-associated biases) is encoded by different words and text spans across languages. The discoveries of our work highlight the necessity of culture awareness and linguistic artifacts when modeling language representations, in order to better understand the encoded biases.

cs.CL↗

arXiv2020

Towards Efficient Interactive Computation of Dynamic Time Warping Distance

The dynamic time warping (DTW) is a widely-used method that allows us to efficiently compare two time series that can vary in speed. Given two strings $A$ and $B$ of respective lengths $m$ and $n$, there is a fundamental dynamic programming algorithm that computes the DTW distance for $A$ and $B$ together with an optimal alignment in $Θ(mn)$ time and space. In this paper, we tackle the problem of interactive computation of the DTW distance for dynamic strings, denoted $\mathrm{D^2TW}$, where character-wise edit operation (insertion, deletion, substitution) can be performed at an arbitrary position of the strings. Let $M$ and $N$ be the sizes of the run-length encoding (RLE) of $A$ and $B$, respectively. We present an algorithm for $\mathrm{D^2TW}$ that occupies $Θ(mN+nM)$ space and uses $O(m+n+\#_{\mathrm{chg}}) \subseteq O(mN + nM)$ time to update a compact differential representation $\mathit{DS}$ of the DP table per edit operation, where $\#_{\mathrm{chg}}$ denotes the number of cells in $\mathit{DS}$ whose values change after the edit operation. Our method is at least as efficient as the algorithm recently proposed by Froese et al. running in $Θ(mN + nM)$ time, and is faster when $\#_{\mathrm{chg}}$ is smaller than $O(mN + nM)$ which, as our preliminary experiments suggest, is likely to be the case in the majority of instances.

cs.DS↗

arXiv2018

$O(n \log n)$-time text compression by LZ-style longest first substitution

Mauer et al. [A Lempel-Ziv-style Compression Method for Repetitive Texts, PSC 2017] proposed a hybrid text compression method called LZ-LFS which has both features of Lempel-Ziv 77 factorization and longest first substitution. They showed that LZ-LFS can achieve better compression ratio for repetitive texts, compared to some state-of-the-art compression algorithms. The drawback of Mauer et al.'s method is that their LZ-LFS compression algorithm takes $O(n^2)$ time on an input string of length $n$. In this paper, we show a faster LZ-LFS compression algorithm that works in $O(n \log n)$ time. We also propose a simpler version of LZ-LFS that can be computed in $O(n)$ time.

cs.DS↗