Vocalizations can vary as a function of their context of production and provide an immediate measure of an animal's affective states. If vocal expression of emotions has been conserved throughout evolution, direct...
详细信息
Vocalizations can vary as a function of their context of production and provide an immediate measure of an animal's affective states. If vocal expression of emotions has been conserved throughout evolution, direct between-species comparisons using the same set of acoustic indicators should be possible. The present study used a machine learning algorithm (eXtreme Gradient Boosting [XGBoost]) to distinguish between contact calls indicating positive (pleasant) and negative (unpleasant) emotional valence, produced in various contexts by seven species of ungulates. With an accuracy of 89.49% (balanced accuracy: 83.90%), we found that the most important predictors of emotional valence were acoustic variables reflecting changes in duration, energy quartiles, fundamental frequency, and amplitude modulation. This approach is critical in the field of emotional communication, where more information is needed to reach a better understanding of the emotional origins of human language. In addition, these results can help toward the development of automated tools for animal well-being monitoring.
algorithms for locating accessory pathways (APs) are widely used in planning catheter ablation for patients with ventricular preexcitation. These algorithms typically rely on characteristics of the delta wave and QRS ...
详细信息
Preserving individual privacy is crucial when interacting with Large Language Models (LLMs) during both training and inference stages. Privacy leakage at either stage can lead to irreversible negative consequences. Al...
详细信息
ISBN:
(纸本)9798400712746
Preserving individual privacy is crucial when interacting with Large Language Models (LLMs) during both training and inference stages. Privacy leakage at either stage can lead to irreversible negative consequences. Although data-level privacy-preserving algorithms have been developed for smaller Natural Language Processing (NLP) models, their application to LLMs has not been extensively explored. Moreover, with plenty of algorithms emerging, it brings challenges for organizations or researchers to compare and evaluate these different algorithms to select the most suitable one for their certain requirements. To address these challenges, we introduce ''Privacy-preserving4LLM Benchmarking'', a systematic evaluation framework that systematically assesses different privacy-preserving algorithms' utility-privacy trade-offs across different LLM architectures. Our framework evaluates these algorithms in three practical scenarios: protecting training data only, user queries only, and both. We also introduce a novel Parameter Optimizer to ensure fair comparisons. To quantify privacy protection levels, we use exposure metrics, where canary data sequences are intentionally inserted into training data to measure information memorization and potential leakage. Our study presents a comprehensive empirical analysis comparing three privacy-preserving algorithms across three LLM architectures (Mistral-7B, Llama2-7b, Falcon-7b) using three different datasets. Our findings reveal that algorithm selection, protection scenarios, LLM architectures, and privacy budget settings all impact the utility and privacy level.
Educational games have emerged as powerful tools for enhancing learning experiences across various subjects and age groups. This project represents an innovative approach to combining educational content with highly i...
详细信息
ISBN:
(纸本)9798400705328
Educational games have emerged as powerful tools for enhancing learning experiences across various subjects and age groups. This project represents an innovative approach to combining educational content with highly interactive and engaging gameplay to teach important Computer Science concepts such as data structures and algorithms. Existing implementations offer visualizations of algorithms and data structures, but lack the interactivity required for active participation of students. This project aims to foster active learning by allowing students to solve examples through directly selecting, swapping and manipulating nodes and edges of graphs. This game presents a platform for college students to learn and actively practice the Heap data structure, Heapsort algorithm, and minimum weight spanning tree algorithms such as Kruskal and Prim. Built using JavaScript, HTML, CSS, and ***, the game incorporates a 3D interactive environment to pilot these concepts. The platform includes two modes: a training mode, allowing penalty-free practice, and a regular mode with a scoring system, a leaderboard and limited lives to foster engagement, challenge and competitiveness with peers. Multiple levels of difficulty, immediate feedback, and a carefully tailored hint system were designed to support stepwise learning and enhance interactive experience. Anonymous Data collection of number of mistakes, time spent and change of score per level, runs in the background without storing identifying information on the students. An interface for instructors allows professors to visualize students' data, track improvement, and analyze students' comprehension and performance. Gamification of additional algorithms is currently being developed.
The article explores the role of algorithmic procedures in the implementation of the programme of precision medicine (PM), currently pursued in molecular tumour boards (MTBs) that emerged from the confluence of previo...
详细信息
In general the problem of verifying whether a structured business process is compliant with a given set of regulations is NP-hard. The present paper focuses on identifying a tractable subset of this problem, namely ve...
详细信息
In general the problem of verifying whether a structured business process is compliant with a given set of regulations is NP-hard. The present paper focuses on identifying a tractable subset of this problem, namely verifying whether a structured business process is compliant with a single global obligation. Global obligations are those whose validity spans for the entire execution of a business process. We identify two types of obligations： achievement and maintenance.
In the present paper we firstly define an abstract framework capable to model the problem and secondly we define procedures and algorithms to deal with the compliance problem of checking the compliance of a structured business process with respect to a single global obligation. We show that the algorithms proposed in the paper run in polynomial time.
The algebras considered in this paper are commutative rings of which the additive group is a finite-dimensional vector space over the field of rational numbers. We present deterministic polynomial-time algorithms that...
详细信息
The algebras considered in this paper are commutative rings of which the additive group is a finite-dimensional vector space over the field of rational numbers. We present deterministic polynomial-time algorithms that, given such an algebra, determine its nilradical, all of its prime ideals, as well as the corresponding localizations and residue class fields, its largest separable subalgebra, and its primitive idempotents. We also solve the discrete logarithm problem in the multiplicative group of the algebra. While deterministic polynomial-time algorithms were known earlier, our approach is different from previous ones. One of our tools is a primitive element algorithm;it decides whether the algebra has a primitive element and, if so, finds one, all in polynomial time. A methodological novelty is the use of derivations to replace a Hensel-Newton iteration. It leads to an explicit formula for lifting idempotents against nilpotents that is valid in any commutative ring.
In this paper, we consider the problems for covering multiple intervals on a line. Given a set B of m line segments (called "barriers") on a horizontal line L and another set S of n horizontal line segments ...
详细信息
In this paper, we consider the problems for covering multiple intervals on a line. Given a set B of m line segments (called "barriers") on a horizontal line L and another set S of n horizontal line segments of the same length in the plane, we want to move all segments of S to L so that their union covers all barriers and the maximum movement of all segments of S is minimized. Previously, an O(n(3) log n)-time algorithm was given for the case m =1. In this paper, we propose an O(n(2) log n log log n+ nm log m)-time algorithm for a mare general setting with any m >= 1, which also improves the previous work when m =1. We then consider a line-constrained version of the problem in which the segments of S are all initially on the line I.. Previously, an O(n log n)-time algorithm was known for the case m = 1. We present an algorithm of O(m log m + n log m log n) time for any m >= 1. These problems may have applications in mobile sensor barrier coverage in wireless sensor networks. (C) 2018 Elsevier B.V. All rights reserved.
In this paper we study the computational aspects of coherence and extension of partial possibility assessments, both in an unconditional and a conditional setting, providing complexity results and algorithms for each ...
详细信息
In this paper we study the computational aspects of coherence and extension of partial possibility assessments, both in an unconditional and a conditional setting, providing complexity results and algorithms for each problem. In particular, we propose an algorithm to check the coherence of a partial unconditional assessment which is based on propositional satisfiability. For the conditional case, we firstly prove a new characterization of coherent conditional assessments that allows us to define an algorithm again based on propositional satisfiability. The extension problem, in both settings, is solved by means of a search algorithm which relies on the corresponding coherence procedure. (c) 2011 Elsevier B.V. All rights reserved.
This opinion piece explores the issues of accountability and transparency in relation to the growing use of machine learning algorithms. Citing the recent work of the Royal Society and the British Academy, it looks at...
详细信息
This opinion piece explores the issues of accountability and transparency in relation to the growing use of machine learning algorithms. Citing the recent work of the Royal Society and the British Academy, it looks at the legal protections for individuals afforded by the EU General Data Protection Regulation and asks whether the legal system will be able to adapt to rapid technological change. It concludes by calling for continuing debate that is itself accountable, transparent and public. This article is part of a discussion meeting issue 'The growing ubiquity of algorithms in society: implications, impacts and innovations'.
暂无评论