Search

In the training and inference of spiking neural networks (SNNs), direct training and lightweight computation methods have been orthogonally developed, aimed at reducing power consumption. However, only a limited number of approaches have applied these two mechanisms simultaneously and failed to fully leverage the advantages of SNN-based vision transformers (ViTs) since they were originally designed for convolutional neural networks (CNNs). In this paper, we propose AT-SNN designed to dynamically adjust the number of tokens processed during inference in SNN-based ViTs with direct training, wherein power consumption is proportional to the number of tokens. We first demonstrate the applicability of adaptive computation time (ACT), previously limited to RNNs and ViTs, to SNN-based ViTs, enhancing it to discard less informative spatial tokens selectively. Also, we propose a new token-merge mechanism that relies on the similarity of tokens, which further reduces the number of tokens while enhancing accuracy. We implement AT-SNN to Spikformer and show the effectiveness of AT-SNN in achieving high energy efficiency and accuracy compared to state-of-the-art approaches on the image classification tasks, CIFAR10, CIFAR-100, and TinyImageNet. For example, our approach uses up to 42.4% fewer tokens than the existing best-performing method on CIFAR-100, while conserving higher accuracy., Comment: 8 pages

Estimating power consumption in modern Cloud environments is essential for carbon quantification toward green computing. Specifically, it is important to properly account for the power consumed by each of the running applications, which are packaged as containers. This paper examines multiple challenges associated with this goal. The first challenge is that multiple customers are sharing the same hardware platform (multi-tenancy), where information on the physical servers is mostly obscured. The second challenge is the overhead in power consumption that the Cloud platform control plane induces. This paper addresses these challenges and introduces a novel pipeline framework for power model training. This allows versatile power consumption approximation of individual containers on the basis of available performance counters and other metrics. The proposed model utilizes machine learning techniques to predict the power consumed by the control plane and associated processes, and uses it for isolating the power consumed by the user containers, from the server power consumption. To determine how well the prediction results in an isolation, we introduce a metric termed isolation goodness. Applying the proposed power model does not require online power measurements, nor does it need information on the physical servers, configuration, or information on other tenants sharing the same machine. The results of cross-workload, cross-platform experiments demonstrated the higher accuracy of the proposed model when predicting power consumption of unseen containers on unknown platforms, including on virtual machines., Comment: This is a full-version (8-page) paper of our previous publication in IEEE MASCOTS 2023, which has been accepted as a 4-page short paper (https://ieeexplore.ieee.org/document/10387542)

The widespread adoption and transformative effects of large language models (LLMs) have sparked concerns regarding their capacity to produce inaccurate and fictitious content, referred to as `hallucinations'. Given the potential risks associated with hallucinations, humans should be able to identify them. This research aims to understand the human perception of LLM hallucinations by systematically varying the degree of hallucination (genuine, minor hallucination, major hallucination) and examining its interaction with warning (i.e., a warning of potential inaccuracies: absent vs. present). Participants (N=419) from Prolific rated the perceived accuracy and engaged with content (e.g., like, dislike, share) in a Q/A format. Participants ranked content as truthful in the order of genuine, minor hallucination, and major hallucination, and user engagement behaviors mirrored this pattern. More importantly, we observed that warning improved the detection of hallucination without significantly affecting the perceived truthfulness of genuine content. We conclude by offering insights for future tools to aid human detection of hallucinations. All survey materials, demographic questions, and post-session questions are available at: https://github.com/MahjabinNahar/fakes-of-varying-shades-survey-materials, Comment: Accepted at COLM 2024

Large language models (LLMs) have revolutionized the state-of-the-art of many different natural language processing tasks. Although serving LLMs is computationally and memory demanding, the rise of Small Language Models (SLMs) offers new opportunities for resource-constrained users, who now are able to serve small models with cutting-edge performance. In this paper, we present a set of experiments designed to benchmark SLM inference at performance and energy levels. Our analysis provides a new perspective in serving, highlighting that the small memory footprint of SLMs allows for reaching the Pareto-optimal throughput within the resource capacity of a single accelerator. In this regard, we present an initial set of findings demonstrating how model replication can effectively improve resource utilization for serving SLMs., Comment: It is going to be published at EuroMLSys'24

Federated learning (FL) is a machine learning paradigm that allows multiple clients to collaboratively train a shared model while keeping their data on-premise. However, the straggler issue, due to slow clients, often hinders the efficiency and scalability of FL. This paper presents FedCore, an algorithm that innovatively tackles the straggler problem via the decentralized selection of coresets, representative subsets of a dataset. Contrary to existing centralized coreset methods, FedCore creates coresets directly on each client in a distributed manner, ensuring privacy preservation in FL. FedCore translates the coreset optimization problem into a more tractable k-medoids clustering problem and operates distributedly on each client. Theoretical analysis confirms FedCore's convergence, and practical evaluations demonstrate an 8x reduction in FL training time, without compromising model accuracy. Our extensive evaluations also show that FedCore generalizes well to existing FL frameworks.

Real-world network datasets are typically obtained in ways that fail to capture all edges. The patterns of missing data are often non-uniform as they reflect biases and other shortcomings of different data collection methods. Nevertheless, uniform missing data is a common assumption made when no additional information is available about the underlying missing-edge pattern, and link prediction methods are frequently tested against uniformly missing edges. To investigate the impact of different missing-edge patterns on link prediction accuracy, we employ 9 link prediction algorithms from 4 different families to analyze 20 different missing-edge patterns that we categorize into 5 groups. Our comparative simulation study, spanning 250 real-world network datasets from 6 different domains, provides a detailed picture of the significant variations in the performance of different link prediction algorithms in these different settings. With this study, we aim to provide a guide for future researchers to help them select a link prediction algorithm that is well suited to their sampled network data, considering the data collection process and application domain., Comment: Submitted to PLOS ONE

In cultured meat (CM) products the paramount significance lies in the fundamental attributes like texture and sensory of the processed end product. To cater to the tactile and gustatory preferences of real meat, the product needs to be designed to incorporate its texture and sensory attributes. Presently CM products are mainly grounded products like sausage, nugget, frankfurter, burger patty, surimi, and steak with less sophistication and need to mimic real meat to grapple with the traditional meat market. The existence of fibrous microstructure in connective and muscle tissues has attracted considerable interest in the realm of tissue engineering. Scaffolding plays an important role in CM production by aiding cell adhesion, growth, differentiation, and alignment. A wide array of scaffolding technologies has been developed for implementation in the realm of biomedical research. In recent years researchers also focus on edible scaffolding to ease the process of CM. However, it is imperative to implement cutting edge technologies like 3D scaffolds, 3D printing, electrospun nanofibers in order to advance the creation of sustainable and edible scaffolding methods in CM production, with the ultimate goal of replicating the sensory and nutritional attributes to mimic real meat cut. This review discusses recent advances in scaffolding techniques and biomaterials related to structured CM production and required advances to create muscle fiber structures to mimic real meat. Keywords: Cultured meat, Scaffolding, Biomaterials, Edible scaffolding, Electrospinning, 3D bioprinting, real meat.

In this paper we prove the analytic continuation of a two variable zeta function defined using the vector space of binary forms of degree $d$ to the entire two dimensional complex space as a meromorphic function.

The absolute depth values of surrounding environments provide crucial cues for various assistive technologies, such as localization, navigation, and 3D structure estimation. We propose that accurate depth estimated from panoramic images can serve as a powerful and light-weight input for a wide range of downstream tasks requiring 3D information. While panoramic images can easily capture the surrounding context from commodity devices, the estimated depth shares the limitations of conventional image-based depth estimation; the performance deteriorates under large domain shifts and the absolute values are still ambiguous to infer from 2D observations. By taking advantage of the holistic view, we mitigate such effects in a self-supervised way and fine-tune the network with geometric consistency during the test phase. Specifically, we construct a 3D point cloud from the current depth prediction and project the point cloud at various viewpoints or apply stretches on the current input image to generate synthetic panoramas. Then we minimize the discrepancy of the 3D structure estimated from synthetic images without collecting additional data. We empirically evaluate our method in robot navigation and map-free localization where our method shows large performance enhancements. Our calibration method can therefore widen the applicability under various external conditions, serving as a key component for practical panorama-based machine vision systems. Code is available through the following link: \url{https://github.com/82magnolia/panoramic-depth-calibration}., Comment: Accepted to ICCV 2023

Health outcomes, such as body mass index and cholesterol levels, are known to be dependent on age and exhibit varying effects with their associated risk factors. In this paper, we propose a novel framework for dynamic modeling of the associations between health outcomes and risk factors using varying-coefficients (VC) regional quantile regression via K-nearest neighbors (KNN) fused Lasso, which captures the time-varying effects of age. The proposed method has strong theoretical properties, including a tight estimation error bound and the ability to detect exact clustered patterns under certain regularity conditions. To efficiently solve the resulting optimization problem, we develop an alternating direction method of multipliers (ADMM) algorithm. Our empirical results demonstrate the efficacy of the proposed method in capturing the complex age-dependent associations between health outcomes and their risk factors.

We establish a sharp inequality between the blocks of positive partitioned matrices and conjecture a triangle type inequality for contractions: Given three contactions A,B,C, we conjecture that the constant c=3/4 is sharp in the triangle inequality |A+B+C| < cI + |A| +|B| + |C|., Comment: Final version, to appear in LAA, 13 pages

Ballet, a mainstream performing art predominantly associated with women, exhibits significant gender imbalances in leading positions. However, the collaboration's structural composition on gender representation in the field remains unexplored. Our study investigates the gendered labor force composition and collaboration patterns in ballet creations. Our findings reveal gender disparities in ballet creations aligned with gendered collaboration patterns and women occupying more peripheral network positions respect to men. Productivity disparities show women accessing 20-25\% of ballet creations compared to men. Mathematically derived perception errors show the underestimation of women artists' representation within ballet collaboration networks, potentially impacting women's careers in the field. Our study highlights the structural disadvantages that women face in ballet and emphasizes the need for a more inclusive and equal professional environment to improve the career development of women in the ballet industry. These insights contribute to a broader understanding of structural gender imbalances in artistic domains and can inform cultural organizations about potential affirmative actions towards a better representation of women leaders in ballet., Comment: 29 pages, 2 tables, 5 figures, 7 supplementary figures

The potential social harms that large language models pose, such as generating offensive content and reinforcing biases, are steeply rising. Existing works focus on coping with this concern while interacting with ill-intentioned users, such as those who explicitly make hate speech or elicit harmful responses. However, discussions on sensitive issues can become toxic even if the users are well-intentioned. For safer models in such scenarios, we present the Sensitive Questions and Acceptable Response (SQuARe) dataset, a large-scale Korean dataset of 49k sensitive questions with 42k acceptable and 46k non-acceptable responses. The dataset was constructed leveraging HyperCLOVA in a human-in-the-loop manner based on real news headlines. Experiments show that acceptable response generation significantly improves for HyperCLOVA and GPT-3, demonstrating the efficacy of this dataset., Comment: 19 pages, 10 figures, ACL 2023

Quality control in the manufacturing industry has improved with the use of artificial intelligence (AI). However, the manual inspection of trimming die designs, which is time-consuming and prone to errors, is still done by engineers. This study introduces an automatic design inspection system for automobile trimming dies by integrating AI modules and computer-aided design (CAD) software. The AI modules replace engineers' judgment, and the CAD software carries out operations requested by the AI modules. The inspection process involves a zigzag interaction between the AI modules and CAD software, enabling one-click operation without expert intervention. The AI modules are CAD-independent and data-efficient, making them adaptable to other CAD software. They achieve high performance even with limited training data, with an average length measurement error of only 2.4%. The inspection time is reduced to approximately one-fifth of the time required for manual inspection by experts.

This paper proposes a modeling structure for the relativistic constitutive equations of inelastic deformation in materials moving at high speeds. While the theory of relativity has successfully approximated material motion in space-time, most existing models only consider elastic behavior, neglecting inelastic deformation. A comprehensive relativistic inelasticity model is necessary to objectively measure inelastic deformation for observers with varying velocities, including those near the speed of light. The proposed model expands classical deformation tensors to four-dimensional tensors in space-time, referred to as relativistic Cauchy-Green deformation tensors. By introducing constitutive equations based on these tensors, objective measurements of inelastic deformation are achieved while satisfying the second law of thermodynamics. An illustrative example demonstrates the application of this theory and explains the physical interpretations of the model. Moreover, it is shown that the proposed relativistic inelasticity model converges to a classical inelasticity model under slower motion compared to the speed of light.