Post-training quantization reduces the computational demand of Large Language Models (LLMs) but can weaken some of their capabilities. Since LLM abilities emerge with scale, smaller LLMs are more sensitive to quantization. In this paper, we explore how quantization affects smaller LLMs’ ability to perform retrieval-augmented generation (RAG), specifically in longer contexts. We chose personalization for evaluation because it is a challenging domain to perform using RAG as it requires long-context reasoning over multiple documents. We compare the original FP16 and the quantized INT4 performance of multiple 7B and 8B LLMs on two tasks while progressively increasing the number of retrieved documents to test how quantized models fare against longer contexts. To better understand the effect of retrieval, we evaluate three retrieval models in our experiments. Our findings reveal that if a 7B LLM performs the task well, quantization does not impair its performance and long-context reasoning capabilities. We conclude that it is possible to utilize RAG with quantized smaller LLMs.
MULTIFILE
Introduction: Visuospatial neglect (VSN) is common after stroke and can seriously hamper everyday life. One of the most commonly used and highly recommended rehabilitation methods is Visual Scanning Training (VST) which requires a lot of repetition which makes the treatment intensive and less appealing for the patient. The use of eHealth in healthcare can increase options regarding improved treatment in the areas of patient satisfaction, treatment efficacy and effectiveness. One solution to motivational issues might be Augmented Reality (AR), which offers new opportunities for increasing natural interactions with the environment during treatment of VSN. Aim: The development of an AR-based scanning training program that will improve visuospatial search strategies in individuals affected by VSN. Method: We used a Design Research approach, which is characterized by the iterative and incremental use of prototypes as research instruments together with a strong human-centered focus. Several design thinking methods were used to explore which design elements the AR game should comply with. Seven patients with visuospatial neglect, eight occupational therapists, a game design professional and seven other healthcare professionals participated in this research by means of co-creation based on their own perspectives. Results: Fundamental design choices for an AR game for VSN patients included the factors extrinsic motivation, nostalgia, metaphors, direct feedback, independent movement, object contrast, search elements and competition. Designing for extrinsic motivation was considered the most important design choice, because due to less self-awareness the target group often does not fully understand and accept the consequences of VSN. Conclusion: This study produced a prototype AR game for people with VSN after stroke. The AR game and method used illustrate the promising role of AR tools in geriatric rehabilitation, specifically those aimed at increasing the independence of patients with VSN after stroke. 2020 The Authors. Publishing services by Elsevier B.V. on behalf of KeAi Communications Co. Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
DOCUMENT
Dealing with and maintaining high-quality standards in the design and construction phases is challenging, especially for on-site construction. Issues like improper implementation of building components and poor communication can widen the gap between design specifications and actual conditions. To prevent this, particularly for energy-efficient buildings, it is vital to develop resilient, sustainable strategies. These should optimize resource use, minimize environmental impact, and enhance livability, contributing to carbon neutrality by 2050 and climate change mitigation. Traditional post-occupancy evaluations, which identify defects after construction, are impractical for addressing energy performance gaps. A new, real-time inspection approach is necessary throughout the construction process. This paper suggests an innovative guideline for prefabricated buildings, emphasizing digital ‘self-instruction’ and ‘self-inspection’. These procedures ensure activities impacting quality adhere to specific instructions, drawings, and 3D models, incorporating the relevant acceptance criteria to verify completion. This methodology, promoting alignment with planned energy-efficient features, is supported by BIM-based software and Augmented Reality (AR) tools, embodying Industry 4.0 principles. BIM (Building Information Modeling) and AR bridge the gap between virtual design and actual construction, improving stakeholder communication and enabling real-time monitoring and adjustments. This integration fosters accuracy and efficiency, which are key for energy-efficient and nearly zero-energy buildings, marking a shift towards a more precise, collaborative, and environmentally sensible construction industry.
DOCUMENT
The huge number of images shared on the Web makes effective cataloguing methods for efficient storage and retrieval procedures specifically tailored on the end-user needs a very demanding and crucial issue. In this paper, we investigate the applicability of Automatic Image Annotation (AIA) for image tagging with a focus on the needs of database expansion for a news broadcasting company. First, we determine the feasibility of using AIA in such a context with the aim of minimizing an extensive retraining whenever a new tag needs to be incorporated in the tag set population. Then, an image annotation tool integrating a Convolutional Neural Network model (AlexNet) for feature extraction and a K-Nearest-Neighbours classifier for tag assignment to images is introduced and tested. The obtained performances are very promising addressing the proposed approach as valuable to tackle the problem of image tagging in the framework of a broadcasting company, whilst not yet optimal for integration in the business process.
DOCUMENT
To study the ways in which compounds can induce adverse effects, toxicologists have been constructing Adverse Outcome Pathways (AOPs). An AOP can be considered as a pragmatic tool to capture and visualize mechanisms underlying different types of toxicity inflicted by any kind of stressor, and describes the interactions between key entities that lead to the adverse outcome on multiple biological levels of organization. The construction or optimization of an AOP is a labor intensive process, which currently depends on the manual search, collection, reviewing and synthesis of available scientific literature. This process could however be largely facilitated using Natural Language Processing (NLP) to extract information contained in scientific literature in a systematic, objective, and rapid manner that would lead to greater accuracy and reproducibility. This would support researchers to invest their expertise in the substantive assessment of the AOPs by replacing the time spent on evidence gathering by a critical review of the data extracted by NLP. As case examples, we selected two frequent adversities observed in the liver: namely, cholestasis and steatosis denoting accumulation of bile and lipid, respectively. We used deep learning language models to recognize entities of interest in text and establish causal relationships between them. We demonstrate how an NLP pipeline combining Named Entity Recognition and a simple rules-based relationship extraction model helps screen compounds related to liver adversities in the literature, but also extract mechanistic information for how such adversities develop, from the molecular to the organismal level. Finally, we provide some perspectives opened by the recent progress in Large Language Models and how these could be used in the future. We propose this work brings two main contributions: 1) a proof-of-concept that NLP can support the extraction of information from text for modern toxicology and 2) a template open-source model for recognition of toxicological entities and extraction of their relationships. All resources are openly accessible via GitHub (https://github.com/ontox-project/en-tox).
DOCUMENT
In recent years, a step change has been seen in the rate of adoption of Industry 4.0 technologies by manufacturers and industrial organizations alike. This article discusses the current state of the art in the adoption of Industry 4.0 technologies within the construction industry. Increasing complexity in onsite construction projects coupled with the need for higher productivity is leading to increased interest in the potential use of Industry 4.0 technologies. This article discusses the relevance of the following key Industry 4.0 technologies to construction: data analytics and artificial intelligence, robotics and automation, building information management, sensors and wearables, digital twin, and industrial connectivity. Industrial connectivity is a key aspect as it ensures that all Industry 4.0 technologies are interconnected allowing the full benefits to be realized. This article also presents a research agenda for the adoption of Industry 4.0 technologies within the construction sector, a three-phase use of intelligent assets from the point of manufacture up to after build, and a four-staged R&D process for the implementation of smart wearables in a digital enhanced construction site.
DOCUMENT
With artificial intelligence (AI) systems entering our working and leisure environments with increasing adaptation and learning capabilities, new opportunities arise for developing hybrid (human-AI) intelligence (HI) systems, comprising new ways of collaboration. However, there is not yet a structured way of specifying design solutions of collaboration for hybrid intelligence (HI) systems and there is a lack of best practices shared across application domains. We address this gap by investigating the generalization of specific design solutions into design patterns that can be shared and applied in different contexts. We present a human-centered bottom-up approach for the specification of design solutions and their abstraction into team design patterns. We apply the proposed approach for 4 concrete HI use cases and show the successful extraction of team design patterns that are generalizable, providing re-usable design components across various domains. This work advances previous research on team design patterns and designing applications of HI systems.
MULTIFILE
The field of data science and artificial intelligence (AI) is growing at an unprecedented rate. Manual tasks that for thousands of years could only be performed by humans are increasingly being taken over by intelligent machines. But, more importantly, tasks that could never be performed manually by humans, such as analysing big data, can now be automated while generating valuable knowledge for humankind
DOCUMENT
ABSTRACT Purpose: This short paper describes the dashboard design process for online hate speech monitoring for multiple languages and platforms. Methodology/approach: A case study approach was adopted in which the authors followed a research & development project for a multilingual and multiplatform online dashboard monitoring online hate speech. The case under study is the project for the European Observatory of Online Hate (EOOH). Results: We outline the process taken for design and prototype development for which a design thinking approach was followed, including multiple potential user groups of the dashboard. The paper presents this process's outcome and the dashboard's initial use. The identified issues, such as obfuscation of the context or identity of user accounts of social media posts limiting the dashboard's usability while providing a trade-off in privacy protection, may contribute to the discourse on privacy and data protection in (big data) social media analysis for practitioners. Research limitations/implications: The results are from a single case study. Still, they may be relevant for other online hate speech detection and monitoring projects involving big data analysis and human annotation. Practical implications: The study emphasises the need to involve diverse user groups and a multidisciplinary team in developing a dashboard for online hate speech. The context in which potential online hate is disseminated and the network of accounts distributing or interacting with that hate speech seems relevant for analysis by a part of the user groups of the dashboard. International Information Management Association
LINK
The aim of this research is to explore the potential of Mixed Reality (MR) technologies for Operator Support in order to progress towards Industry 4.0 (I4.0) particularly for SMEs. Through a series of interventions and interviews conducted with local SMEs, potential use cases and their drawbacks have been identified. From this, insights were derived that serve as a starting point for conducting further experiments with MR technology in the smart manufacturing laboratory at the THUAS in Delft. The intervention consisted of a free form workshop in which the participants get ‘tinkering’ time to explore MR in their own work environment. The various levels of awareness were assessed in three stages: during an introductory interview, and after an instruction meeting and some ‘tinkering’. The study took place in the period from January 2022 to July 2022 with 10 local SMEs in the Netherlands. The results show that for all SMEs the awareness and understanding increased. The use cases identified by operators themselves concerned Quality Control, Diagnostics, Instruction, Specification and Improvement of Operations. Drawbacks foreseen related to Ergonomic Concerns, Resistance from operators, Technical considerations, Unavailability of MR device and an insufficient digital infrastructure to support MR in full extent. The use case most promising to the participants was further developed into a physical prototype for an ‘assisted assembly cell’ by which the aspects of ergonomics and the mentioned technical considerations could be analysed.
MULTIFILE
