151398 research outputs found
Sort by
Ionic liquid gels: catalysts for sustainability in synthesis, energy, electronics and medicine
No full textThis opinion paper reflects upon highlights in ionic liquid gel research for green and sustainable chemistry. Discussed are some key observations from research into ionic liquid gels in catalysis, and insights into potential future uses and areas for development. It is noted that some IL gels have a remarkable ability to hold onto precious metals when used to entrap catalysts (nanoparticulate metals and complexes with minimal leaching) allowing heterogenized homogeneous catalysts to be reused, recycled and easily separated. Herein we present observations that ionic liquid gels can actively absorb and retain metals from solution. Evolving trends in research into liquids comprised of ions will be reviewed and a future in which the ionic liquid gels themselves are green and sustainable materials is hypothesized, which will accelerate their adoption as NaturIL gels in synthesis, materials, electronics and medicine
Attention eclipse: manipulating attention to bypass LLM safety-alignment
No full textRecent research has shown that carefully crafted jailbreak inputs can induce large language models to produce harmful outputs, despite safety measures such as alignment. It is important to anticipate the range of potential Jailbreak attacks to guide effective defenses and accurate assessment of model safety. In this paper, we present a new approach for generating highly effective Jailbreak attacks that manipulate the attention of the model to selectively strengthen or weaken attention among different parts of the prompt. By harnessing attention loss, we develop more effective jailbreak attacks, that are also transferrable. The attacks amplify the success rate of existing Jailbreak algorithms including GCG, AutoDAN, and ReNeLLM, while lowering their generation cost (for example, the amplified GCG attack achieves 91.2% ASR, vs. 67.9% for the original attack on Llama2-7B/AdvBench, using less than a third of the generation time)
MIX: a multi-view time-frequency interactive explanation framework for time series classification
No full textDeep learning models for time series classification (TSC) have achieved impressive performance, but explaining their decisions remains a significant challenge. Existing post-hoc explanation methods typically operate solely in the time domain and from a single-view perspective, limiting both faithfulness and robustness. In this work, we propose MIX (Multi-view Time-Frequency Interactive EXplanation Framework), a novel framework that helps to explain deep learning models in a multi-view setting by leveraging multi-resolution, time-frequency views constructed using the Haar Discrete Wavelet Transform (DWT). MIX introduces an interactive cross-view refinement scheme, where explanation's information from one view is propagated across views to enhance overall interpretability. To align with user-preferred perspectives, we propose a greedy selection strategy that traverses the multi-view space to identify the most informative features. Additionally, we present OSIGV, a user-aligned segment-level attribution mechanism based on overlapping windows for each view, and introduce keystone-first IG, a method that refines explanations in each view using additional information from another view. Extensive experiments across multiple TSC benchmarks and model architectures demonstrate that MIX significantly outperforms state-of-the-art (SOTA) methods in terms of explanation faithfulness and robustness
RejSCore: rejection sampling core for multivariate-based public key cryptography
No full textPost-quantum multivariate public key cryptography (MPKC) schemes resist quantum threats but require heavy operations, such as rejection sampling, which challenge resource-limited devices. Prior hardware designs have addressed various aspects of MPKC signature generation. However, rejection sampling remains largely unexplored. This paper presents RejSCore, a lightweight hardware accelerator for rejection sampling in post-quantum cryptography. It targets the QR-UOV scheme, which is a prominent candidate under the second-round of the National Institute of Standards and Technology (NIST) additional digital signature standardization process. The architecture includes an AES-CTR-128-based pseudorandom number generator. Moreover, a lightweight iterative method is employed in rejection sampling, offering reduced resource consumption and area overhead while slightly increasing latency. The performance of RejSCore is comprehensively evaluated on Artix-7 FPGAs and 65 nm CMOS technology using the Area-Delay Product (ADP) and Power-Delay Product (PDP). On Artix-7 and 65 nm CMOS, RejSCore achieves an area of 2042slices and 464,866 μm2, with operating frequencies of 222 MHz and 565 MHz, respectively. Using the QR-UOV parameters for security level I (q = 127, v = 156, m = 54, l = 3), the core completes its operation in 8525 clock cycles. The ADP and PDPevaluations confirm RejSCore’s suitability for deployment in resource-constrained and security-critical environments
Automated data extraction for CAD based weld time estimation
No full textDespite the growing use of digital tools for production planning, the use of CAD models for the purpose of automated data provision to determine production times and resource requirements remains unrealised. Current approaches rely on the experience of production operatives or time-and-motion studies, both of which are resource-intensive, error-prone, and disconnected from design data. This breaks the continuity of the digital thread between CAD and manufacturing, complicating integration with new digital manufacturing technologies and approaches. Drawing on literature and workshops with our industrial partner, this paper presents an approach for extracting manufacturing-relevant features directly from CAD models, enabling early-stage estimation of process times. These estimations can be used to support production simulation and planning activities, helping to bridge the gap between design intent and shopfloor realities. The results show that automated methods for weld length recognition are not only effective, but are also computationally efficient and significantly faster than existing manual methods
Exploring AACA members' perceptions of integrating educational technologies in anatomy education
No full textINTRODUCTION. The adoption of technological tools in anatomy education has grown significantly, particularly in response to the shift toward online modalities during the pandemic. While prior studies often focus on specific tools, this work evaluates the broader integration of diverse educational technologies, based on feedback from AACA members. The study aims to provide insights into the effectiveness, challenges, and potential of these tools, offering practical guidelines for educators. METHODS. An online survey utilized Qualtrics software was distributed to AACA members. It explored faculty perceptions across four domains: (1) Current Educational Technology;(2) Impact on Teaching Anatomy; (3) Technical Challenges; and (4) Comparison with Traditional Dissection. Data were collected and analyzed to identify key trends. SUMMARY. Among 63 respondents (~20% of AACA membership), digital textbooks and e-resources (e.g., Access Anatomy) were the most utilized tools, followed by online platforms. The majority (70%) agreed that technology enhances the explanation of complex anatomical concepts, with 74% reporting increased student engagement and 72% citing improved accommodation of diverse learning styles. However, 66% identified cost as a significant barrier, and 45% reported challenges with technical support. Notably, 81% preferred traditional cadaver dissection over technology-based methods, with 83% endorsing its superior educational value in understanding anatomical variability.CONCLUSIONS. This study reveals strong support for the role of technology in anatomy education while reaffirming the value of cadaver dissection. The findings underscore the need for balanced integration, addressing cost and technical challenges, and prioritizing faculty development programs to optimize the effective use of technology. This study offers a comprehensive perspective on current practices and challenges, guiding the effective use of technology in anatomy curricula.<br/
A tale of carbon and economic unicorns: techno-optimism and dangerous mythic thinking
No full textA spectre is haunting our world and worldbuilding imaginaries in the twenty-first century (this age of the Anthropocene/Capitalocene/Plantationocene/ Trumpocene, the ‘Its too late o cene’) – the spectre of the collapse of the life supporting systems of a habitable earth. We live in a time of denial and erasure, ‘truth decay’ and an age of history, peoples and places being disappeared. Welcome to our fluxed futures. If batshit crazy ideas of geoengineering and managing the planet or bioengineering ourselves to be adapted to a 3 or 4 degree warmer world are being proposed by supposed ‘serious people in charge’ (including by colleagues in the academy), maybe we should have some of our own batshit crazy alternatives. This would seem to me to be our minimum proposition for those of us who advocate for structural and revolutionary change. The hard scientific/technocratic/malestream economic approach to address the planetary crisis can be summarised as saying that ‘we’ (a homogenised human collective subject straight of ‘Anthropocene’ central casting, no troubling issues of gender inequality, legacies of colonialisation, class, spatial or racial injustices or power disparities), this ‘we’ or rather an elite from this ‘we’ should via technological means ‘control’ the earth. Control and management are indeed important. But where we differ is that instead of ‘controlling nature’ (and in the process creating sacrificed zones and peoples), what humanity (in all its heterogeneity) should do is control our always provisional and revisable relations to and relationships with nature and each other. <br/
An X-band leaky-wave dynamic metasurface antenna for integrated sensing and communication
No full textThe work demonstrates a dynamic metasurface antenna (DMA) that consists of hexagonal-shaped meta-atoms.The hexagonal-shaped meta-atoms are arranged in a diagonal fashion in the top layer of the substrate integrated waveguide(SIW). The SIW feeds an array of 32 such meta-atoms. By using p-i-n diodes, the radiation characteristic and the interaction with the propagating wave of each meta-atom are varied in two states. The DMA is capable of generating 7 distinct beam patterns by biasing the p-i-n diodes with the binary distributions obtained through holographic approximation. The DMA offers a wide beam coverage of 136° and an average gain of 6.3 dBi for all the binary distributions. Furthermore, the DMA demonstrates direction-of-arrival (DoA) detection for one or two simultaneous sources with an error as low as 1°. Full-wave simulations validate the design, highlighting its potential for integrated sensing and communication (ISAC) application