Search

The widespread of Large Language Models (LLMs) marks a significant milestone in generative AI. Nevertheless, the increasing context length and batch size in offline LLM inference escalate the memory requirement of the key-value (KV) cache, which imposes a huge burden on the GPU VRAM, especially for resource-constraint scenarios (e.g., edge computing and personal devices). Several cost-effective solutions leverage host memory or SSDs to reduce storage costs for offline inference scenarios and improve the throughput. Nevertheless, they suffer from significant performance penalties imposed by intensive KV cache accesses due to limited PCIe bandwidth. To address these issues, we propose InstInfer, a novel LLM inference system that offloads the most performance-critical computation (i.e., attention in decoding phase) and data (i.e., KV cache) parts to Computational Storage Drives (CSDs), which minimize the enormous KV transfer overheads. InstInfer designs a dedicated flash-aware in-storage attention engine with KV cache management mechanisms to exploit the high internal bandwidths of CSDs instead of being limited by the PCIe bandwidth. The optimized P2P transmission between GPU and CSDs further reduces data migration overheads. Experimental results demonstrate that for a 13B model using an NVIDIA A6000 GPU, InstInfer improves throughput for long-sequence inference by up to 11.1$\times$, compared to existing SSD-based solutions such as FlexGen.

Foundational vision models, such as the Segment Anything Model (SAM), have achieved significant breakthroughs through extensive pre-training on large-scale visual datasets. Despite their general success, these models may fall short in specialized tasks with limited data, and fine-tuning such large-scale models is often not feasible. Current strategies involve incorporating adaptors into the pre-trained SAM to facilitate downstream task performance with minimal model adjustment. However, these strategies can be hampered by suboptimal learning approaches for the adaptors. In this paper, we introduce a novel Multi-scale Contrastive Adaptor learning method named MCA-SAM, which enhances adaptor performance through a meticulously designed contrastive learning framework at both token and sample levels. Our Token-level Contrastive adaptor (TC-adaptor) focuses on refining local representations by improving the discriminability of patch tokens, while the Sample-level Contrastive adaptor (SC-adaptor) amplifies global understanding across different samples. Together, these adaptors synergistically enhance feature comparison within and across samples, bolstering the model's representational strength and its ability to adapt to new tasks. Empirical results demonstrate that MCA-SAM sets new benchmarks, outperforming existing methods in three challenging domains: camouflage object detection, shadow segmentation, and polyp segmentation. Specifically, MCA-SAM exhibits substantial relative performance enhancements, achieving a 20.0% improvement in MAE on the COD10K dataset, a 6.0% improvement in MAE on the CAMO dataset, a 15.4% improvement in BER on the ISTD dataset, and a 7.9% improvement in mDice on the Kvasir-SEG dataset.

Implementing cross-modal hashing between 2D images and 3D point-cloud data is a growing concern in real-world retrieval systems. Simply applying existing cross-modal approaches to this new task fails to adequately capture latent multi-modal semantics and effectively bridge the modality gap between 2D and 3D. To address these issues without relying on hand-crafted labels, we propose contrastive masked autoencoders based self-supervised hashing (CMAH) for retrieval between images and point-cloud data. We start by contrasting 2D-3D pairs and explicitly constraining them into a joint Hamming space. This contrastive learning process ensures robust discriminability for the generated hash codes and effectively reduces the modality gap. Moreover, we utilize multi-modal auto-encoders to enhance the model's understanding of multi-modal semantics. By completing the masked image/point-cloud data modeling task, the model is encouraged to capture more localized clues. In addition, the proposed multi-modal fusion block facilitates fine-grained interactions among different modalities. Extensive experiments on three public datasets demonstrate that the proposed CMAH significantly outperforms all baseline methods., Comment: Accepted by ICME 2024

Urban parks provide significant health benefits by offering spaces and facilities for various recreational and leisure activities. However, the capacity of specific park spaces and elements to foster health remains underexamined. Traditional studies have focused on parks' size, greenery, and accessibility, often overlooking their ability to facilitate specific health-promoting activities. To address this gap, we propose a taxonomy consisting of six categories of health-promoting activities in parks: physical, mind-body, nature appreciation, environmental, social, and cultural. We estimate the capacity of parks in 35 global cities to promote health by establishing a lexicon linking park spaces and elements with specific health-promoting activities from our taxonomy. Using this lexicon, we collected data on elements and spaces in all parks in 35 cities from OpenStreetMap. Our analysis covers 23,477 parks with a total of 827,038 elements and spaces. By first comparing similarly sized parks across cities, we found that North American parks offer more spaces for physical activities, while European parks focus more on nature appreciation. Second, by scoring parks based on both elements and spaces, we investigated the variability in their health-promoting potential. We found the most uniform provision across parks for physical activities and the highest disparities regarding social activities. Additionally, parks offering a variety of activities are usually located in city centers, while offerings diminish in parks towards the suburbs. Lastly, we identified significant inequalities in park standards across cities, regardless of their continental location: Tokyo and Paris offer the most uniform park standards, while Copenhagen and Rio de Janeiro exhibit the most pronounced disparities. Our study provides insights for making urban parks more equitable, engaging, and health-promoting., Comment: 29 pages main paper, 10 pages appendix

High-temperature cuprates superconductors are characterised by the complex interplay between superconductivity (SC) and charge density wave (CDW) in the context of intertwined competing orders. In contrast to abundant studies for hole-doped cuprates, the exact nature of CDW and its relationship to SC was much less explored in electron-doped counterparts. Here, we performed resonant inelastic x-ray scattering (RIXS) experiments to investigate the relationship between CDW and SC in electron-doped La$_{2-x}$Ce$_x$CuO$_4$. The short-range CDW order with a correlation length $\sim35$~\AA~was found in a wide range of temperature and doping concentration. Near the optimal doping, the CDW order is weakened inside the SC phase, implying an intimate relationship between the two orders. This interplay has been commonly reported in hole-doped La-based cuprates near the optimal doping. We reconciled the diverging behaviour of CDW across the superconducting phase in various cuprate materials by introducing the CDW correlation length as a key parameter. Our study paves the way for establishing a unified picture to describe the phenomenology of CDW and its relationship with SC in the cuprate family., Comment: 24 pages 5 figures; Supplementary Materials available upon request

Estimating many-body effects that deviate from an independent particle approach, has long been a key research interest in condensed matter physics. Layered cuprates are prototypical systems, where electron-electron interactions are found to strongly affect the dynamics of single-particle excitations. It is however, still unclear how the electron correlations influence charge excitations, such as plasmons, which have been variously treated with either weak or strong correlation models. In this work, we demonstrate the hybridised nature of collective valence charge fluctuations leading to dispersing acoustic-like plasmons in hole-doped La$_{1.84}$Sr$_{0.16}$CuO$_{4}$ and electron-doped La$_{1.84}$Ce$_{0.16}$CuO$_{4}$ using the two-particle probe, resonant inelastic x-ray scattering. We then describe the plasmon dispersions in both systems, within both the weak mean-field Random Phase Approximation (RPA) and strong coupling $t$-$J$-$V$ models. The $t$-$J$-$V$ model, which includes the correlation effects implicitly, accurately describes the plasmon dispersions as resonant excitations outside the single-particle intra-band continuum. In comparison, a quantitative description of the plasmon dispersion in the RPA approach is obtained only upon explicit consideration of re-normalized electronic band parameters. Our comparative analysis shows that electron correlations significantly impact the low-energy plasmon excitations across the cuprate doping phase diagram, even at long wavelengths. Thus, complementary information on the evolution of electron correlations, influenced by the rich electronic phases in condensed matter systems, can be extracted through the study of two-particle charge response., Comment: 6 Figures

AI regulations are expected to prohibit machine learning models from using sensitive attributes during training. However, the latest Natural Language Processing (NLP) classifiers, which rely on deep learning, operate as black-box systems, complicating the detection and remediation of such misuse. Traditional bias mitigation methods in NLP aim for comparable performance across different groups based on attributes like gender or race but fail to address the underlying issue of reliance on protected attributes. To partly fix that, we introduce NLPGuard, a framework for mitigating the reliance on protected attributes in NLP classifiers. NLPGuard takes an unlabeled dataset, an existing NLP classifier, and its training data as input, producing a modified training dataset that significantly reduces dependence on protected attributes without compromising accuracy. NLPGuard is applied to three classification tasks: identifying toxic language, sentiment analysis, and occupation classification. Our evaluation shows that current NLP classifiers heavily depend on protected attributes, with up to $23\%$ of the most predictive words associated with these attributes. However, NLPGuard effectively reduces this reliance by up to $79\%$, while slightly improving accuracy., Comment: Paper accepted at CSCW 2024

The relation between charge order, its quantum fluctuations and optical phonon modes in cuprate superconductors remains an unsolved problem. The exploration of these excitations is however complicated by the presence of twinned domains. Here, we use uniaxial strain in combination with ultra-high-resolution Resonant Inelastic X-ray Scattering (RIXS) at the oxygen K- and copper L3-edges to study the excitations stemming from the charge ordering wave vector in La1.875Sr0.125CuO4. By detwinning stripe ordering, we demonstrate that the optical phonon anomalies do not show any stripe anisotropy. The low-energy charge excitations also retain an in-plane four-fold symmetry. As such, we find that both phonon and charge excitations are decoupled entirely from the strength of static charge ordering. The almost isotropic character of charge excitations remains a possible source for the strange metal properties found in the normal state of cuprate superconductors., Comment: 9 pages, 4 figures

The recently released Segment Anything Model (SAM) has shown powerful zero-shot segmentation capabilities through a semi-automatic annotation setup in which the user can provide a prompt in the form of clicks or bounding boxes. There is growing interest around applying this to medical imaging, where the cost of obtaining expert annotations is high, privacy restrictions may limit sharing of patient data, and model generalisation is often poor. However, there are large amounts of inherent uncertainty in medical images, due to unclear object boundaries, low-contrast media, and differences in expert labelling style. Currently, SAM is known to struggle in a zero-shot setting to adequately annotate the contours of the structure of interest in medical images, where the uncertainty is often greatest, thus requiring significant manual correction. To mitigate this, we introduce \textbf{Sim}ulated Interaction for \textbf{S}egment \textbf{A}nything \textbf{M}odel (\textsc{\textbf{SimSAM}}), an approach that leverages simulated user interaction to generate an arbitrary number of candidate masks, and uses a novel aggregation approach to output the most compatible mask. Crucially, our method can be used during inference directly on top of SAM, without any additional training requirement. Quantitatively, we evaluate our method across three publicly available medical imaging datasets, and find that our approach leads to up to a 15.5\% improvement in contour segmentation accuracy compared to zero-shot SAM. Our code is available at \url{https://github.com/BenjaminTowle/SimSAM}., Comment: Published at ISBI 2024. Awarded Top 12 Oral Presentation

Unlike Li-ion transport in the bulk of carbonaceous materials, little is known about Li-ion diffusion on their surface. In this study, we have discovered an ultra-fast Li-ion transport phenomenon on the surface of carbonaceous materials, particularly when they have limited Li insertion capacity along with a high surface area. This is exemplified by a carbon black, Ketjen Black (KB). An ionic conductivity of 18.1 mS cm-1 at room temperature is observed, far exceeding most solid-state ion conductors. Theoretical calculations reveal a low diffusion barrier for the surface Li species. The species is also identified as Li*, which features a partial positive charge. As a result, lithiated KB functions effectively as an interlayer between Li and solid-state electrolytes (SSE) to mitigate dendrite growth and cell shorting. This function is found to be electrolyte agnostic, effective for both sulfide and halide SSEs. Further, lithiated KB can act as a high-performance mixed ion/electron conductor that is thermodynamically stable at potentials near Li metal. A graphite anode mixed with KB instead of a solid electrolyte demonstrates full utilization with a capacity retention of ~85% over 300 cycles. The discovery of this surface-mediated ultra-fast Li-ion transport mechanism provides new directions for the design of solid-state ion conductors and solid-state batteries., Comment: 21 pages, 6 figures

Heterostructures from complex oxides allow one to combine various electronic and magnetic orders as to induce new quantum states. A prominent example is the coupling between superconducting and magnetic orders in multilayers from high-Tc cuprates and manganites. A key role is played here by the interfacial CuO2 layer whose distinct properties remain to be fully understood. Here, we study with resonant inelastic X-ray scattering (RIXS) the magnon excitations of this interfacial CuO2 layer. In particular, we show that the underlying antiferromagnetic exchange interaction at the interface is strongly suppressed to J ~ 70 meV, as compared to J ~ 130 meV for the CuO2 layers away from the interface. Moreover, we observe an anomalous momentum dependence of the intensity of the interfacial magnon mode and show that it suggests that the antiferromagnetic order is accompanied by a particular kind of orbital order that yields a so-called altermagnetic state. Such a two-dimensional altermagnet has recently been predicted to enable new spintronic applications and superconducting proximity effects.

Strongly-correlated transition-metal oxides are widely known for their various exotic phenomena. This is exemplified by rare-earth nickelates such as LaNiO$_{3}$, which possess intimate interconnections between their electronic, spin, and lattice degrees of freedom. Their properties can be further enhanced by pairing them in hybrid heterostructures, which can lead to hidden phases and emergent phenomena. An important example is the LaNiO$_{3}$/LaTiO$_{3}$ superlattice, where an interlayer electron transfer has been observed from LaTiO$_{3}$ into LaNiO$_{3}$ and is predicted to result in a high-spin state. However, macroscopic emergence of magnetic order has so far not been observed. Here, by using muon spin rotation, x-ray absorption, and resonant inelastic x-ray scattering, we present direct evidence of an emergent antiferromagnetic order with high magnon energy and exchange interactions at the LaNiO$_{3}$/LaTiO$_{3}$ interface. As the magnetism is purely interfacial, a single LaNiO$_{3}$/LaTiO$_{3}$ interface can essentially behave as an atomically thin quasi-two-dimensional antiferromagnet, potentially allowing its technological utilisation in advanced spintronic devices. Furthermore, its strong quasi-two-dimensional magnetic correlations and orbitally-polarized planar ligand holes make its electronic and magnetic configurations resemble the precursor states of superconducting cuprates and nickelates, but with an S $\rightarrow$ 1 spin state instead., Comment: 41 pages, 13 figures

Recent x-ray spectroscopic studies have revealed spin excitations and charge density waves in thin films of infinite-layer (IL) nickelates. However, clarifying whether the origin of these phenomena is intrinsic to the material class or attributable to impurity phases in the films has presented a major challenge. Here we utilize topotactic methods to synthesize bulk crystals of the IL nickelate LaNiO$_2$ with crystallographically oriented surfaces. We examine these crystals using resonant inelastic x-ray scattering (RIXS) at the Ni $L_3$-edge to elucidate the spin and charge correlations in the bulk of the material. While we detect the presence of prominent spin excitations in the crystals, fingerprints of charge order are absent at the ordering vectors identified in previous in thin-film studies. These results contribute to the understanding of the bulk properties of LaNiO$_2$ and establish topotactically synthesized crystals as viable complementary specimens for spectroscopic investigations., Comment: 11 pages, 7 figures with supplemental materials

As Artificial Intelligence (AI) becomes ubiquitous, the need for Explainable AI (XAI) has become critical for transparency and trust among users. A significant challenge in XAI is catering to diverse users, such as data scientists, domain experts, and end-users. Recent research has started to investigate how users' characteristics impact interactions with and user experience of explanations, with a view to personalizing XAI. However, are we heading down a rabbit hole by focusing on unimportant details? Our research aimed to investigate how user characteristics are related to using, understanding, and trusting an AI system that provides explanations. Our empirical study with 149 participants who interacted with an XAI system that flagged inappropriate comments showed that very few user characteristics mattered; only age and the personality trait openness influenced actual understanding. Our work provides evidence to reorient user-focused XAI research and question the pursuit of personalized XAI based on fine-grained user characteristics., Comment: 20 pages, 4 tables, 2 figures

In this paper, we define the upper (resp. lower) covariance under multiple probabilities via a corresponding max-min-max (resp. min-max-min) optimization problem and the related properties of covariances are obtained. In particular, we propose a fast algorithm of calculation for upper and lower covariances under the finite number of probabilities. As an application, our algorithm can be used to solve a class of quadratic programming problem exactly, and we obtain a probabilistic representation of such quadratic programming problem., Comment: 20 pages

The service provided by content delivery networks (CDNs) may overlook content locality, leaving room for potential performance improvement. In this study, we explore the feasibility of leveraging generated data as a replacement for fetching data in missing scenarios based on content locality. Due to sufficient local computing resources and reliable generation efficiency, we propose a content-oriented generative-hit framework (CoGenT) for CDNs. CoGenT utilizes idle computing resources on edge nodes to generate requested data based on similar or related cached data to achieve hits. Our implementation in a real-world system demonstrates that CoGenT reduces the average access latency by half. Additionally, experiments conducted on a simulator also confirm that CoGenT can enhance existing caching algorithms, resulting in reduced latency and bandwidth usage.

The striking discovery of high-temperature superconductivity (HTSC) of 80 K in a bilayer nickelate La$_3$Ni$_2$O$_7$ under a moderately high pressure of about 14 GPa ignited a new wave of studying HTSC in nickelates. The properties of the parental phase at ambient pressure may contain key information on basic interactions therein and bosons that may mediate pairing giving birth to superconductivity. Moreover, the bilayer structure of La$_3$Ni$_2$O$_7$ may suggest a distinct minimal model in comparison to cuprate superconductors. Here using X-ray absorption spectroscopy and resonant inelastic X-ray scattering, we studied La$_3$Ni$_2$O$_7$ at ambient pressure, and found that Ni 3$d_{x^2-y^2}$, Ni 3$d_{z^2}$, and ligand oxygen 2$p$ orbitals dominate the low-energy physics with a small charge-transfer energy. Remarkably, well-defined optical-like magnetic excitations were found to soften into a quasi-static spin-density-wave ordering, evidencing the strong electronic correlations and rich magnetic properties. Based on a Heisenberg spin model, we found that the inter-layer effective magnetic superexchange interaction is much larger than the intra-layer ones, and proposed two viable magnetic structures. Our results set the foundation for further exploration of La$_3$Ni$_2$O$_7$ superconductor.

Misinformation proliferates in the online sphere, with evident impacts on the political and social realms, influencing democratic discourse and posing risks to public health and safety. The corporate world is also a prime target for fake news dissemination. While recent studies have attempted to characterize corporate misinformation and its effects on companies, their findings often suffer from limitations due to qualitative or narrative approaches and a narrow focus on specific industries. To address this gap, we conducted an analysis utilizing social media quantitative methods and crowd-sourcing studies to investigate corporate misinformation across a diverse array of industries within the S\&P 500 companies. Our study reveals that corporate misinformation encompasses topics such as products, politics, and societal issues. We discovered companies affected by fake news also get reputable news coverage but less social media attention, leading to heightened negativity in social media comments, diminished stock growth, and increased stress mentions among employee reviews. Additionally, we observe that a company is not targeted by fake news all the time, but there are particular times when a critical mass of fake news emerges. These findings hold significant implications for regulators, business leaders, and investors, emphasizing the necessity to vigilantly monitor the escalating phenomenon of corporate misinformation., Comment: Accepted in ICWSM 2024