Search

Large Language Models (LLMs) have significantly improved personalized conversational capabilities. However, existing datasets like Persona Chat, Synthetic Persona Chat, and Blended Skill Talk rely on static, predefined personas. This approach often results in dialogues that fail to capture human personalities' fluid and evolving nature. To overcome these limitations, we introduce a novel dataset with around 400,000 dialogues and a framework for generating personalized conversations using long-form journal entries from Reddit. Our approach clusters journal entries for each author and filters them by selecting the most representative cluster, ensuring that the retained entries best reflect the author's personality. We further refine the data by capturing the Big Five personality traits --openness, conscientiousness, extraversion, agreeableness, and neuroticism --ensuring that dialogues authentically reflect an individual's personality. Using Llama 3 70B, we generate high-quality, personality-rich dialogues grounded in these journal entries. Fine-tuning models on this dataset leads to an 11% improvement in capturing personality traits on average, outperforming existing approaches in generating more coherent and personality-driven dialogues., Comment: Accepted in COLING 2025

We study $H\to ZZ^\star$ production process in final four lepton states at $13$ TeV LHC in SMEFT framework. The anomalous $HZZ$ couplings are parameterized with dimension-6 $SU(2)_L\times U(1)_Y$ gauge invariant operators. We compute the eight polarizations of each $Z$ boson and $64$ spin-correlations as asymmetries in angular functions of final decayed leptons in the rest frame of the $Z$ boson. These asymmetries are further used to construct the joint density matrix (DM) for $ZZ^\star$ system. However, such DM suffers from negative probability and eigenvalues. To alleviate the negativity issues, we reconstruct the DM using asymmetries of symmetrized angular functions owing to the indistinguishability of two $Z$ bosons. The symmetrized DM is further employed to compute lower bound for concurrence as a witness of entanglement measurable at the collider experiments. The $ZZ^\star$ system is found to be in an entangled state for all values of the anomalous couplings. Notably, while the lower bound exhibits poorer sensitivity to anomalous couplings compared to asymmetries, it demonstrates distinct behavior for CP-even and odd couplings., Comment: 21 pages, 4 figures

In recent years, Cyber attacks have increased in number, and with them, the intensity of the attacks and their potential to damage the user have also increased significantly. In an ever-advancing world, users find it difficult to keep up with the latest developments in technology, which can leave them vulnerable to attacks. To avoid such situations we need tools to deter such attacks, for this machine learning models are among the best options. This paper presents a Browser Extension that uses machine learning models to enhance online security by integrating three crucial functionalities: Malicious URL detection, Spam Email detection and Network logs analysis. The proposed solution uses LGBM classifier for classification of Phishing websites, the model has been trained on a dataset with 87 features, this model achieved an accuracy of 96.5% with a precision of 96.8% and F1 score of 96.49%. The Model for Spam email detection uses Multinomial NB algorithm which has been trained on a dataset with over 5500 messages, this model achieved an accuracy of 97.09% with a precision of 100%. The results demonstrate the effectiveness of using machine learning models for cyber security., Comment: 5 Pages, 2 figures

Observations of the Jovian upper atmosphere at high latitudes in the UV, IR and mm/sub-mm all indicate that the chemical distributions and thermal structure are broadly influenced by auroral particle precipitations. Mid-IR and UV observations have shown that several light hydrocarbons (up to 6 carbon atoms) have altered abundances near Jupiter's main auroral ovals. Ion-neutral reactions influence the hydrocarbon chemistry, with light hydrocarbons produced in the upper stratosphere, and heavier hydrocarbons as well as aerosols produced in the lower stratosphere. One consequence of the magnetosphere-ionosphere coupling is the existence of ionospheric jets that propagate into the neutral middle stratosphere, likely acting as a dynamical barrier to the aurora-produced species. As the ionospheric jets and the background atmosphere do not co-rotate at the same rate, this creates a complex system where chemistry and dynamics are intertwined. The ion-neutral reactions produce species with a spatial distribution following the SIII longitude system in the upper stratosphere. As these species sediment down to the lower stratosphere, and because of the progressive dynamical decoupling between the ionospheric flows and the background atmosphere, the spatial distribution of the auroral-related species progressively follows a zonal distribution with increasing pressures that ultimately produces a system of polar and subpolar hazes that extends down to the bottom of the stratosphere. This paper reviews the most recent work addressing different aspects of this environment., Comment: Accepted for publication in Space Science Reviews. 65 pages, 20 figures (including appendix)

In recent years, there has been significant effort to align large language models with human preferences. This work focuses on developing a chatbot specialized in the real estate domain, with an emphasis on incorporating compliant behavior to ensure it can be used without perpetuating discriminatory practices like steering and redlining, which have historically plagued the real estate industry in the United States. Building on prior work, we present a method for generating a synthetic general instruction-following dataset, along with safety data. Through extensive evaluations and benchmarks, we fine-tuned a llama-3-8B-instruct model and demonstrated that we can enhance it's performance significantly to match huge closed-source models like GPT-4o while making it safer and more compliant. We open-source the model, data and code to support further development and research in the community.

Urban air pollution hotspots pose significant health risks, yet their detection and analysis remain limited by the sparsity of public sensor networks. This paper addresses this challenge by combining predictive modeling and mechanistic approaches to comprehensively monitor pollution hotspots. We enhanced New Delhi's existing sensor network with 28 low-cost sensors, collecting PM2.5 data over 30 months from May 1, 2018, to Nov 1, 2020. Applying established definitions of hotspots to this data, we found the existence of additional 189 hidden hotspots apart from confirming 660 hotspots detected by the public network. Using predictive techniques like Space-Time Kriging, we identified hidden hotspots with 95% precision and 88% recall with 50% sensor failure rate, and with 98% precision and 95% recall with 50% missing sensors. The projected results of our predictive models were further compiled into policy recommendations for public authorities. Additionally, we developed a Gaussian Plume Dispersion Model to understand the mechanistic underpinnings of hotspot formation, incorporating an emissions inventory derived from local sources. Our mechanistic model is able to explain 65% of observed transient hotspots. Our findings underscore the importance of integrating data-driven predictive models with physics-based mechanistic models for scalable and robust air pollution management in resource-constrained settings.

This paper studies the pricing of contingent claims of American style, using indifference pricing by fully dynamic convex risk measures. We provide a general definition of risk-indifference prices for buyers and sellers in continuous time, in a setting where buyer and seller have potentially different information, and show that these definitions are consistent with no-arbitrage principles. Specifying to stochastic volatility models, we characterize indifference prices via solutions of Backward Stochastic Differential Equations reflected at Backward Stochastic Differential Equations and show that this characterization provides a basis for the implementation of numerical methods using deep learning., Comment: 18 pages, 3 figures

We examine the parameter space of the Next to Minimal Supersymmetric Standard Model (NMSSM) with Singlino-dominated neutralino $\widetilde\chi_1^0$ as the lightest supersymmetric particle (LSP). Our study focuses on identifying the regions within this parameter space that produce a thermal relic abundance of $\widetilde\chi_1^0$ smaller than the observed cold dark matter relic density while remaining consistent with constraints from LEP measurements, low-energy experiments, Higgs measurements, LHC data, and dark matter direct detection experiments. We identify the dominant annihilation modes of the LSP neutralino across varying LSP mass ranges $\sim \mathcal{O}(1)-\mathcal{O}(10^{3})~$GeV. Furthermore, we conduct a benchmark study to assess the production rates of triple-boson final states emerging from direct electroweakino pair production at the LHC. Drawing insights from these findings, we perform a detailed collider analysis to explore the future potential of probing the triple-boson final states involving a light Higgs boson at the high-luminosity LHC (HL-LHC)., Comment: 31 pages, 17 figures, 4 tables

Efficient intravascular access in trauma and critical care significantly impacts patient outcomes. However, the availability of skilled medical personnel in austere environments is often limited. Autonomous robotic ultrasound systems can aid in needle insertion for medication delivery and support non-experts in such tasks. Despite advances in autonomous needle insertion, inaccuracies in vessel segmentation predictions pose risks. Understanding the uncertainty of predictive models in ultrasound imaging is crucial for assessing their reliability. We introduce MSU-Net, a novel multistage approach for training an ensemble of U-Nets to yield accurate ultrasound image segmentation maps. We demonstrate substantial improvements, 18.1% over a single Monte Carlo U-Net, enhancing uncertainty evaluations, model transparency, and trustworthiness. By highlighting areas of model certainty, MSU-Net can guide safe needle insertions, empowering non-experts to accomplish such tasks., Comment: Accepted for the 5th International Workshop of Advances in Simplifying Medical UltraSound (ASMUS), held in conjunction with MICCAI 2024, the 27th International Conference on Medical Image Computing and Computer Assisted Intervention

Widespread air source heat pump (ASHP) adoption faces several challenges that on-site thermal energy storage (TES), particularly thermochemical salt hydrate TES, can mitigate. No techno-economic analyses for salt-hydrate-based TES in residential applications exist. We quantify the residential space heating value of four salt hydrate TES materials - MgSO4, MgCl2, K2CO3, and SrBr2 - coupled with ASHPs across 4,800 representative households in 12 U.S. cities by embedding salt-hydrate-specific Ragone plots into a techno-economic model of coupled ASHP-TES operations. In Detroit, salt hydrate TES is projected to reduce household annual electricity costs by up to $\$$241 (8$\%$). Cost savings from TES can differ by over an order of magnitude between households and salt hydrates. We identify the most promising salt in this study, SrBr2, due to its high energy density and low humidification parasitic load. Break-even capital costs of SrBr2-based TES range from $\$$13/kWh to $\$$17/kWh, making it the only salt hydrate studied to reach and exceed the U.S. Department of Energy's $\$$15/kWh TES cost target. Sensitivities highlight the importance of variable TES sizing and efficiency losses in the value of TES.

Between 2001 and 2023, we obtained high spectral resolution mid-infrared observations of Io using the TEXES instrument at NASA's Infrared Telescope Facility. These observations were centered at 529.8 cm-1 (18.88 {\mu}m) and include several SO2 absorption lines. By modeling the shapes and strengths of these absorption lines, we are able to determine how Io's SO2 atmospheric density varies over the 22-year time period, covering nearly two Jovian years. Previous analysis has shown that the density of Io's atmosphere on the anti-Jovian hemisphere exhibits clear seasonal temporal variability, which can be modeled as the sum of a seasonally-varying frost sublimation component and a constant component, assumed to be volcanic. The new data show that the seasonal pattern repeats during the second Jovian year, confirming the importance of sublimation support. The considerable longitudinal variability in Io's atmospheric density found in previous work is also stable over the second Jovian year with the SO2 column density on the Jupiter-facing hemisphere being 5--8 times lower than the anti-Jovian hemisphere. For the first time, we detect seasonal variability on the Jupiter-facing hemisphere as well. This can also be modeled as a combination of sublimation and a small constant source. The lower atmospheric density on the Jupiter-facing hemisphere can plausibly be explained by the daily Jupiter eclipses, which decrease the surface temperature and therefore reduce the sublimation-driven component of the atmosphere, combined with a lower level of volcanic activity directly emitting SO2 into the atmosphere., Comment: Accepted in Icarus

We study the connection between collider and dark matter phenomenology in the singlet extension of the Georgi-Machacek model. In this framework, the singlet scalar serves as a suitable thermal dark matter (DM) candidate. Our focus lies on the region $v_{\chi}<1$ GeV, where $v_{\chi}$ is the common vacuum expectation value of the neutral components of the scalar triplets of the model. Setting bounds on the model parameters from theoretical, electroweak precision and LHC experimental constraints, we find that the BSM Higgs sector is highly constrained. Allowed values for the masses of the custodial fiveplets, triplets and singlet are restricted to the range $140~ {\rm GeV }< M_{H_5} < 350~ {\rm GeV }$, $150~ {\rm GeV }< M_{H_3} < 270 ~{\rm GeV }$ and $145~ {\rm GeV }< M_{H} < 300~ {\rm GeV }$. The extended scalar sector provides new channels for DM annihilation into BSM scalars that allow to satisfy the observed relic density constraint while being consistent with direct DM detection limits. The allowed region of the parameter space of the model can be explored in the upcoming DM detection experiments, both direct and indirect. In particular, the possible high values of BR$(H^0_5\to\gamma\gamma)$ can lead to an indirect DM signal within the reach of CTA. The same feature also provides the possibility of exploring the model at the High-Luminosity run of the LHC. In a simple cut-based analysis, we find that a signal of about $4\sigma$ significance can be achieved in final states with at least two photons for one of our benchmark points., Comment: 44 pages, 20 figures, 7 tables, Published as JHEP10(2024)058

Can web-based image processing and visualization tools easily integrate into existing websites without significant time and effort? Our Boostlet.js library addresses this challenge by providing an open-source, JavaScript-based web framework to enable additional image processing functionalities. Boostlet examples include kernel filtering, image captioning, data visualization, segmentation, and web-optimized machine-learning models. To achieve this, Boostlet.js uses a browser bookmark to inject a user-friendly plugin selection tool called PowerBoost into any host website. Boostlet also provides on-site access to a standard API independent of any visualization framework for pixel data and scene manipulation. Web-based Boostlets provide a modular architecture and client-side processing capabilities to apply advanced image-processing techniques using consumer-level hardware. The code is open-source and available., Comment: 5 pages, 5 figures

This study introduces 'clickbait spoiling', a novel technique designed to detect, categorize, and generate spoilers as succinct text responses, countering the curiosity induced by clickbait content. By leveraging a multi-task learning framework, our model's generalization capabilities are significantly enhanced, effectively addressing the pervasive issue of clickbait. The crux of our research lies in generating appropriate spoilers, be it a phrase, an extended passage, or multiple, depending on the spoiler type required. Our methodology integrates two crucial techniques: a refined spoiler categorization method and a modified version of the Question Answering (QA) mechanism, incorporated within a multi-task learning paradigm for optimized spoiler extraction from context. Notably, we have included fine-tuning methods for models capable of handling longer sequences to accommodate the generation of extended spoilers. This research highlights the potential of sophisticated text processing techniques in tackling the omnipresent issue of clickbait, promising an enhanced user experience in the digital realm., Comment: Accepted in ICON 2023

Granular flows with solid particles play an important role in energy and catalysis applications. To predict thermal performance, this study develops a comprehensive discrete particle tracking flow model fully coupled with conductive and radiative heat transfer. Leveraging LIGGGHTS, our model uniquely integrates particle-particle and particle-wall thermal radiation for grey radiative surfaces up to nearly 10 times of particle size. Models include: (a) plug flows of particles with a near-constant streamwise velocity and (b) gravity-driven, dense, moving beds of particles to determine wall-to-particle heat-transfer coefficients with isothermal channel walls. The plug flow model is used to interrogate the interdependent effects of flow-regime-dependent solid volume fractions (0.02 -- 0.48), particle size (0.4 -- 1 mm), operating wall temperatures (700 -- 1300 K), and thermophysical particles properties including emissivity (0.1 -- 1) and thermal conductivity (0.16, 33 W/m/K). The overall and conductive heat-transfer coefficients increase with solid volume fraction, while radiation contribution decreases. For a fixed solid volume fraction, heat-transfer coefficients exhibit linear dependence on particle emissivity and cubic dependence on wall temperature, driven by enhanced thermal radiation. For a fixed mass flow rate, the dependence of radiative heat-transfer coefficient on particle size changes with solid volume fraction. To generalize the applicability of our results, heat-transfer coefficients are transformed into a dimensionless Nusselt number, and establish its dependence on solid volume fraction. These results reveal the high sensitivity of the overall Nusselt number to even small changes in solid volume fraction in the dense flow regime, particularly for the high-conductivity alumina particles. In contrast, the radiative Nusselt number asymptotes for large solid volume fractions.

We describe an application of tropical moduli spaces to complex dynamics. A post-critically finite branched covering $\varphi$ of $S^2$ induces a pullback map on the Teichm\"uller space of complex structures of $S^2$; this descends to an algebraic correspondence on the moduli space of point-configurations of $\mathbb{C}\mathbb{P}^1$. We make a case for studying the action of the tropical moduli space correspondence by making explicit the connections between objects that have come up in one guise in tropical geometry and in another guise in complex dynamics. For example, a Thurston obstruction for $\varphi$ corresponds to a ray that is fixed by the tropical moduli space correspndence, and scaled by a factor $\ge 1$. This article is intended to be accessible to algebraic and tropical geometers as well as to complex dynamicists., Comment: 15 pages

In this study, we address the challenge of enabling large language models (LLMs) to consistently adhere to emotional support strategies in extended conversations. We focus on the steerability of the Llama-2 and Llama-3 suite of models, examining their ability to maintain these strategies throughout interactions. To assess this, we introduce the Strategy Relevant Attention (SRA) metric, which quantifies the model's adherence to the prompted strategy through attention maps. To facilitate our study, we create a strategy-conditioned synthetic conversational dataset derived from the ESConv dataset. We also propose various baselines informed by our proposed SRA metric to address the challenge and propose a fine-tuned model that significantly enhances the steerability of the base model in following the strategy throughout the conversation. The code and data are publicly available on our GitHub.

In a previous publication, we studied the parameter space of the phenomenological Minimal Supersymmetric Standard Model (pMSSM) with a light neutralino thermal dark matter ($M_{\tilde{\chi}_1^0} \leq M_h/2$) and observed that the recent results from the dark matter and collider experiments put strong constraints on this scenario. In this work, we present in detail the arguments behind the robustness of this result against scanning over the large number of parameters in pMSSM. The Run-3 of LHC will be crucial in probing the surviving regions of the parameter space. We further investigate the impact of light staus on our parameter space and also provide benchmarks which can be interesting for Run-3 of LHC. We analyse these benchmarks at the LHC using the machine learning framework of \texttt{XGBOOST}. Finally, we also discuss the effect of non-standard cosmology on the parameter space., Comment: 34 pages, 22 figures, 8 tables

State-of-the-art conversational AI systems raise concerns due to their potential risks of generating unsafe, toxic, unethical, or dangerous content. Previous works have developed datasets to teach conversational agents the appropriate social paradigms to respond effectively to specifically designed hazardous content. However, models trained on these adversarial datasets still struggle to recognize subtle unsafe situations that appear naturally in conversations or introduce an inappropriate response in a casual context. To understand the extent of this problem, we study prosociality in both adversarial and casual dialog contexts and audit the response quality of general-purpose language models in terms of propensity to produce unsafe content. We propose a dual-step fine-tuning process to address these issues using a socially aware n-pair contrastive loss. Subsequently, we train a base model that integrates prosocial behavior by leveraging datasets like Moral Integrity Corpus (MIC) and ProsocialDialog. Experimental results on several dialog datasets demonstrate the effectiveness of our approach in generating socially appropriate responses., Comment: SCI-CHAT@EACL2024