Hasil untuk "Municipal refuse. Solid wastes"

Chenyue Zhou, Jiayi Tuo, Shitong Qin et al.

The automated extraction of structured questions from paper-based mathematics exams is fundamental to intelligent education, yet remains challenging in real-world settings due to severe visual noise. Existing benchmarks mainly focus on clean documents or generic layout analysis, overlooking both the structural integrity of mathematical problems and the ability of models to actively reject incomplete inputs. We introduce MathDoc, the first benchmark for document-level information extraction from authentic high school mathematics exam papers. MathDoc contains \textbf{3,609} carefully curated questions with real-world artifacts and explicitly includes unrecognizable samples to evaluate active refusal behavior. We propose a multi-dimensional evaluation framework covering stem accuracy, visual similarity, and refusal capability. Experiments on SOTA MLLMs, including Qwen3-VL and Gemini-2.5-Pro, show that although end-to-end models achieve strong extraction performance, they consistently fail to refuse illegible inputs, instead producing confident but invalid outputs. These results highlight a critical gap in current MLLMs and establish MathDoc as a benchmark for assessing model reliability under degraded document conditions. Our project repository is available at \href{https://github.com/winnk123/papers/tree/master}{GitHub repository}

Alex Frankel, Navin Kartik

Signaling is wasteful. But how wasteful? We study the fraction of surplus dissipated in a separating equilibrium. For isoelastic environments, this waste ratio has a simple formula: $β/(β+σ)$, where $β$ is the benefit elasticity (reward to higher perception) and $σ$ is the elasticity of higher types' relative cost advantage. The ratio is constant across types and independent of other parameters, including convexity of cost in the signal. A constant waste ratio characterizes the isoelastic class. In winner-take-all signaling tournaments with $N$ candidates, exactly $(N-1)/N$ of the surplus dissipates -- the same as in Tullock contests.

Pravin Sankhwar, Khushabu Sankhwar

Processes for generating clean hydrogen from waste plastics through thermochemical methods such as pyrolysis and gasification are a promising solution for both waste management and clean energy initiatives. Then, this derived hydrogen powers the fuel cell, which produces electricity that can be directly fed to charge electric vehicles (EVs). Although this complex process has many challenges related to energy efficiency during the conversion processes—starting from the generation of hydrogen from thermochemical processes and hydrogen storage and followed by fueling the fuel cells and charging EV infrastructure—the simplistic conceptual modeling developed for this research demonstrates how an ecosystem of such processes can be made feasible commercially. Clean hydrogen generated using known techniques reported in the literature is promising for commercialization, but harnessing hydrogen from plastics offers additional benefits, such as reducing greenhouse gas (GHG) emissions. Overall, the feasibility of clean hydrogen using this methodology is not limited by potential cost inefficiencies, especially when savings from GHG emissions reduction are taken into account. EVs have become commercially viable thanks to high-energy-density Li-ion batteries. And therefore, research continues to optimize charging performance through the integration of renewable energy and battery storage systems. This study examines another potential of clean hydrogen: its use as a power source in grids, especially V-2-G (vehicle-to-grid) systems. Additionally, direct current (DC) power from a fuel cell powers an EV charger at DC input voltages for e-ambulances. In particular, this designed system operates on DC voltages throughout the power system, combining high-voltage direct current (HVDC) lines, renewable energy sources, DC-DC converters, DC EV chargers, and other supporting components. The literature review identified gaps in plastics production, waste management, and processes for converting them into useful energy. The presented model is a stepping stone towards a novel, innovative process for clean hydrogen production to power electric vehicle charging infrastructure for emergency response systems in healthcare, thereby improving public safety. The limitations of the study would be governed by the effective establishment of locations where waste management services are performed (for example, landfills) and adoption by local government authorities with deregulated power systems.

Ioannis Kontodimos, Christos Evaggelou, Anatoli Rontogianni et al.

A rapidly emerging approach within the scientific community involves the utilization of waste streams for renewable energy generation, particularly through biomethane production. A key aspect of this approach lies in the co-digestion of diverse waste streams, which can enhance process efficiency and contribute to a more effective reduction in the organic load. The present study investigates the anaerobic digestion of a mixture of food waste leachates and dairy waste (cheese whey wastewater), with a dual objective: to evaluate the reduction in organic-load efficiency of the mixed substrate and to assess the production of biogas enriched in biomethane content. Three distinct mixing ratios by volume of the two waste streams (25%/75%, 50%/50% and 75%/25%) were subjected to an anaerobic digestion process under the same SIR. The performance of each mixture was assessed in terms of both reduction in organic-load efficiency and biomethane yield, followed by a comparative analysis to identify the optimal mixing ratio. The results indicate that while the organic-load reduction remains consistently effective across all mixing ratios, the biomethane production potential is notably higher for the 25%/75% waste mixture, highlighting it as the most promising configuration for both energy recovery and waste treatment efficiency.

Fabian Hildebrandt, Andreas Maier, Patrick Krauss et al.

Refusal behavior in large language models (LLMs) enables them to decline responding to harmful, unethical, or inappropriate prompts, ensuring alignment with ethical standards. This paper investigates refusal behavior across six LLMs from three architectural families. We challenge the assumption of refusal as a linear phenomenon by employing dimensionality reduction techniques, including PCA, t-SNE, and UMAP. Our results reveal that refusal mechanisms exhibit nonlinear, multidimensional characteristics that vary by model architecture and layer. These findings highlight the need for nonlinear interpretability to improve alignment research and inform safer AI deployment strategies.

Shengyun Si, Xinpeng Wang, Guangyao Zhai et al.

Recent advancements in large language models (LLMs) have demonstrated that fine-tuning and human alignment can render LLMs harmless. In practice, such "harmlessness" behavior is mainly achieved by training models to reject harmful requests, such as "Explain how to burn down my neighbor's house", where the model appropriately declines to respond. However, this approach can inadvertently result in false refusal, where models reject benign queries as well, such as "Tell me how to kill a Python process". In this work, we demonstrate that prompting safety reflection before generating a response can mitigate false refusal behavior. Building on this finding, we introduce the Think-Before-Refusal (TBR) schema and conduct safety-aware instruction fine-tuning incorporating safety reflection. In an ablation study across 15 pre-trained models, we show that models fine-tuned with safety reflection significantly reduce false refusal behavior while maintaining safety and overall performance compared to those fine-tuned without safety reflection.

Rom Himelstein, Amit LeVi, Brit Youngmann et al.

Safety-aligned large language models (LLMs) are becoming increasingly widespread, especially in sensitive applications where fairness is essential and biased outputs can cause significant harm. However, evaluating the fairness of models is a complex challenge, and approaches that do so typically utilize standard question-answer (QA) styled schemes. Such methods often overlook deeper issues by interpreting the model's refusal responses as positive fairness measurements, which creates a false sense of fairness. In this work, we introduce the concept of silenced biases, which are unfair preferences encoded within models' latent space and are effectively concealed by safety-alignment. Previous approaches that considered similar indirect biases often relied on prompt manipulation or handcrafted implicit queries, which present limited scalability and risk contaminating the evaluation process with additional biases. We propose the Silenced Bias Benchmark (SBB), which aims to uncover these biases by employing activation steering to reduce model refusals during QA. SBB supports easy expansion to new demographic groups and subjects, presenting a fairness evaluation framework that encourages the future development of fair models and tools beyond the masking effects of alignment training. We demonstrate our approach over multiple LLMs, where our findings expose an alarming distinction between models' direct responses and their underlying fairness issues.

Alexandra Abbas, Nora Petrova, Helios Ael Lyons et al.

Recent work has shown that language models' refusal behavior is primarily encoded in a single direction in their latent space, making it vulnerable to targeted attacks. Although Latent Adversarial Training (LAT) attempts to improve robustness by introducing noise during training, a key question remains: How does this noise-based training affect the underlying representation of refusal behavior? Understanding this encoding is crucial for evaluating LAT's effectiveness and limitations, just as the discovery of linear refusal directions revealed vulnerabilities in traditional supervised safety fine-tuning (SSFT). Through the analysis of Llama 2 7B, we examine how LAT reorganizes the refusal behavior in the model's latent space compared to SSFT and embedding space adversarial training (AT). By computing activation differences between harmful and harmless instruction pairs and applying Singular Value Decomposition (SVD), we find that LAT significantly alters the refusal representation, concentrating it in the first two SVD components which explain approximately 75 percent of the activation differences variance - significantly higher than in reference models. This concentrated representation leads to more effective and transferable refusal vectors for ablation attacks: LAT models show improved robustness when attacked with vectors from reference models but become more vulnerable to self-generated vectors compared to SSFT and AT. Our findings suggest that LAT's training perturbations enable a more comprehensive representation of refusal behavior, highlighting both its potential strengths and vulnerabilities for improving model safety.

Vincent Siu, Nicholas Crispino, Zihao Yu et al.

Large Language Models (LLMs) encode behaviors such as refusal within their activation space, yet identifying these behaviors remains a significant challenge. Existing methods often rely on predefined refusal templates detectable in output tokens or require manual analysis. We introduce \textbf{COSMIC} (Cosine Similarity Metrics for Inversion of Concepts), an automated framework for direction selection that identifies viable steering directions and target layers using cosine similarity - entirely independent of model outputs. COSMIC achieves steering performance comparable to prior methods without requiring assumptions about a model's refusal behavior, such as the presence of specific refusal tokens. It reliably identifies refusal directions in adversarial settings and weakly aligned models, and is capable of steering such models toward safer behavior with minimal increase in false refusals, demonstrating robustness across a wide range of alignment conditions.

Utsav Maskey, Sumit Yadav, Mark Dras et al.

LLMs increasingly exhibit over-refusal behavior, where safety mechanisms cause models to reject benign instructions that seemingly resemble harmful content. This phenomenon diminishes utility in production applications that repeatedly rely on common prompt templates or applications that frequently rely on LLMs for specific tasks (e.g. sentiment analysis, language translation). Through extensive evaluation, we demonstrate that LLMs persist in refusing inputs containing harmful content, even when they are reframed with tasks that have benign intent. Our mechanistic analysis reveals that LLMs follow distinct "constellation" patterns in embedding space as representations traverse layers, with each NLP task maintaining consistent trajectories that shift predictably between refusal and non-refusal cases. We introduce SafeConstellations, an inference-time trajectory-shifting approach that tracks task-specific trajectory patterns and guides representations toward non-refusal pathways. By selectively guiding model behavior only on tasks prone to over-refusal, our method reduces over-refusal rates by up to 73% with minimal impact on utility -- offering a principled and conditional approach to mitigating over-refusals.

Eric Hanchen Jiang, Weixuan Ou, Run Liu et al.

Safety alignment of large language models currently faces a central challenge: existing alignment techniques often prioritize mitigating responses to harmful prompts at the expense of overcautious behavior, leading models to incorrectly refuse benign requests. A key goal of safe alignment is therefore to improve safety while simultaneously minimizing false refusals. In this work, we introduce Energy Landscape Steering (ELS), a novel, fine-tuning free framework designed to resolve this challenge through dynamic, inference-time intervention. We train a lightweight external Energy-Based Model (EBM) to assign high energy to undesirable states (false refusal or jailbreak) and low energy to desirable states (helpful response or safe reject). During inference, the EBM maps the LLM's internal activations to an energy landscape, and we use the gradient of the energy function to steer the hidden states toward low-energy regions in real time. This dynamically guides the model toward desirable behavior without modifying its parameters. By decoupling behavioral control from the model's core knowledge, ELS provides a flexible and computationally efficient solution. Extensive experiments across diverse models demonstrate its effectiveness, raising compliance on the ORB-H benchmark from 57.3 percent to 82.6 percent while maintaining baseline safety performance. Our work establishes a promising paradigm for building LLMs that simultaneously achieve high safety and low false refusal rates.

Yuki Uematsu

Hydrophobic interfaces have unique physicochemical properties and are used in various chemical products such as food, cosmetics, soap, and medicine and technologies such as pan coating and ski wax. In this chapter, we describe the fundamental concept of hydrophobic interfaces and explain their ion adsorption and zeta potential by using experimental data from the literature. Thus far, these electrical properties are considered universal for solid/water, liquid/water, and gas/water interfaces; however, a careful comparison in this chapter will reveal significant differences among them. To confirm that the affinity of H$^+$ ions for all hydrophobic interfaces is stronger than that of OH$^-$ ions, more experimental data on hydrophobic liquid/water and solid/water interfaces are required.

Leonel Díaz-Tato, Jesús Fernando López-Perales, Yadira González-Carranza et al.

Developing sustainable ceramic formulations that integrate industrial by-products addresses the high energy and raw material demands of refractory manufacturing while advancing circular economy goals. This study investigates the recycling of chamotte waste from rejected fired electrical porcelain as a partial substitute (5 and 10 wt.%) for flint clay in aluminosilicate refractory castables. Samples were fired at 110, 815, 1050, and 1400 °C and evaluated for bulk density, apparent porosity, cold crushing strength, and flexural strength. Microstructural and mineralogical changes were analyzed by SEM and XRD. Incorporating 10 wt.% chamotte waste fostered an in situ mullite-reinforced microstructure, enhancing mechanical strength (58 MPa—CCS, 18.8 MPa—MOR) and lowering porosity (24.4%), demonstrating chamotte’s dual role as recycled raw material and reinforcement phase for densification and durability. These properties matched or surpassed those of the conventional formulation, with strength improvements of up to 44%. The findings demonstrate that high-temperature industrial waste can be effectively valorized in advanced refractories, reducing reliance on virgin raw materials, diverting waste from landfills, and promoting industrial symbiosis within the ceramics and metallurgical sectors.

Andrei Stsiapanau

Amidst a period marked by growing volumes of nuclear waste and ongoing discussions regarding its management, technologies that utilise natural materials for containment are gaining prominence. This article takes a historical view of Russian nuclear waste management practices with a focus on the role of clay as a natural material for containing nuclear waste. In particular, it explores the use of clay in multi-barrier technology, highlighting its dual role as a protective layer and a resource for managing nuclear safety risks. The siting of the liquid nuclear waste disposal at the Ignalina NPP site in Lithuania (1976–1980) and of solid nuclear waste disposal at Leningrad NPP in Sosnovy Bor, Russia (2013–2018) are the main foci of this article. These cases contribute to understanding nuclear waste disposal siting in the USSR and modern Russia and enable analysis of nuclear waste discourses describing the sites’ geology as a static or dynamic environment within active or passive safety systems.

Pérez Belmonte Nancy Merab, Sandoval Torres Sadoth, Belmonte Jiménez Salvador Isidro

Methane is a powerful greenhouse gas and short-lived climate pollutant generated in landfills. In this work, five first-order decay models were implemented to estimate the methane emissions from a landfill near Oaxaca city. The five models were the simple first-order decay model, the modified first-order decay model, the multiphase model, the LandGem model, and the Intergovernmental Panel on Climate Change (IPCC) model. An autoregressive integrated moving average (ARIMA) model was built to predict waste generation, and a gravimetric method was used to estimate the volume of stored waste. The ARIMA model correctly predicted the generation of municipal solid waste, calculating 108,202 tons of solid waste in the landfill for the year 2022. In terms of the models and considering the experimental data measured in 2020, the simple model and the simple modified model were more accurate, with 3.50 × 10⁶ m³ (relative error = 1.0) and 3.76 × 10⁶ m³ of methane (relative error = 6.3), respectively. The multiphase model calculated a value of 3.09 × 10⁶ m³ of methane (relative error = 12.6), the LandGEM model calculated a value of 4.97 × 10⁶ m³ (40.7), and the IPCC model calculated a value of 3.19 × 10⁶ m³ (relative error = 9.7). The LandGEM model was improved when the standard values proposed by the Environmental Protection Agency (EPA) were considered. According to the simple model and the simple modified model, by 2050, the landfill will emit 1.22 × 10⁶ m³ and 1.37 × 10⁶ m³, demonstrating that important methane emissions will be released in the decades to come. This information is important for the implementation of methane mitigation strategies, to which Mexico has committed in the Global Methane Initiative.

Pin-Han Chen, Jun-Yi Wu

This study introduces a novel two-stage circular economy model to transform organic waste materials—bean dregs and coffee grounds—into functional products: eco-friendly cat litter and organic fertilizer. The hypothesis was that integrating vermicompost and diatomaceous earth with these waste materials would enhance the functional properties of cat litter while ensuring its recyclability into high-quality fertilizer. In the first stage, a combination of bean dregs, coffee grounds, vermicompost, and diatomaceous earth was optimized using the Taguchi method, achieving cat litter with superior water absorption and clumping performance. In the second stage, the spent cat litter was rapidly composted, producing a nutrient-rich organic fertilizer. The fertilizer’s efficacy was validated through a potting experiment with lettuce, where a 10% application rate promoted optimal growth without causing nutrient toxicity. This innovative approach offers a sustainable solution to waste management challenges while contributing to environmentally friendly agricultural practices. Future research could investigate incorporating additional waste streams and enhancing composting efficiency for broader implementation.

James Darmey, Satyanarayana Narra, Osei-Wusu Achaw et al.

Anaerobic digestion (AD) is a sustainable method of treating organic waste to generate methane-rich biogas. However, the complex lignocellulosic nature of organic waste in most cases limits its biodegradability and methane potential. This review evaluates pretreatment technology to optimize AD performance, particularly in developing countries like Ghana, where organic waste remains underutilized. A narrative synthesis of the literature between 2010 and 2024 was conducted through ScienceDirect and Scopus, categorizing pretreatment types as mechanical, thermal, chemical, biological, enzymatic, and hybrid. A bibliometric examination using VOSviewer also demonstrated global trends in research and co-authorship networks. Mechanical and thermal pretreatments increased biogas production by rendering the substrate more available, while chemical treatment degraded lignin and hemicellulose, sometimes more than 100% in methane yield. Biological and enzymatic pretreatments were energy-consuming and effective, with certain enzymatic blends achieving 485% methane yield increases. The study highlights the synergistic benefits of hybrid approaches and growing global interest, as revealed by bibliometric analysis; hence, the need to explore their potential in Ghana. In Ghana, this study concludes that low-cost, biologically driven pretreatments are practical pathways for advancing anaerobic digestion systems toward sustainable waste management and energy goals, despite infrastructure and policy challenges.

Julia Weißert, Kristina Henzler, Shimelis Kebede Kassahun

In sub-Saharan Africa, municipal solid waste management faces significant challenges, including inadequate infrastructure, increasing waste generation, and limited resources, leading to severe environmental and public health issues. Innovations in waste management are essential to address these pressing problems, as they can enhance efficiency, reduce pollution, and promote sustainable practices while fostering sustainable development. To select sustainable and contextually relevant solutions, it is vital to investigate their potential sustainability impacts based on the Sustainable Development Goals (SDGs) beforehand and to involve local stakeholders in the innovation process. Besides, engaging stakeholders increases community buy-in and fosters collaboration, leading to more effective and sustainable outcomes. This paper develops and applies a sustainability assessment methodology for innovations in municipal solid waste management systems in sub-Saharan Africa, with a case study in Ethiopia. The proposed methodology emphasizes the importance of involving local stakeholders in the SDG-based indicator assessment and offers suggestions for a data collection strategy. The case study on a composting process in Bishoftu Town demonstrates that stakeholder participation in selecting innovations positively influences the outcomes. However, the analysis indicates mixed effects of the innovation in the three sustainability dimensions, highlighting areas for optimization. Consequently, the presented method can support the innovation process of municipal solid waste management systems, fostering sustainable municipal development.

Luiz Eduardo Nochi Castro, Tiago Linhares Cruz Tabosa Barroso, Vanessa Cosme Ferreira et al.

This study investigates the properties of Benitaka grape pomace (<i>Vitis vinifera</i> L.), a byproduct of the wine industry, focusing on its potential for applications in the circular economy and biorefinery processes. The analysis covers a range of physical, chemical, and structural characteristics, including the composition of proteins, moisture, lipids, ash, sugars, fiber fractions (such as neutral-detergent fiber, cellulose, lignin, and hemicellulose), pH, acidity, gross energy, as well as bioactive compounds such as total phenolics, flavonoids, anthocyanins, and antioxidant capacity. Advanced characterization techniques, such as nitrogen adsorption/desorption isotherms, Fourier-transform infrared spectroscopy, differential scanning calorimetry, scanning electron microscopy, and high-performance liquid chromatography coupled with mass spectrometry, were employed. The results revealed an acidic pH of 4.05 and a titratable acidity of 1.25 g of tartaric acid per 100 g. The gross energy was 3764 kcal kg⁻¹, indicating high energy capacity, similar to wood chips. The pomace exhibited high hygroscopicity (31 to 50 g of moisture per 100 g), high levels of fiber, cellulose, and lignin, as well as bioactive compounds with significant values of total phenolics (5956.56 mg GAE 100 g⁻¹), flavonoids (1958.33 mg CAT 100 g⁻¹), and anthocyanins (66.92 mg C3G 100 g⁻¹). Antioxidant analysis showed promising results, with DPPH and FRAP values of 20.12 and 16.85 μmol TEAC g⁻¹ of extract, respectively. This study not only validates existing data but also provides new insights into the composition of hemicellulose and lignocellulosic phase transitions, highlighting grape pomace as a promising resource for sustainability in industry and biorefinery processes.

Marcelo Corrales Compagnucci

On July 14th, 2022, the Danish Data Protection Authority issued a reprimand against Helsingor Municipality. It imposed a general ban on using Google Chromebooks and Google Workspace for education in primary schools in the Municipality. The Danish DPA banned such processing and suspended any related data transfers to the United States (U.S.) until it is brought in line with the General Data Protection Regulation (GDPR). The suspension took effect immediately, and the Municipality had until August 3rd, 2022, to withdraw and terminate the processing, as well as delete data already transferred. Finally, in a new decision on August 18th, 2022, the Danish DPA has ratified the ban to the use of Google Chromebooks and Workspace. In the eyes of the Danish DPA, the Municipality failed for example to document that they have assessed and reduced the relevant risks to the rights and freedoms of the pupils. This article is structured as follows: section II provides the background concerning the unfolding events after the Schrems II ruling. Section III discusses the origins and facts of the Danish DPA case. Section IV examines the reasoning and critical findings of the Danish DPA decision. Finally, section V concludes with some general recommendations the Danish municipalities must follow based on the ensuing effects stemming from this case.