Controllable Automatic Text Simplification (CATS) produces user-tailored outputs, yet controllability is often treated as a decoding problem and evaluated with metrics that are not reflective to the measure of control. We observe that controllability in ATS is significantly constrained by data and evaluation. To this end, we introduce a domain-agnostic CATS framework based on instruction fine-tuning with discrete control tokens, steering open-source models to target readability levels and compression rates. Across three model families with different model sizes (Llama, Mistral, Qwen; 1-14B) and four domains (medicine, public administration, news, encyclopedic text), we find that smaller models (1-3B) can be competitive, but reliable controllability strongly depends on whether the training data encodes sufficient variation in the target attribute. Readability control (FKGL, ARI, Dale-Chall) is learned consistently, whereas compression control underperforms due to limited signal variability in the existing corpora. We further show that standard simplification and similarity metrics are insufficient for measuring control, motivating error-based measures for target-output alignment. Finally, our sampling and stratification experiments demonstrate that naive splits can introduce distributional mismatch that undermines both training and evaluation.
Precise segmentation of irregular and densely arranged components is essential for robotic disassembly and material recovery in electronic waste (e-waste) recycling. This study evaluates the impact of model architecture and scale on segmentation performance by comparing SAM2, a transformer-based vision model, with the lightweight YOLOv8 network. Both models were trained and tested on a newly collected dataset of 1,456 annotated RGB images of laptop components including logic boards, heat sinks, and fans, captured under varying illumination and orientation conditions. Data augmentation techniques, such as random rotation, flipping, and cropping, were applied to improve model robustness. YOLOv8 achieved higher segmentation accuracy (mAP50 = 98.8%, mAP50-95 = 85%) and stronger boundary precision than SAM2 (mAP50 = 8.4%). SAM2 demonstrated flexibility in representing diverse object structures but often produced overlapping masks and inconsistent contours. These findings show that large pre-trained models require task-specific optimization for industrial applications. The resulting dataset and benchmarking framework provide a foundation for developing scalable vision algorithms for robotic e-waste disassembly and circular manufacturing systems.
Artificial intelligence (AI) has made remarkable progress in recent years, yet its rapid expansion brings overlooked environmental and ethical challenges. This review explores four critical areas where AI's impact extends beyond performance: energy consumption, electronic waste (e-waste), inequality in compute access, and the hidden energy burden of cybersecurity systems. Drawing from recent studies and institutional reports, the paper highlights systemic issues such as high emissions from model training, rising hardware turnover, global infrastructure disparities, and the energy demands of securing AI. By connecting these concerns, the review contributes to Responsible AI discourse by identifying key research gaps and advocating for sustainable, transparent, and equitable development practices. Ultimately, it argues that AI's progress must align with ethical responsibility and environmental stewardship to ensure a more inclusive and sustainable technological future.
Nico Hofmann, Julian Dauenhauer, Nils Ole Dietzler
et al.
Scientific text is complex as it contains technical terms by definition. Simplifying such text for non-domain experts enhances accessibility of innovation and information. Politicians could be enabled to understand new findings on topics on which they intend to pass a law, or family members of seriously ill patients could read about clinical trials. The SimpleText CLEF Lab focuses on exactly this problem of simplification of scientific text. Task 1.1 of the 2025 edition specifically handles the simplification of complex sentences, so very short texts with little context. To tackle this task we investigate the identification of complex terms in sentences which are rephrased using small Gemini and OpenAI large language models for non-expert readers.
Ngoc-Bao-Quang Nguyen, Tuan-Minh Do, Cong-Tam Phan
et al.
Automated image-based garbage classification is a critical component of global waste management; however, systematic benchmarks that integrate Machine Learning (ML), Deep Learning (DL), and efficient hybrid solutions remain underdeveloped. This study provides a comprehensive comparison of three paradigms: (1) machine learning algorithms using handcrafted features, (2) deep learning architectures, including ResNet variants and EfficientNetV2S, and (3) a hybrid approach that utilizes deep models for feature extraction combined with classical classifiers such as Support Vector Machine and Logistic Regression to identify the most effective strategy. Experiments on three public datasets - TrashNet, Garbage Classification, and a refined Household Garbage Dataset (with 43 corrected mislabels)- demonstrate that the hybrid method consistently outperforms the others, achieving up to 100% accuracy on TrashNet and the refined Household set, and 99.87% on Garbage Classification, thereby surpassing state-of-the-art benchmarks. Furthermore, feature selection reduces feature dimensionality by over 95% without compromising accuracy, resulting in faster training and inference. This work establishes more reliable benchmarks for waste classification and introduces an efficient hybrid framework that achieves high accuracy while reducing inference cost, making it suitable for scalable deployment in resource-constrained environments.
With the energy crisis and climate warming, the position of a new generation of smart windows is becoming increasingly important, and materials or systems that can have high blocking of near-infrared (NIR) and ultraviolet (UV) and high transmittance of visible light (VIS) are needed. Currently, it is difficult for smart heat-insulation materials to achieve high transmittance of VIS, good UV isolation, outstanding cooling and thermal insulation, and excellent waste heat collection. Here, we design a novel composite hydrogel to achieve an average 92% VIS transmittance, efficient UV absorption , 11 Celsius degree of thermal insulation, and sensing properties. Interestingly, we designed a transparent heat insulation system with this composite hydrogel to obtain about 22 Celsius degree of the record-breaking insulation performance for 168 hours, waste heat collection and reutilization, and temperature sensing. Our findings provide new ideas and possibilities for designing transparent and heat-insulation smart window systems.
Yunzhen Feng, Parag Jain, Anthony Hartshorn
et al.
Reinforcement learning with verifiable rewards (RLVR) has become a standard recipe for improving large language models (LLMs) on reasoning tasks, with Group Relative Policy Optimization (GRPO) widely used in practice. Yet GRPO wastes substantial compute on negative groups: groups in which no sampled response is correct yield zero advantage and thus no gradient. We ask whether negative groups can be leveraged without extra supervision. Starting from a maximum-likelihood (MLE) objective in reward modeling, we show that the MLE gradient is equivalent to a policy gradient for a modified value function. This value function adds a confidence-weighted penalty on incorrect responses, imposing larger penalties on more confident mistakes. We refer to this as \textbf{L}ikelihood \textbf{E}stimation with \textbf{N}egative \textbf{S}amples (\textbf{LENS}). LENS modifies GRPO to assign non-zero, confidence-dependent rewards to incorrect generations, making negative groups informative and converting previously wasted samples into useful gradient updates. On the MATH benchmark with Llama-3.1-8B and Qwen-2.5-3B, the proposed variant consistently outperforms GRPO baseline, with significant gains on harder items. These results demonstrate a principled and practical way to "rescue" negative groups, improving efficiency and performance in RLVR.
The United States Department of Energy has long term goals to develop solutions for managing the nations spent nuclear fuel and high-level waste inventory. The Integrated Waste Management program is employing system-level engineering and analysis principles to inform potential future waste management system architectures. Managing the SNF requires the use of system-level analysis software that considers waste generation, on-site (centralized storage), transportation infrastructure, and long-term disposal. The Next Generation System Analysis Model is an agent-based model that was developed to simulate the transportation and storage of SNF and HLW. NGSAM has the capability to detail the interaction and movement of individual components and groups, such as rail cars and casks. The SNF inventory from commercial nuclear reactors is currently in temporary storage at multiple locations spread across the US. Shipping of SNF from these locations relies on one of three transportation modes: rail, heavy-haul truck, or barge. Rail is the most preferred due to the size of the canisters and casks the SNF would be shipped in. Under some scenarios, a rail route might not be available to a reactor site or improving the rail infrastructure at shutdown sites might be too cost-prohibitive for utilities to opt for a direct rail transfer. Under such scenarios, using a barge or heavy haul truck to de-inventory the site and transfer the SNF to a nearby intermodal transfer site with adequate rail infrastructure where the payload could be transferred to a rail car might prove to be an attractive option. This work presents the various intermodal transportation options to move SNF from reactor sites to rail cars. Next, the operational steps in each of these modes to move the SNF from a reactor site and transfer it to a rail car is explored. The ideology, assumptions, and future steps are presented.
The Spent Fuel and Waste Science and Technology (SFWST) Campaign of the U.S. Department of Energy Office of Nuclear Energy (DOE-NE), Office of Spent Fuel and Waste Disposition is conducting research and development (R&D) on geologic disposal of spent nuclear fuel (SNF) and high-level nuclear waste (HLW). This work includes the Geologic Disposal Safety Assessment (GDSA) program which is charged with development of generic deep geologic repository concepts and system performance assessment models. One part of the GDSA framework is the development of a biosphere model capable of assessing dose to potential receptors exposed to radionuclides released via groundwater from geologic disposal sites. This work includes the development of a biosphere model capable of estimating doses to potential receptors living in the biosphere and exposed to radionuclides released from a hypothetical geologic disposal site. The GDSA biosphere model is being developed so that it is compatible with the GDSA framework including PFLOTRAN, the massively parallel subsurface flow and reactive transport code. PFLOTRAN is a subsurface flow and reactive transport code that solves a system of partial differential equations for multiphase flow and transport of components in porous materials such as shale formation. PFLOTRAN simultaneously simulates energy and mass flow with fluid properties as function of pressure and temperature through equations of state. PFLOTRAN also solves the mass conservation and transport equations of multicomponent formulations of aqueous chemical species, gases, and minerals reactive transport. It contains a waste form process model that simulates the radionuclide inventory under potential failures in a geological repository. PFLOTRAN can thus calculate the source term of radionuclides in ground water which can then be used as the GDSA biosphere model input.
We present an evaluation of text simplification (TS) in Spanish for a production system, by means of two corpora focused in both complex-sentence and complex-word identification. We compare the most prevalent Spanish-specific readability scores with neural networks, and show that the latter are consistently better at predicting user preferences regarding TS. As part of our analysis, we find that multilingual models underperform against equivalent Spanish-only models on the same task, yet all models focus too often on spurious statistical features, such as sentence length. We release the corpora in our evaluation to the broader community with the hopes of pushing forward the state-of-the-art in Spanish natural language processing.
Eleftheria Papadopoulou, Charlene Vance, Paloma S. Rozene Vallespin
et al.
To substitute petroleum-based materials with bio-based alternatives, microbial fermentation combined with inexpensive biomass is suggested. In this study Saccharina latissima hydrolysate, candy-factory waste, and digestate from full-scale biogas plant were explored as substrates for lactic acid production. The lactic acid bacteria Enterococcus faecium, Lactobacillus plantarum, and Pediococcus pentosaceus were tested as starter cultures. Sugars released from seaweed hydrolysate and candy-waste were successfully utilized by the studied bacterial strains. Additionally, seaweed hydrolysate and digestate served as nutrient supplements supporting microbial fermentation. According to the highest achieved relative lactic acid production, a scaled-up co-fermentation of candy-waste and digestate was performed. Lactic acid reached a concentration of 65.65 g/L, with 61.69% relative lactic acid production, and 1.37 g/L/hour productivity. The findings indicate that lactic acid can be successfully produced from low-cost industrial residues.
Ilaria Burò, Valeria Consoli, A. Castellano
et al.
The awareness of the large amount of waste produced along the food chain, starting in the agricultural sector and continuing across industrial transformation to the domestic context, has in recent years also aroused strong concern amongst the public, who are ing about the possible consequences that this could have on environmental sustainability, resource waste and human health. The aim of the present research is the recovery of substances with high added value from waste and by-products typical of the Mediterranean area, such as the residue from the industrial processing of red oranges, called pastazzo (peels, pulps and seeds), which is particularly rich in anthocyanins, flavanones and hydroxycinnamic acids, and has numerous nutraceutical properties, as well as the olive leaves coming from olive-tree pruning, which are rich in substances such as oleuropein, elenolic acid, hydroxytyrosol, tyrosol and rutin. The effect of Red Orange Extract (ROE) and Olive Leaf Extract (OLE) on HepG2 fatty storage capacity was assessed performing Oil Red O’ staining, and antioxidant properties of the extracts were evaluated following the steatosis model onset. Based on the results obtained, the preparation of natural extracts that are derived from these waste products can be useful for preventing, counteracting or delaying the onset of the complications of fatty liver disease, such as hepatic steatosis.
We propose GRS: an unsupervised approach to sentence simplification that combines text generation and text revision. We start with an iterative framework in which an input sentence is revised using explicit edit operations, and add paraphrasing as a new edit operation. This allows us to combine the advantages of generative and revision-based approaches: paraphrasing captures complex edit operations, and the use of explicit edit operations in an iterative manner provides controllability and interpretability. We demonstrate these advantages of GRS compared to existing methods on the Newsela and ASSET datasets.
Mertol Gökelma, Alicia Vallejo-Olivares, G. Tranell
The increasing use of aluminimum in packaging applications results in many different aluminium-based products ending up in consumer mixed-waste bins. This waste is typically incinerated, generating an aluminium-containing bottom ash. The current work investigates the recyclability of the aluminium fraction in the bottom ash from waste incineration plants in the USA, UK and Denmark. Incinerated Al-samples from different size fractions (2-6 mm, 6-12 mm and 12-30 mm) were characterized in terms of inherent oxide thickness, re-melting yield/coagulation and composition. The measured average oxide thickness on Al particles was 68 µm (SD=100), with the metal yield and coagulation efficiency measured to between 76 and 92% and 87-99% respectively. Larger particle size fractions resulted in a higher metal yield due to their higher mass to surface ratio. A simplified model correlating metal yield and particle size was proposed. The aluminium content of the melted material was determined to between 95.6 and 98.5% with main impurities being Fe, Si, Mn, Zn, Mg and Cu, corresponding to major aluminium alloying elements and waste charge components.