Search

The evolution of natural life is guided by a perpetually adaptive set of rules, encompassing natural laws, human policies, and game mechanics. Automated game design, through the creation of simulated environments populated by AI agents, embodies these rules, aligning with the objectives of artificial life research that seeks to replicate the dynamics of biological life through computational models. This paper presents a comprehensive framework, the Rule Generation Networks (RGN), devised for automated rule design, evaluation, and evolution in line with controllable expectations. We refine and formalize three cardinal elements - rules, strategies, and evaluation - to elucidate the intricate relationships inherent in rule generation tasks. The RGN integrates generative neural networks for rule design and a suite of reinforcement learning models for rule evaluation. To exemplify rule evolution and adaptation across varying environments, we introduce a controllability metric to gauge game dynamics and evolve the rule designer accordingly. Furthermore, we develop two game environments, Maze Run and Trust Evolution, modelling human exploration and societal trade dynamics, to gamify and evaluate the generated rules.

Medical image segmentation is a critical task for clinical diagnosis and research. However, dealing with highly imbalanced data remains a significant challenge in this domain, where the region of interest (ROI) may exhibit substantial variations across different slices. This presents a significant hurdle to medical image segmentation, as conventional segmentation methods may either overlook the minority class or overly emphasize the majority class, ultimately leading to a decrease in the overall generalization ability of the segmentation results. To overcome this, we propose a novel approach based on multi-step reinforcement learning, which integrates prior knowledge of medical images and pixel-wise segmentation difficulty into the reward function. Our method treats each pixel as an individual agent, utilizing diverse actions to evaluate its relevance for segmentation. To validate the effectiveness of our approach, we conduct experiments on four imbalanced medical datasets, and the results show that our approach surpasses other state-of-the-art methods in highly imbalanced scenarios. These findings hold substantial implications for clinical diagnosis and research.

It is uncertain whether the power of transformer architectures can complement existing convolutional neural networks. A few recent attempts have combined convolution with transformer design through a range of structures in series, where the main contribution of this paper is to explore a parallel design approach. While previous transformed-based approaches need to segment the image into patch-wise tokens, we observe that the multi-head self-attention conducted on convolutional features is mainly sensitive to global correlations and that the performance degrades when these correlations are not exhibited. We propose two parallel modules along with multi-head self-attention to enhance the transformer. For local information, a dynamic local enhancement module leverages convolution to dynamically and explicitly enhance positive local patches and suppress the response to less informative ones. For mid-level structure, a novel unary co-occurrence excitation module utilizes convolution to actively search the local co-occurrence between patches. The parallel-designed Dynamic Unary Convolution in Transformer (DUCT) blocks are aggregated into a deep architecture, which is comprehensively evaluated across essential computer vision tasks in image-based classification, segmentation, retrieval and density estimation. Both qualitative and quantitative results show our parallel convolutional-transformer approach with dynamic and unary convolution outperforms existing series-designed structures.

Multifocus image fusion (MFIF), as an efficient way to improve the visual effect of images with partial focus defects, is of great significance in the field of image enhancement. According to the imaging principle of the lens, we summarize the visual salience priors (VSP) from the daily photo scene and two relationships from MFIF. Thereby, an edge-sensitive model for MFIF is presented in this study. Supported by VSP, we consider the correlation between salience object detection (SOD) and MFIF, and select the former as a pre-training task. SOD provides the network with realistic depth of field and bokeh effects to learn, and enhances the network’s ability to extract and express the edges of focused objects. Meanwhile, given the scarcity of real multifocus training sets, we propose a randomized approach to generate massive training sets and pseudo-labels based on limited unlabeled data. Besides, two attention modules are designed based on isometric domain transformation (IDT) in the traditional edge-preservation field. IDT removes interference information from feature maps in a low-cost manner, thereby facilitating channel-wise and spatial-wise weight assignments. Experimental results on four datasets show that the performance of our model is superior to that of many supervised models, without the need of any real MFIF training set.

Achieving stable long-term mainstream nitrite oxidizing bacteria (NOB) suppression is the bottleneck for the novel partial nitrification (PN) process toward energy- and carbon-efficient wastewater treatment. However, long-term PN stability remains a challenge due to NOB adaptation. This study proposed and demonstrated a novel strategy for achieving NOB suppression by the primary treatment of mainstream wastewater with a forward osmosis (FO) membrane process, which facilitated two external NOB inhibition factors (salinity and free nitrous acid, FNA). To evaluate the proposed strategy, a lab-scale sequencing batch reactor was operated for 200 days. A stable PN operation was achieved with a nitrite accumulation ratio of 97.7 ± 2.8%. NOB were suppressed under the combined inhibition effect of NaCl (7.9 ± 0.2 g/L, as introduced by the FO direct filtration) and FNA (0.11 ± 0.02 mg of HNO2–N/L, formed as a result of the increased NH4+–N concentration after the FO process). The two inhibition factors worked in synergy to achieve a more stable PN operation. The microbial analysis showed that the elevated salinity and accumulation of FNA reshaped the microbial community and selectively eliminated NOB. Finally, an economic and feasibility analysis was conducted, which suggests that the integration of an FO unit into PN/A is a feasible and economically viable wastewater treatment process.

Anaerobic ammonium oxidation (anammox) was discovered as a new microbial reaction in the late 1990s, which led to the development of an innovative energy- and carbon-efficient technology─partial nitritation and anammox (PN/A)─for nitrogen removal. PN/A was first applied to remove the nitrogen from high-strength wastewaters, e.g., anaerobic digestion liquor (i.e., sidestream), and further expanded to the main line of wastewater treatment plants (i.e., mainstream). While sidestream PN/A has been well-established with extensive full-scale installations worldwide, practical application of PN/A in mainstream treatment has been proven extremely challenging to date. A key challenge is achieving stable suppression of nitrite-oxidizing bacteria (NOB). This study examines the progress of NOB suppression in both sidestream- and mainstream PN/A over the past two decades. The successful NOB suppression in sidestream PN/A was reviewed, and these successes were evaluated in terms of their transferability into mainstream PN/A. Drawing on the learning over the past decades, we anticipate that a hybrid process, comprised of biofilm and floccular sludge, bears great potential to achieve efficient mainstream PN/A, while a combination of strategies is entailed for stable NOB suppression. Furthermore, the recent discovery of novel nitrifiers would trigger new opportunities and new challenges for mainstream PN/A.

Mathematical modeling plays a critical role toward the mitigation of nitrous oxide (N2O) emissions from wastewater treatment plants (WWTPs). In this work, we proposed a novel hybrid modeling approach by integrating the first principal model with deep learning techniques to predict N2O emissions. The hybrid model was successfully implemented and validated with the N2O emission data from a full-scale WWTP. This hybrid model is demonstrated to have higher accuracy for N2O emission modeling in the WWTP than the mechanistic model or pure deep learning model. Equally important, the hybrid model is more applicable than the pure deep learning model due to the lower requirement of data and the pure mechanistic model due to the less calibration requirement. This superior performance was due to the hybrid nature of the proposed model. It integrated the essential wastewater treatment knowledge as the first principal component and the less understood N2O production processes by the data-driven deep learning approach. The developed hybrid model was also successfully implemented under different circumstances for the prediction of N2O flux, which showed the generalizability of the model. The hybrid model also showed great potential to be applied for the N2O mitigation work. Nevertheless, the capability of the hybrid model in evaluating N2O mitigation strategies still requires validation with experiments. Going beyond N2O modeling in WWTP, the novel hybridization modeling concept can potentially be applied to other environmental systems.

Convolutional neural networks (CNNs), such as the time-delay neural network (TDNN), have shown their remarkable capability in learning speaker embedding. However, they meanwhile bring a huge computational cost in storage size, processing, and memory. Discovering the specialized CNN that meets a specific constraint requires a substantial effort of human experts. Compared with hand-designed approaches, neural architecture search (NAS) appears as a practical technique in automating the manual architecture design process and has attracted increasing interest in spoken language processing tasks such as speaker recognition. In this paper, we propose EfficientTDNN, an efficient architecture search framework consisting of a TDNN-based supernet and a TDNN-NAS algorithm. The proposed supernet introduces temporal convolution of different ranges of the receptive field and feature aggregation of various resolutions from different layers to TDNN. On top of it, the TDNN-NAS algorithm quickly searches for the desired TDNN architecture via weight-sharing subnets, which surprisingly reduces computation while handling the vast number of devices with various resources requirements. Experimental results on the VoxCeleb dataset show the proposed EfficientTDNN enables approximate $10^{13}$ architectures concerning depth, kernel, and width. Considering different computation constraints, it achieves a 2.20% equal error rate (EER) with 204 M multiply-accumulate operations (MACs), 1.41% EER with 571 M MACs as well as 0.94% EER with 1.45 G MACs. Comprehensive investigations suggest that the trained supernet generalizes subnets not sampled during training and obtains a favorable trade-off between accuracy and efficiency.

Nitrous oxide (N2O) emissions account for the majority of the carbon footprint of wastewater treatment plants (WWTPs). Many N2O mitigation strategies have since been developed while a holistic view is still missing. This article reviews the state-of-the-art of N2O mitigation studies in wastewater treatment. Through analyzing existing studies, this article presents the essential knowledge to guide N2O mitigations, and the logics behind mitigation strategies. In practice, mitigations are mainly carried out by aeration control, feed scheme optimization, and process optimization. Despite increasingly more studies, real implementation remains rare, which is a combined result of unclear climate change policies/incentives, as well as technical challenges. Five critical technical challenges, as well as opportunities, of N2O mitigations were identified. It is proposed that (i) quantification methods for overall N2O emissions and pathway contributions need improvement; (ii) a reliable while straightforward mathematical model is required to quantify benefits and compare mitigation strategies; (iii) tailored risk assessment needs to be conducted for WWTPs, in which more long-term full-scale trials of N2O mitigation are urgently needed to enable robust assessments of the resulting operational costs and impact on nutrient removal performance; (iv) current mitigation strategies focus on centralized WWTPs, more investigations are warranted for decentralised systems, especially decentralized activated sludge WWTPs; and (v) N2O may be mitigated by adopting novel strategies promoting N2O reduction denitrification or microorganisms that emit less N2O. Overall, we conclude N2O mitigation research is reaching a maturity while challenges still exist for a wider implementation, especially in relation to the reliability of N2O mitigation strategies and potential risks to nutrient removal performances of WWTPs.

Aerobic sludge digestion of waste activated sludge (WAS) is widely used as a stabilization option in small- and midsized wastewater treatment plants. However, the digestion process is often limited by low volatile solids (VS) destruction and poor pathogen removal efficiency. This study presents a novel operational strategy that achieves enhanced VS destruction and nitrogen removal by inducing sustained nitrite accumulation via a single spike of nitrite to aerobic digester operated at a natively low pH (<5.5). The strategy was demonstrated through the use of three laboratory aerobic sludge digesters, each continuously operated for over 300 days. Compared to control reactors, the strategy enhanced volatile solids destruction by 35.0–38.4%, nitrogen removal by 58.5–70.8%, and pathogen reduction by approximately 1 log. The standard oxygen uptake rate (SOUR) was reduced to 0.49 ± 0.03 mgO2/gVS/h, compared to 0.85 ± 0.01–1.68 ± 0.02 mgO2/gVS/h in the control, indicating enhanced stabilization. Free nitrous acid formed from nitrite at low pH, rather than nitrite itself, was identified to be the cause of improved digestion performance. Since the nitrite production is self-supporting, no additional ongoing costs are incurred.

Stable suppression of nitrite oxidizing bacteria (NOB) is one of the major bottlenecks for achieving mainstream nitrite shunt or partial nitritation/anammox (PN/A). It is increasingly experienced that NOB could develop resistance to suppressions over an extended time, leading to failure of nitrite shunt or PN/A. This study reports and demonstrates the first effective strategy to overcome NOB adaptation through alternating sludge treatment with free nitrous acid (FNA) and free ammonia (FA). During over 650 days of reactor operation, NOB adaptation to both FNA and FA was observed, but the adaptation was successfully overcome by deploying the alternate treatment strategy. Microbial community analysis showed Nitrospiraand Nitrobacter, the key NOB populations in the reactor, have the ability to adapt to FNA and FA, respectively, but do not adapt to the alternation. Stable nitrite shunt with nitrite accumulation ratio over 95% and excellent nitrogen removal were maintained for the last 10 months with only one alternation applied. N2O emission increased initially as the attainment of nitrite shunt but exhibited a declining trend during the study. By using on-site-produced nitrite and ammonium, the proposed strategy is feasible and sustainable. This study brings the mainstream nitrite shunt and PN/A one step closer to wide applications.

This study aims to demonstrate a new technology roadmap to support the ongoing paradigm shift in wastewater management from pollutant removal to resource recovery. This is achieved by developing a novel use of an iron salt (i.e., FeCl3) in an integrated anaerobic wastewater treatment and mainstream anammox process. FeCl3was chosen to be dosed in a proposed sidestream unit rather than in a primary settler or a mainstream reactor. This causes acidification of returned activated sludge and enables stable suppression of nitrite-oxidizing bacterial activity and excess sludge reduction. A laboratory-scale system, which comprised an anaerobic baffled reactor, a continuous-flow anoxic–aerobic (A/O) reactor, and a secondary settler, was designed to treat real domestic wastewater, with the performance of the system comprehensively monitored under a steady-state condition. The experimental assessments showed that the system had good effluent quality, with total nitrogen and phosphorus concentrations of 12.6 ± 1.3 mg N/L and 0.34 ± 0.05 mg P/L, respectively. It efficiently retained phosphorus in excess sludge (0.18 ± 0.03 g P/g dry sludge), suggesting its potential for further recovery. About half of influent organic carbon was recovered in the form of bioenergy (i.e., methane). This together with low energy consumption revealed that the system could produce a net energy of about 0.11 kWh/m3-wastewater, assessed by an energy balance analysis.

Herein we report a self-cleaning coating derived from zwitterionic poly(2-methacryloyloxylethyl phosphorylcholine) (PMPC) brushes grafted on a solid substrate. The PMPC surface not only exhibits complete oil repellency in a water-wetted state (i.e., underwater superoleophobicity), but also allows effective cleaning of oil fouled on dry surfaces by water alone. The PMPC surface was compared with typical underwater superoleophobic surfaces realized with the aid of surface roughening by applying hydrophilic nanostructures and those realized by applying smooth hydrophilic polyelectrolyte multilayers. We show that underwater superoleophobicity of a surface is not sufficient to enable water to clean up oil fouling on a dry surface, because the latter circumstance demands the surface to be able to strongly bond water not only in its pristine state but also in an oil-wetted state. The PMPC surface is unique with its described self-cleaning performance because the zwitterionic phosphorylcholine groups exhibit exceptional binding affinity to water even when they are already wetted by oil. Further, we show that applying this PMPC coating onto steel meshes produces oil–water separation membranes that are resilient to oil contamination with simply water rinsing. Consequently, we provide an effective solution to the oil contamination issue on the oil–water separation membranes, which is an imperative challenge in this field. Thanks to the self-cleaning effect of the PMPC surface, PMPC-coated steel meshes can not only separate oil from oil–water mixtures in a water-wetted state, but also can lift oil out from oil–water mixtures even in a dry state, which is a very promising technology for practical oil-spill remediation. In contrast, we show that oil contamination on conventional hydrophilic oil–water separation membranes would permanently induce the loss of oil–water separation function, and thus they have to be always used in a completely water-wetted state, which significantly restricts their application in practice.

In-sewer stability of biomarkers is a critical factor for wastewater-based epidemiology, as it could affect the accuracy of the estimated prevalence of substances in the community. The spatiotemporal variations of environmental and biological conditions in sewers can influence the transformation of biomarkers. To date, the relationship between environmental variables and biomarker stability in sewers is poorly understood. Therefore, this study evaluated the transformation of common illicit drug and pharmaceutical biomarkers in laboratory sewer reactors with different levels of pH, temperature, and suspended solids. The correlations between degradation rates of 14 biomarkers, 3 controlled environmental variables (pH, temperature, and suspended solids concentration), and 3 biological activity indicators (sulfide production rate, methane production rate, and the removal rate of soluble chemical oxygen demand (SCOD)) were assessed using correlation matrix, stepwise regression method, and principal component analysis. The consistent results affirmed the dominant effects of biological activities and pH on biomarker transformation in sewers, particularly for labile compounds, whereas the impact of temperature or suspended solids was less significant. This study enhances the understanding of factors affecting the fate of micropollutants in sewer systems and facilitates the interpretation of WBE results for assessing drug use and public health in communities.