YOLOv8-C2f-Faster-EMA: An Improved Underwater Trash Detection Model Based on YOLOv8.

Anthropogenic waste deposition in aquatic environments precipitates a decline in water quality, engendering pollution that adversely impacts human health, ecological integrity, and economic endeavors. The evolution of underwater robotic technologies heralds a new era in the timely identification and extraction of submerged litter, offering a proactive measure against the scourge of water pollution. This study introduces a refined YOLOv8-based algorithm tailored for the enhanced detection of small-scale underwater debris, aiming to mitigate the prevalent challenges of high miss and false detection rates in aquatic settings. The research presents the YOLOv8-C2f-Faster-EMA algorithm, which optimizes the backbone, neck layer, and C2f module for underwater characteristics and incorporates an effective attention mechanism. This algorithm improves the accuracy of underwater litter detection while simplifying the computational model. Empirical evidence underscores the superiority of this method over the conventional YOLOv8n framework, manifesting in a significant uplift in detection performance. Notably, the proposed method realized a 6.7% increase in precision (P), a 4.1% surge in recall (R), and a 5% enhancement in mean average precision (mAP). Transcending its foundational utility in marine conservation, this methodology harbors potential for subsequent integration into remote sensing ventures. Such an adaptation could substantially enhance the precision of detection models, particularly in the realm of localized surveillance, thereby broadening the scope of its applicability and impact.

Hash Encoding and Brightness Correction in 3D Industrial and Environmental Reconstruction of Tidal Flat Neural Radiation.

We present an innovative approach to mitigating brightness variations in the unmanned aerial vehicle (UAV)-based 3D reconstruction of tidal flat environments, emphasizing industrial applications. Our work focuses on enhancing the accuracy and efficiency of neural radiance fields (NeRF) for 3D scene synthesis. We introduce a novel luminance correction technique to address challenging illumination conditions, employing a convolutional neural network (CNN) for image enhancement in cases of overexposure and underexposure. Additionally, we propose a hash encoding method to optimize the spatial position encoding efficiency of NeRF. The efficacy of our method is validated using diverse datasets, including a custom tidal flat dataset and the Mip-NeRF 360 dataset, demonstrating superior performance across various lighting scenarios.

Mutation detection dual correlation filter with an object-awareness module for real-time target tracking.

Discriminative correlation filter (DCF) based methods have recently been widely used for visual tracking tasks. The adaptive spatiotemporal-regulation based tracker (AutoTrack) can only partially solve some limitations of the DCF framework including filter degradation and the boundary effect, but its application scenarios need to be broadened, and performance improvements are also required. To further surmount these difficulties, this paper provides an object-awareness-module based mutation detection dual correlation filter (MDDCF-OAM). The main innovation points of this work are: (1) an object-mask based context enhancer is proposed to formulate a more robust appearance model; (2) a dual filter training-learning structure is adopted to allow the dual filters to restrict each other and suppress the filter degradation effect; (3) a Gaussian label map is updated with the refined joint response map to detect and attenuate the response mutation effects. Exhaustive experiments have been conducted to test the efficiency of the suggested MDDCF-OAM on four benchmarks, namely, OTB2015, UAV123, TC128, and VOT2019. The results indicate that: (1) the introduced MDDCF-OAM surpasses nine state-of-the-art trackers; (2) the MDDCF-OAM has a real-time speed of 32 frames per second, which is sufficient for target tracking tasks in numerous scenarios, especially unmanned aerial vehicles and camera tracking.

Single-Stage Underwater Target Detection Based on Feature Anchor Frame Double Optimization Network.

OBJECTIVE: The shallow underwater environment is complex, with problems of color shift, uneven illumination, blurring, and distortion in the imaging process. These scenes are very unfavorable for the reasoning of the detection network. Additionally, typical object identification algorithms struggle to maintain high resilience in underwater environments due to picture domain offset, making underwater object detection problematic. METHODS: This paper proposes a single-stage detection method with the double enhancement of anchor boxes and features. The feature context relevance is improved by proposing a composite-connected backbone network. The receptive field enhancement module is introduced to enhance the multi-scale detection capability. Finally, a prediction refinement strategy is proposed, which refines the anchor frame and features through two regressions, solves the problem of feature anchor frame misalignment, and improves the detection performance of the single-stage underwater algorithm. RESULTS: We achieved an effect of 80.2 mAP on the Labeled Fish in the Wild dataset, which saves some computational resources and time while still improving accuracy. On the original basis, UWNet can achieve 2.1 AP accuracy improvement due to the powerful feature extraction function and the critical role of multi-scale functional modules. At an input resolution of 300 × 300, UWNet can provide an accuracy of 32.4 AP. When choosing the number of prediction layers, the accuracy of the four and six prediction layer structures is compared. The experiments show that on the Labeled Fish in the Wild dataset, the six prediction layers are better than the four. CONCLUSION: The single-stage underwater detection model UWNet proposed in this research has a double anchor frame and feature optimization. By adding three functional modules, the underwater detection of the single-stage detector is enhanced to address the issue that it is simple to miss detection while detecting small underwater targets.

Super-resolution reconstruction of biometric features recognition based on manifold learning and deep residual network.

BACKGROUND AND OBJECTIVE: In daily life, face information has the characteristics of uniqueness and universality. However, in a real-world scene, the image information of the face acquired by the acquisition device often contains noises such as blurring and sharpening. As such, super-resolution reconstruction of face features recognition based on manifold learning is proposed in this paper. METHODS: We reconstruct low-resolution facial expression images, introduce a simplified residual block network and manifold learning, and propose joint supervision through a new hybrid loss function, which not only retains the color and characteristics of the image, but also retains the high-frequency information. The ResNet50 network uses the weight feature of information entropy to optimize the information of the pooling layer, and the esNet50 network uses the improved PSO algorithm to optimize the initial weight of the error back-propagation phase. RESULTS: In the case of inputting extremely low resolution (6 × 6) facial expression images, the accuracy rate is increased by 9.091%. The accuracy of the high-resolution facial expressions after reconstruction with a size of 12×12 is 96.970%. The accuracy rate for happy expressions is 100%, the accuracy rate for anger, disgust, sadness, and surprise recognition is 97%, the accuracy rate for contempt is 94%, and the accuracy rate for fear is 88%. CONCLUSIONS: The experimental results verify the feasibility and superiority of the system, and effectively improve the accuracy of low-resolution facial expressions.

Facial expression recognition based on deep learning.

BACKGROUND AND OBJECTIVE: Facial expression recognition technology will play an increasingly important role in our daily life. Autonomous driving, virtual reality and all kinds of robots integrated into our life depend on the development of facial expression recognition technology. Many tasks in the field of computer vision are based on deep learning technology and convolutional neural network. The paper proposes an occluded expression recognition model based on the generated countermeasure network. The model is divided into two modules, namely, occluded face image restoration and face recognition. METHODS: Firstly, this paper summarizes the research status of deep facial expression recognition methods in recent ten years and the development of related facial expression database. Then, the current facial expression recognition methods based on deep learning are divided into two categories: Static facial expression recognition and dynamic facial expression recognition. The two methodswill be introduced and summarized respectively. Aiming at the advanced deep expression recognition algorithms in the field, the performance of these algorithms on common expression databases is compared, and the strengths and weaknesses of these algorithms are analyzed in detail. DISCUSSION AND RESULTS: As the task of facial expression recognition is gradually transferred from the controlled laboratory environment to the challenging real-world environment, with the rapid development of deep learning technology, deep neural network can learn discriminative features, and is gradually applied to automatic facial expression recognition task. The current deep facial expression recognition system is committed to solve the following two problems: (1) Overfitting due to lack of sufficient training data; (2) In the real world environment, other variables that have nothing to do with expression bring interference problems. CONCLUSION: From the perspective of algorithm, combining other expression models, such as facial action unit model and pleasure arousal dimension model, as well as other multimodal models, such as audio mode, 3D face depth information and human physiological information, can make expression recognition more practical.

Effective Face Detector Based on YOLOv5 and Superresolution Reconstruction.

The application of face detection and recognition technology in security monitoring systems has made a huge contribution to public security. Face detection is an essential first step in many face analysis systems. In complex scenes, the accuracy of face detection would be limited because of the missing and false detection of small faces, due to image quality, face scale, light, and other factors. In this paper, a two-level face detection model called SR-YOLOv5 is proposed to address some problems of dense small faces in actual scenarios. The research first optimized the backbone and loss function of YOLOv5, which is aimed at achieving better performance in terms of mean average precision (mAP) and speed. Then, to improve face detection in blurred scenes or low-resolution situations, we integrated image superresolution technology on the detection head. In addition, some representative deep-learning algorithm based on face detection is discussed by grouping them into a few major categories, and the popular face detection benchmarks are enumerated in detail. Finally, the wider face dataset is used to train and test the SR-YOLOv5 model. Compared with multitask convolutional neural network (MTCNN), Contextual Multi-Scale Region-based CNN (CMS-RCNN), Finding Tiny Faces (HR), Single Shot Scale-invariant Face Detector (S3FD), and TinaFace algorithms, it is verified that the proposed model has higher detection precision, which is 0.7%, 0.6%, and 2.9% higher than the top one. SR-YOLOv5 can effectively use face information to accurately detect hard-to-detect face targets in complex scenes.

Robust face recognition based on multi-task convolutional neural network.

PURPOSE: Due to the lack of prior knowledge of face images, large illumination changes, and complex backgrounds, the accuracy of face recognition is low. To address this issue, we propose a face detection and recognition algorithm based on multi-task convolutional neural network (MTCNN). METHODS: In our paper, MTCNN mainly uses three cascaded networks, and adopts the idea of candidate box plus classifier to perform fast and efficient face recognition. The model is trained on a database of 50 faces we have collected, and Peak Signal-to-Noise Ratio (PSNR), Structural Similarity Index Measurement (SSIM), and receiver operating characteristic (ROC) curve are used to analyse MTCNN, Region-CNN (R-CNN) and Faster R-CNN. RESULTS: The average PSNR of this technique is 1.24 dB higher than that of R-CNN and 0.94 dB higher than that of Faster R-CNN. The average SSIM value of MTCNN is 10.3% higher than R-CNN and 8.7% higher than Faster R-CNN. The Area Under Curve (AUC) of MTCNN is 97.56%, the AUC of R-CNN is 91.24%, and the AUC of Faster R-CNN is 92.01%. MTCNN has the best comprehensive performance in face recognition. For the face images with defective features, MTCNN still has the best effect. CONCLUSIONS: This algorithm can effectively improve face recognition to a certain extent. The accuracy rate and the reduction of the false detection rate of face detection can not only be better used in key places, ensure the safety of property and security of the people, improve safety, but also better reduce the waste of human resources and improve efficiency.

Miniaturized dielectric barrier discharge-molecular emission spectrometer for determination of total sulfur dioxide in food.

SO2 is a type of additive widely used in the food processing industry as preservative and anti-browning, bleaching, or effective antibacterial agent. However, the SO2 residue can have adverse effects on human health. In this work, a low-temperature microplasma generated by dielectric barrier discharge was used for the direct, in situ excitation of the molecular emission of SO2 in food samples. The food samples were acidified and heated to release SO2 and a miniaturized charge-coupled device spectrometer recorded the characteristic emission line at 301.9 nm. The linear correlation coefficient of the method was greater than 0.99 in the range of 10 to 100 mg L-1. Moreover, the limit of detection was 0.01 mg L-1, with recoveries between 72% and 108% and relative standard deviations of 1.5%-7.6%. The method is simple, accurate, low-cost, involves miniaturized and compact equipment and is suitable for the determination of total SO2 in food samples.

Novel Segmented Concentration Addition Method to Predict Mixture Hormesis of Chlortetracycline Hydrochloride and Oxytetracycline Hydrochloride to Aliivibrio fischeri.

Hormesis is a concentration-response phenomenon characterized by low-concentration stimulation and high-concentration inhibition, which typically has a nonmonotonic J-shaped concentration-response curve (J-CRC). The concentration addition (CA) model is the gold standard for studying mixture toxicity. However, the CA model had the predictive blind zone (PBZ) for mixture J-CRC. To solve the PBZ problem, we proposed a segmented concentration addition (SCA) method to predict mixture J-CRC, which was achieved through fitting the left and right segments of component J-CRC and performing CA prediction subsequently. We selected two model compounds including chlortetracycline hydrochloride (CTCC) and oxytetracycline hydrochloride (OTCC), both of which presented J-CRC to Aliivibrio fischeri (AVF). The seven binary mixtures (M1-M7) of CTCC and OTCC were designed according to their molar ratios of 12:1, 10:3, 8:5, 1:1, 5:8, 3:10, and 1:12 referring to the direct equipartition ray design. These seven mixtures all presented J-CRC to AVF. Based on the SCA method, we obtained mixture maximum stimulatory effect concentration (ECm) and maximum stimulatory effect (Em) predicted by SCA, both of which were not available for the CA model. The toxicity interactions of these mixtures were systematically evaluated by using a comprehensive approach, including the co-toxicity coefficient integrated with confidence interval method (CTCICI), CRC, and isobole analysis. The results showed that the interaction types were additive and antagonistic action, without synergistic action. In addition, we proposed the cross point (CP) hypothesis for toxic interactive mixtures presenting J-CRC, that there was generally a CP between mixture observed J-CRC and CA predicted J-CRC; the relative positions of observed and predicted CRCs on either side of the CP would exchange, but the toxic interaction type of mixtures remained unchanged. The CP hypothesis needs to be verified by more mixtures, especially those with synergism. In conclusion, the SCA method is expected to have important theoretical and practical significance for mixture hormesis.

Efficient extraction of uranium in organics-containing wastewater over g-C3N4/GO hybrid nanosheets with type-II band structure.

The presence of organics in radioactive wastewater decreased the extraction efficiency in traditional adsorption treatment of nuclide. Herein, we developed g-C3N4/graphene oxide hybrid nanosheets (g-C3N4/GO) with the typical type-Ⅱ band structure for the phtodegradation-extraction treatment of hexavalent uranium (U(VI)) in the tannic acid (TA)-containing wastewater. Due to the staggered band structure, the photoelectron transferred from g-C3N4 to GO under simulated sunlight. Accordingly, g-C3N4/GO hybrid nanosheets exhibited promoted TA degradation and U(VI) extraction compared with the pristine g-C3N4 and GO nanosheets. In light condition, the g-C3N4/GO hybrid nanosheets degraded 92% of the TA and a 352 mg/g of U(VI) extraction capacity in 40 min. Notably, as an adverse factor for the U(VI) extraction in dark condition, the existence of TA increased the extraction capacity under simulated sunlight, which resulted from the consuming of photoinduced holes and promotion of U(VI) reduction.

New Insight Regarding the Relationship Between Enantioselective Toxicity Difference and Enantiomeric Toxicity Interaction from Chiral Ionic Liquids.

Chirality is an important property of molecules. The study of biological activity and toxicity of chiral molecules has important theoretical and practical significance for toxicology, pharmacology, and environmental science. The toxicological significance of chiral ionic liquids (ILs) has not been well revealed. In the present study, the enantiomeric joint toxicities of four pairs of chiral ILs 1-alkyl-3-methylimidazolium lactate to Allivibrio fischeri were systematically investigated by using a comprehensive approach including the co-toxicity coefficient (CTC) integrated with confidence interval (CI) method (CTCICI), concentration-response curve (CRC), and isobole analysis. The direct equipartition ray (EquRay) design was used to design five binary mixtures of enantiomers according to molar ratios of 1:5, 2:4, 3:3, 4:2, and 5:1. The toxicities of chiral ILs and their mixtures were determined using the microplate toxicity analysis (MTA) method. Concentration addition (CA) and independent action (IA) were used as the additive reference models to construct the predicted CRC and isobole of mixtures. On the whole, there was an enantioselective toxicity difference between [BMIM]D-Lac and [BMIM]L-Lac, and [HMIM]D-Lac and [HMIM]L-Lac, while no enantioselective toxicity difference was observed for [EMIM]D-Lac and [EMIM]L-Lac, and [OMIM]D-Lac and [OMIM]L-Lac. Thereinto, the enantiomer mixtures of [BMIM]D-Lac and [BMIM]L-Lac, and [HMIM]D-Lac and [HMIM]L-Lac presented antagonistic action, and the enantiomer mixtures of [EMIM]D-Lac and [EMIM]L-Lac, and [OMIM]D-Lac and [OMIM]L-Lac overall presented additive action. Moreover, the greatest antagonistic toxicity interaction occurred at the equimolar ratio of enantiomers. Based on these results, we proposed two hypotheses, (1) chiral molecules with enantioselective toxicity difference tended to produce toxicity interactions, (2) the highest or lowest toxicity was usually at the equimolar ratio and its adjacent ratio for the enantiomer mixture. These hypotheses will need to be further validated by other enantiomer mixtures.

Integrative Assessment of Mixture Toxicity of Three Ionic Liquids on Acetylcholinesterase Using a Progressive Approach from 1D Point, 2D Curve, to 3D Surface.

The joint toxicities of [BMIM]BF4, [BMIM]PF6, and [HMIM]BF4 on acetylcholinesterase (AChE) were systematically investigated by using a progressive approach from 1D single effect point, 2D concentration-response curve (CRC), to 3D equivalent-surface (ES) level. The equipartition equivalent-surface design (EESD) method was used to design 10 ternary mixtures, and the direct equipartition ray (EquRay) design was used to design 15 binary mixtures. The toxicities of ionic liquids (ILs) and their mixtures were determined using the microplate toxicity analysis (MTA) method. The concentration addition (CA), independent action (IA), and co-toxicity coefficient (CTC) were used as the additive reference model to analyze the toxic interaction of these mixtures. The results showed that the Weibull function fitted well the CRCs of the three ILs and their mixtures with the coefficient of determination (R2) greater than 0.99 and root-mean-square error (RMSE) less than 0.04. According to the CTC integrated with confidence interval (CI) method (CTCICI) developed in this study, the 25 mixtures were almost all additive action at 20% and 80% effect point levels. At 50% effect, at least half of the 25 mixtures were slightly synergistic action, and the remaining mixtures were additive action. Furthermore, the ESs and CRCs predicted by CA and IA were all within the CIs of mixture observed ESs and CRCs, respectively. Therefore, the toxic interactions of these 25 mixtures were actually additive action. The joint toxicity of the three ILs can be effectively evaluated by the ES method. We also studied the relationship between the mixture toxicities and component concentration proportions. This study can provide reference data for IL risk assessment of combined pollution.

Fluorometric aptasensor for cadmium(II) by using an aptamer-imprinted polymer as the recognition element.

A fluorometric assay is described for the determination of Cd(II) in environmental and agricultural samples. It is making use of a molecularly imprinted polymer (MIP) and aptamer as dual recognition units, while carbon quantum dots (co-doped with sulphur and nitrogen) and gold nanoparticles (SN-CQD/Au) act as the fluorophores. The aptamer-modified MIP was placed on an SN-CQD/Au-modified indium tin oxide glass electrode. Cd(II) was detected with high selectivity by the recognition sites of the aptamer in the MIP. Fluorescence, with excitation/emission peaks at 370/430 nm, is quenched by Cd(II). Response is linear in the 20 pM to 12 nM concentration range. The detection limit is 1.2 pM. The sensor is selective for Cd(II), and recoveries from spiked waters, soils and vegetables real-world samples range between 82.1 and 113.9%. Graphical abstractA fluorescence sensor composed of a molecularly imprinted polymer and an aptamer as a dual identification system for Cd2+ coupled with and carbon quantum dots (co-doped with sulphur and nitrogen) and gold nanoparticles (SN-CQDs/Au) as fluorescent element that can detect Cd2+ with high selectivity by a dual-recognition mechanism.

Biomass-derived composite aerogels with novel structure for removal/recovery of uranium from simulated radioactive wastewater.

With the development of nuclear energy, the removal/recovery of radionuclides has attracted increasing attention. Here, an ultra-light, super-elastic, konjac glucomannan/graphene oxide composite aerogel (KGCA) as a high performance adsorbent for radionuclide removal/recovery was fabricated by a three-step process of freeze-casting, freeze-drying, and carbonization. The as-prepared bionic structured KGCA showed ultralow density, high specific surface area, desirable super-elasticity, and abundant oxygen-containing functional groups. Batch adsorption results demonstrated the maximum adsorption capacity of uranium (U(VI)) on KGCA is as high as 513.4 mg g-1, far exceeding other biomass carbon aerogels. Furthermore, KGCA showed good radiation stability, selective adsorption of U(VI), and high recycling performance. The KGCA also showed good adsorption properties even under simulated seawater or high salt concentration. Thus, these ultra-light and super-elastic biomass-derived composite aerogels could have a wide range of applications for nuclear wastewater treatment in the future.

A liquid chromatography detector based on continuous-flow chemical vapor generation coupled glow discharge atomic emission spectrometry: Determination of organotin compounds in food samples.

A novel liquid chromatography (LC) detector based on continuous-flow chemical vapor generation (CVG)-coupled glow discharge (GD) atomic emission spectrometry (AES) was proposed herein for the simultaneous determination of organotins (OTs) in food samples. OTs were separated on an LC column and converted into volatile gases by the CVG technique. The GD microplasma was used to excite the Sn atoms to generate the special atomic emission lines of Sn at 317.66 nm, which were recorded by a charge-coupled device (CCD) spectrometer. For an injection volume of 1 mL, the linear correlation coefficients (R) were higher than 0.99 in the concentration range from 0.1 to 10 µg mL-1. The recoveries of OTs from spiked samples were 70-103%, and the relative standard deviations (RSD) were 0.2-8.7%. The limits of detection (LOD) for the tested OTs, i.e. trimethyltin chloride (TMT) and dimethyltin dichloride (DMTC), were determined to be 0.59 and 0.93 µg L-1, respectively. The CVG-GD-AES detector can be used to simultaneously determine TMT and DMTC with high sensitivity, in a simple and cost-effective manner.

Nano-Montmorillonite Regulated Crystallization of Hierarchical Strontium Carbonate in a Microbial Mineralization System.

In this paper, nano-montmorillonite (nano-MMT) was introduced into the microbial mineralization system of strontium carbonate (SrCO3). By changing the nano-MMT concentration and the mineralization time, the mechanism of mineralization was studied. SrCO3 superstructures with complex forms were acquired in the presence of nano-MMT as a crystal growth regulator. At low concentrations of nano-MMT, a cross-shaped SrCO3 superstructure was obtained. As the concentration increased, flower-like SrCO3 crystals formed via the dissolution and recrystallization processes. An emerging self-assembly process and crystal polymerization mechanism have been proposed by forming complex flower-like SrCO3 superstructures in high concentrations of nano-MMT. The above research indicated that unique bionic synthesis strategies in microbial systems could not only provide a useful route for the production of inorganic or inorganic/organic composites with a novel morphology and unique structure but also provide new ideas for the treatment of radionuclides.

Compartmentalizing Incompatible Tandem Reactions in Pickering Emulsions To Enable Visual Colorimetric Detection of Nitramine Explosives Using a Smartphone.

We report a visual colorimetric assay for detection of nitramine explosives such as 1,3,5-trinitro-1,3,5-triazinane (RDX) and 1,3,5,7-tetranitro-1,3,5,7-tetrazocane (HMX) using a smartphone. This assay is based on compartmentalizing incompatible tandem reactions in Pickering emulsions. The alkaline hydrolysis of RDX or HMX in one Pickering emulsion produces nitrite ions, which autodiffuse into the other Pickering emulsion to form nitrous acid. It oxidizes the 3,3',5,5'-tetramethylbenzidine (TMB) to generate yellow TMB diimine. The RGB component change of the optical images is applied to quantitatively determine the RDX and HMX at different reaction temperatures. A distinct color change occurs at RDX and HMX concentrations of 1.2 and 12 µM, respectively. The adjusted intensity increases linearly with the increase of the logarithms of the concentrations of RDX and HMX in the range of 1.2-90 µM and 12-90 µM, respectively. The limits of detection of RDX and HMX are 96 and 110 nM, respectively. Importantly, this assay is employed for the detection of RDX and HMX in real water, proving the applicability of the assay in real-world samples.

Mixture cytotoxicity assessment of ionic liquids and heavy metals in MCF-7 cells using mixtox.

Ionic liquids (ILs) are widely used as extractants for heavy metals. However, the effect of mixtures of ILs and heavy metals is rarely understood. In this study, we tested the cytotoxicity of four ILs, four heavy metals and their mixtures on human MCF-7 cells in 96-well microplates. The toxicity of single compounds in MCF-7 cells ranges from 3.07 × 10(-6) M for Cu(II) to 2.20 × 10(-3) M for 1-ethyl-3-methylimidazolium tetrafluoroborate. The toxicity of heavy metals in MCF-7 is generally higher than the toxicity of ILs. A uniform experimental design was used to simulate environmentally realistic mixtures. Two classical reference models (concentration addition and independent action) were used to predict their mixture. The experiments to evaluate the toxicity of the mixture revealed antagonism among four ILs and four heavy metals in MCF-7 cells. Pearson correlation analysis showed that Ni(II) and 1-dodecyl-3-methylimidazolium chloride are positively correlated with the extent of antagonism, while 1-hexyl-3-methylimidazolium tetrafluoroborate showed a negative correlation. Data analysis was conducted in the R package mixtox, which integrates features such as curve fitting, experimental design, and mixture toxicity prediction. The international community of toxicologists is welcome to use this package and provide feedback as suggestions and comments.

Recursive Random Forests Enable Better Predictive Performance and Model Interpretation than Variable Selection by LASSO.

Variable selection is of crucial significance in QSAR modeling since it increases the model predictive ability and reduces noise. The selection of the right variables is far more complicated than the development of predictive models. In this study, eight continuous and categorical data sets were employed to explore the applicability of two distinct variable selection methods random forests (RF) and least absolute shrinkage and selection operator (LASSO). Variable selection was performed: (1) by using recursive random forests to rule out a quarter of the least important descriptors at each iteration and (2) by using LASSO modeling with 10-fold inner cross-validation to tune its penalty λ for each data set. Along with regular statistical parameters of model performance, we proposed the highest pairwise correlation rate, average pairwise Pearson's correlation coefficient, and Tanimoto coefficient to evaluate the optimal by RF and LASSO in an extensive way. Results showed that variable selection could allow a tremendous reduction of noisy descriptors (at most 96% with RF method in this study) and apparently enhance model's predictive performance as well. Furthermore, random forests showed property of gathering important predictors without restricting their pairwise correlation, which is contrary to LASSO. The mutual exclusion of highly correlated variables in LASSO modeling tends to skip important variables that are highly related to response endpoints and thus undermine the model's predictive performance. The optimal variables selected by RF share low similarity with those by LASSO (e.g., the Tanimoto coefficients were smaller than 0.20 in seven out of eight data sets). We found that the differences between RF and LASSO predictive performances mainly resulted from the variables selected by different strategies rather than the learning algorithms. Our study showed that the right selection of variables is more important than the learning algorithm for modeling. We hope that a standard procedure could be developed based on these proposed statistical metrics to select the truly important variables for model interpretation, as well as for further use to facilitate drug discovery and environmental toxicity assessment.