Recently, recommendation based on causal inference has gained much attention in the industrial community. The introduction of causal techniques into recommender systems (RS) has brought great development to this field and has gradually become a trend. However, a unified causal analysis framework has not been established yet. On one hand, the existing causal methods in RS lack a clear causal and mathematical formalization on the scientific questions of interest. Many confusions need to be clarified: what exactly is being estimated, for what purpose, in which scenario, by which technique, and under what plausible assumptions. On the other hand, technically speaking, the existence of various biases is the main obstacle to drawing causal conclusions from observed data. Yet, formal definitions of the biases in RS are still not clear. Both of the limitations greatly hinder the development of RS.In this paper, we attempt to give a causal analysis framework to accommodate different scenarios in RS, thereby providing a principled and rigorous operational guideline for causal recommendation. We first propose a step-by-step guideline on how to clarify and investigate problems in RS using causal concepts. Then, we provide a new taxonomy and give a formal definition of various biases in RS from the perspective of violating what assumptions are adopted in standard causal analysis. Finally, we find that many problems in RS can be well formalized into a few scenarios using the proposed causal analysis framework.
Spiking Neural Networks (SNNs) have gained great attraction due to their distinctive properties of low power consumption and fast inference on neuromorphic hardware. As the most effective method to get deep SNNs, ANN-SNN conversion has achieved comparable performance as ANNs on large-scale datasets. Despite this, it requires long time-steps to match the firing rates of SNNs to the activation of ANNs. As a result, the converted SNN suffers severe performance degradation problems with short time-steps, which hamper the practical application of SNNs. In this paper, we theoretically analyze ANN-SNN conversion error and derive the estimated activation function of SNNs. Then we propose the quantization clip-floor-shift activation function to replace the ReLU activation function in source ANNs, which can better approximate the activation function of SNNs. We prove that the expected conversion error between SNNs and ANNs is zero, enabling us to achieve high-accuracy and ultra-low-latency SNNs. We evaluate our method on CIFAR-10/100 and ImageNet datasets, and show that it outperforms the state-of-the-art ANN-SNN and directly trained SNNs in both accuracy and time-steps. To the best of our knowledge, this is the first time to explore high-performance ANN-SNN conversion with ultra-low latency (4 time-steps).
Spiking Neural Networks (SNNs) have been attached great importance due to the distinctive properties of low power consumption, biological plausibility, and adversarial robustness. The most effective way to train deep SNNs is through ANN-to-SNN conversion, which have yielded the best performance in deep network structure and large-scale datasets. However, there is a trade-off between accuracy and latency. In order to achieve high precision as original ANNs, a long simulation time is needed to match the firing rate of a spiking neuron with the activation value of an analog neuron, which impedes the practical application of SNN. In this paper, we aim to achieve high-performance converted SNNs with extremely low latency (fewer than 32 time-steps). We start by theoretically analyzing ANN-to-SNN conversion and show that scaling the thresholds does play a similar role as weight normalization. Instead of introducing constraints that facilitate ANN-to-SNN conversion at the cost of model capacity, we applied a more direct way by optimizing the initial membrane potential to reduce the conversion loss in each layer. Besides, we demonstrate that optimal initialization of membrane potentials can implement expected error-free ANN-to-SNN conversion. We evaluate our algorithm on the CIFAR-10, CIFAR-100 and ImageNet datasets and achieve state-of-the-art accuracy, using fewer time-steps. For example, we reach top-1 accuracy of 93.38% on CIFAR-10 with 16 time-steps. Moreover, our method can be applied to other ANN-SNN conversion methodologies and remarkably promote performance when the time-steps is small.
Sulfite(S(IV))-induced advanced oxidation processes (AOPs) have great prospect in the field of removing organic pollutants, yet developing highly efficient sulfite activation systems and optimizing active sites for favorable catalytic processes are important but still challenging. Herein, we have achieved a composite catalyst with modulated Co electron structure for efficient AOPs by decorating Co(OH)2 on ultrathin graphitic carbon nitride (g-C3N4) nanosheet through an adjustable strategy, which exhibits high catalytic performance in S(IV) activation system. At optimal pH 9, 92% of paracetamol (APAP) (0.005 mM) is removed with the degradation rate constant of k1 = 0.193 min−1 within 30 min in presence of the composite material. The in-situ synthesis mode introduces strong heterogeneous interface interaction, resulting in directional electron transfer from cobalt hydroxide layer to g-C3N4 sheet revealed by X-ray photoelectron spectroscopy and density functional theory (DFT) calculations. The underlying activity enhanced mechanisms for APAP in S(IV) activation system using Co(OH)2/g-C3N4 are proposed: (i) The ultrathin g-C3N4 nanosheets provide more anchoring centers for generating small Co(OH)2 nanoparticles with abundant active sites which benefit to form metastable intermediates of Co(II)-SO3; (ii) The strong interface interaction induces charge redistribution between Co(OH)2 and g-C3N4 conformed by DFT calculation, which modulates the d-band center of Co site and strengthens the bind of Co(II)-SO3, thereby giving rise to radicals (•OH, SO4• and O2•) and nonradicals (1O2 and electron transfer) oxidation for highly-efficient removal APAP. Our work will pave the way to build an environmentally friendly strategy for emerging organic pollutant degradation in water through building efficient catalysts in sulfite activation system.
Per-and polyfluoroalkyl substances (PFASs) are manmade chemicals that have wide industrial and commercial application. However, little research has been carried out on PFASs pollution in groundwater from a previously contaminated site. Here, we investigated 43 PFASs in a monitoring campaign from two different aquifers in the North China Plain. Our results revealed that total PFASs concentrations ( n-ary sumation 43PFASs) ranged from 0.22 to 3,776.76 ng/L, with no spatial or compositional differences. Moreover, perfluorooctanoic acid (PFOA) and perfluoroheptane sulfonate (PFHpS) were the dominant pollutants with mean concentrations of 177.33 ng/L and 51 ng/L, respectively. n-ary sumation 43PFAS decreased with well depth due to the adsorption of PFASs to the aquifer materials. Water temperature, total organic carbon, dissolved oxygen, and total phosphorus concentrations were correlated to the PFAS concentrations. Principal component analysis indicated that the main sources of PFASs in groundwater were untreated industrial discharge, untreated domestic wastewater, food packaging, aqueous film forming foams and metal plating, and surface runoff, which overlapped with the industries that previously existed in a nearby city. Human health risks from drinking contaminated groundwater were low to the local residents, with children aged 1-2 years being the most sensitive group. One specific site with a high PFOA concentration was of concern, as it was several orders higher than the 70 ng/L recommended by US Environmental Protection Agency health advisory. This study provided baseline data for PFASs in a previously contaminated site, which will help in the development of effective strategies for controlling PFASs pollution in the North China Plain.
Particulate matter (PM) and gaseous hydrogen peroxide (H2O2) interact ubiquitously to influence atmospheric oxidizing capacity. However, quantitative information on H2O2 loss and its fate on urban aerosols remain unclear. This study investigated the kinetics of heterogeneous reactions of H2O2 on PM2.5, and explored how these processes are affected by various experimental conditions (i.e., relative humidity, temperature, and H2O2 concentration). We observed a persistent uptake of H2O2 by PM2.5 (with the uptake coefficients (γ) of 10-4 to 10-3), exacerbated by aerosol liquid water and temperature, confirming the critical role of water-assisted chemical decomposition during the uptake process. A positive correlation between the γ values and the ratio of dissolved iron concentration to H2O2 concentration suggests that a Fenton catalytic decomposition may be an important pathway for H2O2 conversion on PM2.5 under dark conditions. Furthermore, on the basis of kinetic data gained, the parameterization of H2O2 uptake on PM2.5 was developed, and was applied into a box model. The good agreement between simulated and measured H2O2 uncovered the significant role that heterogeneous uptake plays in the sink of H2O2 in the atmosphere. These findings suggest that the composition-dependent particle reactivity toward H2O2 should be considered in atmospheric models for elucidating the environmental and health effects of H2O2 uptake by ambient aerosols.