Learning Data Streams With Changing Distributions and Temporal Dependency.

In a data stream, concept drift refers to unpredictable distribution changes over time, which violates the identical-distribution assumption required by conventional machine learning methods. Current concept drift adaptation techniques mostly focus on a data stream with changing distributions. However, since each variable of a data stream is a time series, these variables normally have temporal dependency problems in the real world. How to solve concept drift and temporal dependency problems at the same time is rarely discussed in the concept-drift literature. To solve this situation, this article proves and validates that the testing error decreases faster if a predictor is trained on a temporally reconstructed space when drift occurs. Based on this theory, a novel drift adaptation regression (DAR) framework is designed to predict the label variable for data streams with concept drift and temporal dependency. A new statistic called local drift degree (LDD+) is proposed and used as a drift adaptation technique in the DAR framework to discard outdated instances in a timely way, thereby guaranteeing that the most relevant instances will be selected during the training process. The performance of DAR is demonstrated by a set of experimental evaluations on both synthetic data and real-world data streams.

A Drift Region-Based Data Sample Filtering Method.

Concept drift refers to changes in the underlying data distribution of data streams over time. A well-trained model will be outdated if concept drift occurs. Once concept drift is detected, it is necessary to understand where the drift occurs to support the drift adaptation strategy and effectively update the outdated models. This process, called drift understanding, has rarely been studied in this area. To fill this gap, this article develops a drift region-based data sample filtering method to update the obsolete model and track the new data pattern accurately. The proposed method can effectively identify the drift region and utilize information on the drift region to filter the data sample for training models. The theoretical proof guarantees the identified drift region converges uniformly to the real drift region as the sample size increases. Experimental evaluations based on four synthetic datasets and two real-world datasets demonstrate our method improves the learning accuracy when dealing with data streams involving concept drift.

A Segment-Based Drift Adaptation Method for Data Streams.

In concept drift adaptation, we aim to design a blind or an informed strategy to update our best predictor for future data at each time point. However, existing informed drift adaptation methods need to wait for an entire batch of data to detect drift and then update the predictor (if drift is detected), which causes adaptation delay. To overcome the adaptation delay, we propose a sequentially updated statistic, called drift-gradient to quantify the increase of distributional discrepancy when every new instance arrives. Based on drift-gradient, a segment-based drift adaptation (SEGA) method is developed to online update our best predictor. Drift-gradient is defined on a segment in the training set. It can precisely quantify the increase of distributional discrepancy between the old segment and the newest segment when only one new instance is available at each time point. A lower value of drift-gradient on the old segment represents that the distribution of the new instance is closer to the distribution of the old segment. Based on the drift-gradient, SEGA retrains our best predictors with the segments that have the minimum drift-gradient when every new instance arrives. SEGA has been validated by extensive experiments on both synthetic and real-world, classification and regression data streams. The experimental results show that SEGA outperforms competitive blind and informed drift adaptation methods.

Optical Fingerprint Classification of Single Upconversion Nanoparticles by Deep Learning.

Highly controlled synthesis of upconversion nanoparticles (UCNPs) can be achieved in the heterogeneous design, so that a library of optical properties can be arbitrarily produced by depositing multiple lanthanide ions. Such a control offers the potential in creating nanoscale barcodes carrying high-capacity information. With the increasing creation of optical information, it poses more challenges in decoding them in an accurate, high-throughput, and speedy fashion. Here, we reported that the deep-learning approach can recognize the complexity of the optical fingerprints from different UCNPs. Under a wide-field microscope, the lifetime profiles of hundreds of single nanoparticles can be collected at once, which offers a sufficient amount of data to develop deep-learning algorithms. We demonstrated that high accuracies of over 90% can be achieved in classifying 14 kinds of UCNPs. This work suggests new opportunities in handling the diverse properties of nanoscale optical barcodes toward the establishment of vast luminescent information carriers.

Preselectable Optical Fingerprints of Heterogeneous Upconversion Nanoparticles.

The control in optical uniformity of single nanoparticles and tuning their diversity in multiple dimensions, dot to dot, holds the key to unlocking nanoscale applications. Here we report that the entire lifetime profile of the single upconversion nanoparticle (τ2 profile) can be resolved by confocal, wide-field, and super-resolution microscopy techniques. The advances in both spatial and temporal resolutions push the limit of optical multiplexing from microscale to nanoscale. We further demonstrate that the time-domain optical fingerprints can be created by utilizing nanophotonic upconversion schemes, including interfacial energy migration, concentration dependency, energy transfer, and isolation of surface quenchers. We exemplify that three multiple dimensions, including the excitation wavelength, emission color, and τ2 profile, can be built into the nanoscale derivative τ2-dots. Creating a vast library of individually preselectable nanotags opens up a new horizon for diverse applications, spanning from sub-diffraction-limit data storage to high-throughput single-molecule digital assays and super-resolution imaging.

Multiplexed structured illumination super-resolution imaging with lifetime-engineered upconversion nanoparticles.

The emerging optical multiplexing within nanoscale shows super-capacity in encoding information by using lifetime fingerprints from luminescent nanoparticles. However, the optical diffraction limit compromises the decoding accuracy and throughput of the nanoparticles during conventional widefield imaging. This, in turn, challenges the quality of nanoparticles to afford the modulated excitation condition and further retain the multiplexed optical fingerprints for super-resolution multiplexing. Here we report a tailor-made multiplexed super-resolution imaging method using the lifetime-engineered upconversion nanoparticles. We demonstrate that the nanoparticles are bright, uniform, and stable under structured illumination, which supports a lateral resolution of 185 nm, less than 1/4th of the excitation wavelength. We further develop a deep learning algorithm to coordinate with super-resolution images for more accurate decoding compared to a numeric algorithm. We demonstrate a three-channel super-resolution imaging based optical multiplexing with decoding accuracies above 93% for each channel and larger than 60% accuracy for potential seven-channel multiplexing. The improved resolution provides high throughput by resolving the particles within the diffraction-limited spots, which enables higher multiplexing capacity in space. This lifetime multiplexing super-resolution method opens a new horizon for handling the growing amount of information content, disease source, and security risk in modern society.

Development of a hybrid model to predict construction and demolition waste: China as a case study.

Construction and demolition waste (C&DW) is currently a worldwide issue, and the situation is the worst in China due to a rapid increase in the construction industry and the short life span of China's buildings. To create an opportunity out of this problem, comprehensive prevention measures and effective management strategies are urgently needed. One major gap in the literature of waste management is a lack of estimations on future C&DW generation. Therefore, this paper presents a forecasting procedure for C&DW in China that can forecast the quantity of each component in such waste. The proposed approach is based on a GM-SVR model that improves the forecasting effectiveness of the gray model (GM), which is achieved by adjusting the residual series by a support vector regression (SVR) method and a transition matrix that aims to estimate the discharge of each component in the C&DW. Through the proposed method, future C&DW volume are listed and analyzed containing their potential components and distribution in different provinces in China. Besides, model testing process provides mathematical evidence to validate the proposed model is an effective way to give future information of C&DW for policy makers.