A fast and memory‐efficient new method for performing genetic algorithm partial least squares (GA‐PLS) on spectroscopic data preprocessed in multiple different ways is presented. The method, which is primarily intended for datasets containing many observations, involves preprocessing a spectral dataset with several different techniques and concatenating the different versions of the data horizontally into a design matrixXwhich is both tall and wide. The large matrix is then condensed into a substantially smaller covariance matrixX^TXwhose resulting size is unrelated to the number of observations in the dataset, i.e. the height ofX. It is demonstrated that the smaller covariance matrix can be used to efficiently calibrate partial least squares (PLS) models containing feature selections from any of the involved preprocessing techniques. The method is incorporated into GA‐PLS and used to evolve variable selections for a set of different preprocessing techniques concurrently within a single algorithm. This allows a single instance of GA‐PLS to determine which preprocessing technique, within the set of considered methods, is best suited for the spectroscopic dataset. Additionally, the method allows feature selections to be evolved containing variables from a mixture of different preprocessing techniques. The benefits of the introduced GA‐PLS technique can be summarized as threefold: (1) for datasets with many observations, the proposed method is substantially faster compared to conventional GA‐PLS implementations based on NIPALS, SIMPLS, etc. (2) using a single GA‐PLS automatically reveals which of the considered preprocessing techniques results in the lowest model error. (3) it allows the exploration of highly complex solutions composed of features preprocessed using various techniques.

A novel formulation of the wide kernel algorithm for partial least squares regression (PLSR) is proposed. We show how the elimination of redundant calculations in the traditional applications of PLSR helps in speeding up any choice of cross‐validation strategy by utilizing precalculated lookup matrices.The proposed lookup approach is combined with some additional computational shortcuts resulting in highly effective and numerically accurate cross‐validation results. The computational advantages of the proposed method are demonstrated by comparisons to the classical NIPALS and the bidiag2 algorithms for calculating cross‐validated PLSR models. Problems including both one and several responses, double/nested cross‐validated, and one‐vs‐all classification are among the considered applications.

The purpose of this research is to develop a method for estimating the spatially and temporally resolved moisture content of thermally modified Scots pine (Pinus sylvestris) using remote sensing. Hyperspectral time series imaging in the NIR wavelength region (953–2516 nm) was used to gather information about the absorbance of eight thermally modified pine samples each minute as they dried during a period of approximately 20 h. After preprocessing the collected spectral data and identifying an appropriate wavelength selection, partial least squares regression (PLS) was used to map the absorbance data of each pine sample to a distribution of moisture contents within the samples at different time steps during the drying process. To enable separate studying and comparison of the drying dynamics taking place within the early- and latewood regions of the pine samples, the collected images were spatially segmented to separate between early- and latewood pixels. The results of the study indicate that the 1966–2244 nm region of a NIR spectrum, when preprocessed with extended multiplicative scatter correction and first order derivation, can be used to model the average moisture content of thermally modified pine using PLS. The methods presented in this paper allows for estimation and visualization of the intrasample spatial distribution of moisture in thermally modified pine wood.

It is recognised that flame retardant chemicals degrade and leach out of flame-protected wood claddings when exposed to natural weathering. However, the ability to survey the current state of a flame retardant treatment applied to a wood cladding, an arbitrary length of time after the initial application, is limited today. In this study, hyperspectral imaging in the near infrared to short-wavelength infrared region is used to quantify the amount of flame retardant present on wooden surfaces. Several sets of samples were treated with various concentrations of a flame retardant chemical and scanned with a push broom hyperspectral camera. An inductively coupled plasma (ICP) spectroscopy analysis of the outermost layer of the treated samples was then carried out in order to determine each sample’s phosphorus content, the active ingredient in the flame retardant. Spectra from the hyperspectral images were pre-processed with extended multiplicative scatter correction, and the phosphorus content was modelled using a partial least squares (PLS) regression model. The PLS regression yielded robust predictions of surface phosphorus content with a coefficient of determination, R2, between 0.8 and 0.9 on validation data regardless of whether the flame retardant chemical had been applied to the surface of the wood or pressure-impregnated into it. The result from the study indicates that spectral imaging around the 2400–2531 nm wavelength region is favourable for quantifying the amount of phosphorus-based flame retardant contained in the outermost layer of non-coated wooden claddings. The results also reveal that the uptake of phosphorus-based flame retardant does not occur uniformly throughout the wood surface, but is to a larger extent concentrated in the earlywood regions than in the latewood.