The unbiased estimation of the fraction of variance explained by a model

The correlation coefficient squared, r², is commonly used to validate quantitative models on neural data, yet it is biased by trial-to-trial variability: as trial-to-trial variability increases, measured correlation to a model’s predictions decreases. As a result, models that perfectly explain neural tuning can appear to perform poorly. Many solutions to this problem have been proposed, but no consensus has been reached on which is the least biased estimator. Some currently used methods substantially overestimate model fit, and the utility of even the best performing methods is limited by the lack of confidence intervals and asymptotic analysis. We provide a new estimator, r^ER2, that outperforms all prior estimators in our testing, and we provide confidence intervals and asymptotic guarantees. We apply our estimator to a variety of neural data to validate its utility. We find that neural noise is often so great that confidence intervals of the estimator cover the entire possible range of values ([0, 1]), preventing meaningful evaluation of the quality of a model’s predictions. This leads us to propose the use of the signal-to-noise ratio (SNR) as a quality metric for making quantitative comparisons across neural recordings. Analyzing a variety of neural data sets, we find that up to ∼ 40% of some state-of-the-art neural recordings do not pass even a liberal SNR criterion. Moving toward more reliable estimates of correlation, and quantitatively comparing quality across recording modalities and data sets, will be critical to accelerating progress in modeling biological phenomena.     

Estimating T-cell repertoire diversity: limitations of classical estimators and a new approach

A highly diverse T-cell receptor (TCR) repertoire is a fundamental property of an effective immune system, and is associated with efficient control of viral infections and other pathogens. However, direct measurement of total TCR diversity is impossible. The diversity is high and the frequency distribution of individual TCRs is heavily skewed; the diversity therefore cannot be captured in a blood sample. Consequently, estimators of the total number of TCR clonotypes that are present in the individual, in addition to those observed, are essential. This is analogous to the ‘unseen species problem’ in ecology. We review the diversity (species richness) estimators that have been applied to T-cell repertoires and the methods used to validate these estimators. We show that existing approaches have significant shortcomings, and frequently underestimate true TCR diversity. We highlight our recently developed estimator, DivE, which can accurately estimate diversity across a range of immunological and biological systems.     

Is the cardiac monitoring function related to the self in both the default network and right anterior insula?

The self has been proposed to be rooted in the neural monitoring of internal bodily signals and might thus involve interoceptive areas, notably the right anterior insula (rAI). However, studies on the self consistently showed the involvement of midline default network (DN) nodes, without referring to visceral monitoring. Here, we investigate this apparent discrepancy. We previously showed that neural responses to heartbeats in the DN encode two different self-dimensions, the agentive ‘I’ and the introspective ‘Me’, in a whole-brain analysis of magnetoencephalography (MEG) data. Here, we confirm and anatomically refine this result with intracranial recordings (intracranial electroencephalography, iEEG). In two patients, we show a parametric modulation of neural responses to heartbeats by the self-relatedness of thoughts, at the single trial level. A region-of-interest analysis of the insula reveals that MEG responses to heartbeats in the rAI encode the ‘I’ self-dimension. The effect in rAI was weaker than in the DN and was replicated in iEEG data in one patient out of two. We propose that a common mechanism, the neural monitoring of cardiac signals, underlies the self in both the DN and rAI. This might reconcile studies on the self highlighting the DN, with studies on interoception focusing on the insula.This article is part of the themed issue ‘Interoception beyond homeostasis: affect, cognition and mental health’.     

Selective and Efficient Neural Coding of Communication Signals Depends on Early Acoustic and Social Environment

Previous research has shown that postnatal exposure to simple, synthetic sounds can affect the sound representation in the auditory cortex as reflected by changes in the tonotopic map or other relatively simple tuning properties, such as AM tuning. However, their functional implications for neural processing in the generation of ethologically-based perception remain unexplored. Here we examined the effects of noise-rearing and social isolation on the neural processing of communication sounds such as species-specific song, in the primary auditory cortex analog of adult zebra finches. Our electrophysiological recordings reveal that neural tuning to simple frequency-based synthetic sounds is initially established in all the laminae independent of patterned acoustic experience; however, we provide the first evidence that early exposure to patterned sound statistics, such as those found in native sounds, is required for the subsequent emergence of neural selectivity for complex vocalizations and for shaping neural spiking precision in superficial and deep cortical laminae, and for creating efficient neural representations of song and a less redundant ensemble code in all the laminae. Our study also provides the first causal evidence for ‘sparse coding’, such that when the statistics of the stimuli were changed during rearing, as in noise-rearing, that the sparse or optimal representation for species-specific vocalizations disappeared. Taken together, these results imply that a layer-specific differential development of the auditory cortex requires patterned acoustic input, and a specialized and robust sensory representation of complex communication sounds in the auditory cortex requires a rich acoustic and social environment.     

Shrinkage Estimators for Covariance Matrices

Estimation of covariance matrices in small samples has been studied by many authors. Standard estimators, like the unstructured maximum likelihood estimator (ML) or restricted maximum likelihood (REML) estimator, can be very unstable with the smallest estimated eigenvalues being too small and the largest too big. A standard approach to more stably estimating the matrix in small samples is to compute the ML or REML estimator under some simple structure that involves estimation of fewer parameters, such as compound symmetry or independence. However, these estimators will not be consistent unless the hypothesized structure is correct. If interest focuses on estimation of regression coefficients with correlated (or longitudinal) data, a sandwich estimator of the covariance matrix may be used to provide standard errors for the estimated coefficients that are robust in the sense that they remain consistent under misspecification of the covariance structure. With large matrices, however, the inefficiency of the sandwich estimator becomes worrisome. We consider here two general shrinkage approaches to estimating the covariance matrix and regression coefficients. The first involves shrinking the eigenvalues of the unstructured ML or REML estimator. The second involves shrinking an unstructured estimator toward a structured estimator. For both cases, the data determine the amount of shrinkage. These estimators are consistent and give consistent and asymptotically efficient estimates for regression coefficients. Simulations show the improved operating characteristics of the shrinkage estimators of the covariance matrix and the regression coefficients in finite samples. The final estimator chosen includes a combination of both shrinkage approaches, i.e., shrinking the eigenvalues and then shrinking toward structure. We illustrate our approach on a sleep EEG study that requires estimation of a 24 x 24 covariance matrix and for which inferences on mean parameters critically depend on the covariance estimator chosen. We recommend making inference using a particular shrinkage estimator that provides a reasonable compromise between structured and unstructured estimators.     

Tethering-facilitated DNA ‘opening’ and complementary roles of β-hairpin motifs in the Rad4/XPC DNA damage sensor protein

XPC/Rad4 initiates eukaryotic nucleotide excision repair on structurally diverse helix-destabilizing/distorting DNA lesions by selectively ‘opening’ these sites while rapidly diffusing along undamaged DNA. Previous structural studies showed that Rad4, when tethered to DNA, could also open undamaged DNA, suggesting a ‘kinetic gating’ mechanism whereby lesion discrimination relied on efficient opening versus diffusion. However, solution studies in support of such a mechanism were lacking and how ‘opening’ is brought about remained unclear. Here, we present crystal structures and fluorescence-based conformational analyses on tethered complexes, showing that Rad4 can indeed ‘open’ undamaged DNA in solution and that such ‘opening’ can largely occur without one or the other of the β-hairpin motifs in the BHD2 or BHD3 domains. Notably, the Rad4-bound ‘open’ DNA adopts multiple conformations in solution notwithstanding the DNA’s original structure or the β-hairpins. Molecular dynamics simulations reveal compensatory roles of the β-hairpins, which may render robustness in dealing with and opening diverse lesions. Our study showcases how fluorescence-based studies can be used to obtain information complementary to ensemble structural studies. The tethering-facilitated DNA ‘opening’ of undamaged sites and the dynamic nature of ‘open’ DNA may shed light on how the protein functions within and beyond nucleotide excision repair in cells.     

On the Properties of GEE Estimators in the Presence of Invariant Covariates

Since Liang and Zeger (1986) proposed the ‘generalized estimating equations’ approach for the estimation of regression parameters in models with correlated discrete responses, a lot of work has been devoted to the investigation of the properties of the corresponding GEE estimators. However, the effects of different kinds of covariates have often been overlooked. In this paper it is shown that the use of non-singular block invariant matrices of covariates, as e.g. a design matrix in an analysis of variance model, leads to GEE estimators which are identical regardless of the ‘working’ correlation matrix used. Moreover, they are efficient (McCullagh, 1983). If on the other hand only covariates are used which are invariant within blocks, the efficiency gain in choosing the ‘correct’ vs. an ‘incorrect’ correlation structure is shown to be negligible. The results of a simple simulation study suggest that although different GEE estimators are not identical and are not as efficient as a ML estimator, the differences are still negligible if both types of invariant covariates are present.     

Application of Hierarchical Dissociated Neural Network in Closed-Loop Hybrid System Integrating Biological and Mechanical Intelligence

Neural networks are considered the origin of intelligence in organisms. In this paper, a new design of an intelligent system merging biological intelligence with artificial intelligence was created. It was based on a neural controller bidirectionally connected to an actual mobile robot to implement a novel vehicle. Two types of experimental preparations were utilized as the neural controller including ‘random’ and ‘4Q’ (cultured neurons artificially divided into four interconnected parts) neural network. Compared to the random cultures, the ‘4Q’ cultures presented absolutely different activities, and the robot controlled by the ‘4Q’ network presented better capabilities in search tasks. Our results showed that neural cultures could be successfully employed to control an artificial agent; the robot performed better and better with the stimulus because of the short-term plasticity. A new framework is provided to investigate the bidirectional biological-artificial interface and develop new strategies for a future intelligent system using these simplified model systems.     

A comparison of performance of plant miRNA target prediction tools and the characterization of features for genome-wide target prediction

Background

Deep-sequencing has enabled the identification of large numbers of miRNAs and siRNAs, making the high-throughput target identification a main limiting factor in defining their function. In plants, several tools have been developed to predict targets, majority of them being trained on Arabidopsis datasets. An extensive and systematic evaluation has not been made for their suitability for predicting targets in species other than Arabidopsis. Nor, these have not been evaluated for their suitability for high-throughput target prediction at genome level.

Results

We evaluated the performance of 11 computational tools in identifying genome-wide targets in Arabidopsis and other plants with procedures that optimized score-cutoffs for estimating targets. Targetfinder was most efficient [89% ‘precision’ (accuracy of prediction), 97% ‘recall’ (sensitivity)] in predicting ‘true-positive’ targets in Arabidopsis miRNA-mRNA interactions. In contrast, only 46% of true positive interactions from non-Arabidopsis species were detected, indicating low ‘recall’ values. Score optimizations increased the ‘recall’ to only 70% (corresponding ‘precision’: 65%) for datasets of true miRNA-mRNA interactions in species other than Arabidopsis. Combining the results of Targetfinder and psRNATarget delivers high true positive coverage, whereas the intersection of psRNATarget and Tapirhybrid outputs deliver highly ‘precise’ predictions. The large number of ‘false negative’ predictions delivered from non-Arabidopsis datasets by all the available tools indicate the diversity in miRNAs-mRNA interaction features between Arabidopsis and other species. A subset of miRNA-mRNA interactions differed significantly for features in seed regions as well as the total number of matches/mismatches.

Conclusion

Although, many plant miRNA target prediction tools may be optimized to predict targets with high specificity in Arabidopsis, such optimized thresholds may not be suitable for many targets in non-Arabidopsis species. More importantly, non-conventional features of miRNA-mRNA interaction may exist in plants indicating alternate mode of miRNA target recognition. Incorporation of these divergent features would enable next-generation of algorithms to better identify target interactions.

Electronic supplementary material

The online version of this article (doi:10.1186/1471-2164-15-348) contains supplementary material, which is available to authorized users.     

Phenolic compounds and vitamins in wild and cultivated apricot (Prunus armeniaca L.) fruits grown in irrigated and dry farming conditions

Background

Turkey is the main apricot producer in the world and apricots have been produced under both dry and irrigated conditions in the country. In this study, phenolic compounds and vitamins in fruits of one wild (Zerdali) and three main apricot cultivars (‘Cataloglu’, ‘Hacihaliloglu’ and ‘Kabaasi’) grown in both dry and irrigated conditions in Malatya provinces in Turkey were investigated.

Results

The findings indicated that higher content of phenolic compounds and vitamins was found in apricot fruits grown in irrigated conditions. Among the cultivars, ‘Cataloglu’ had the highest rutin contents both in irrigated and dry farming conditions as 2855 μg in irrigated and 6952 μg per 100 g dried weight base in dry conditions and the highest chlorogenic acid content in irrigated and dry farming conditions were measured in fruits of ‘Hacıhaliloglu’ cultivar as 7542 μg and 15251 μg per 100 g dried weight base. Vitamin C contents in homogenates of fruit flesh and skin was found to be higher than β-caroten, retinol, vitamin E and lycopen contents in apricot fruits both in irrigated and dry farming conditions.

Conclusion

The results suggested that apricot fruits grown in both dry and irrigated conditions had high health benefits phytochemicals and phytochemical content varied among cultivars and irrigation conditions as well. However, more detailed biological and pharmacological studies are needed for the demonstration and clarification of health benefits of apricot fruits.     

Can the meaning of multiple words be integrated unconsciously?

What are the limits of unconscious language processing? Can language circuits process simple grammatical constructions unconsciously and integrate the meaning of several unseen words? Using behavioural priming and electroencephalography (EEG), we studied a specific rule-based linguistic operation traditionally thought to require conscious cognitive control: the negation of valence. In a masked priming paradigm, two masked words were successively (Experiment 1) or simultaneously presented (Experiment 2), a modifier (‘not’/‘very’) and an adjective (e.g. ‘good’/‘bad’), followed by a visible target noun (e.g. ‘peace’/‘murder’). Subjects indicated whether the target noun had a positive or negative valence. The combination of these three words could either be contextually consistent (e.g. ‘very bad - murder’) or inconsistent (e.g. ‘not bad - murder’). EEG recordings revealed that grammatical negations could unfold partly unconsciously, as reflected in similar occipito-parietal N400 effects for conscious and unconscious three-word sequences forming inconsistent combinations. However, only conscious word sequences elicited P600 effects, later in time. Overall, these results suggest that multiple unconscious words can be rapidly integrated and that an unconscious negation can automatically ‘flip the sign’ of an unconscious adjective. These findings not only extend the limits of subliminal combinatorial language processes, but also highlight how consciousness modulates the grammatical integration of multiple words.     

Gamma Oscillations as a Neural Signature of Shifting Times in Narrative Language

Verbs and other temporal expressions allow speakers to specify the location of events in time, as well as to move back and forth in time, shifting in a narrative between past, present and future. The referential flexibility of temporal expressions is well understood in linguistics but its neurocognitive bases remain unknown. We aimed at obtaining a neural signature of shifting times in narrative language. We recorded and analyzed event-related brain potentials (ERPs) and oscillatory responses to the adverb ‘now’ and to the second main verb in Punctual (‘An hour ago the boy stole a candy and now he peeled the fruit’) and Iterative (‘The entire afternoon the boy stole candy and now he peeled the fruit’) contexts. ‘An hour ago’ introduces a time frame that lies entirely in the past, ‘now’ shifts the narrative to the present, and ‘peeled’ shifts it back to the past. These two referential shifts in Punctual contexts are expected to leave very similar traces on neural responses. In contrast, ‘The entire afternoon’ specifies a time frame that may encompass past, present and future, such that both ‘now’ and ‘peeled’ are consistent with it. Here, no time shift is required. We found no difference in ERPs between Punctual and Iterative contexts either at ‘now’ or at the second verb. However, reference shifts modulated oscillatory signals. ‘Now’ and the second verb in Punctual contexts resulted in similar responses: an increase in gamma power with a left-anterior distribution. Gamma bursts were absent in Iterative contexts. We propose that gamma oscillations here reflect the binding of temporal variables to the values allowed by constraints introduced by temporal expressions in discourse.     

Enhanced Growth of Endothelial Precursor Cells on PCG-Matrix Facilitates Accelerated,Fibrosis-Free,Wound Healing: A Diabetic Mouse Model

Diabetes mellitus (DM)-induced endothelial progenitor cell (EPC) dysfunction causes impaired wound healing, which can be rescued by delivery of large numbers of ‘normal’ EPCs onto such wounds. The principal challenges herein are (a) the high number of EPCs required and (b) their sustained delivery onto the wounds. Most of the currently available scaffolds either serve as passive devices for cellular delivery or allow adherence and proliferation, but not both. This clearly indicates that matrices possessing both attributes are ‘the need of the day’ for efficient healing of diabetic wounds. Therefore, we developed a system that not only allows selective enrichment and expansion of EPCs, but also efficiently delivers them onto the wounds. Murine bone marrow-derived mononuclear cells (MNCs) were seeded onto a PolyCaprolactone-Gelatin (PCG) nano-fiber matrix that offers a combined advantage of strength, biocompatibility wettability; and cultured them in EGM2 to allow EPC growth. The efficacy of the PCG matrix in supporting the EPC growth and delivery was assessed by various in vitro parameters. Its efficacy in diabetic wound healing was assessed by a topical application of the PCG-EPCs onto diabetic wounds. The PCG matrix promoted a high-level attachment of EPCs and enhanced their growth, colony formation, and proliferation without compromising their viability as compared to Poly L-lactic acid (PLLA) and Vitronectin (VN), the matrix and non-matrix controls respectively. The PCG-matrix also allowed a sustained chemotactic migration of EPCs in vitro. The matrix-effected sustained delivery of EPCs onto the diabetic wounds resulted in an enhanced fibrosis-free wound healing as compared to the controls. Our data, thus, highlight the novel therapeutic potential of PCG-EPCs as a combined ‘growth and delivery system’ to achieve an accelerated fibrosis-free healing of dermal lesions, including diabetic wounds.     

Disturbance Alters the Phylogenetic Composition and Structure of Plant Communities in an Old Field System

The changes in phylogenetic composition and structure of communities during succession following disturbance can give us insights into the forces that are shaping communities over time. In abandoned agricultural fields, community composition changes rapidly when a field is plowed, and is thought to reflect a relaxation of competition due to the elimination of dominant species which take time to re-establish. Competition can drive phylogenetic overdispersion, due to phylogenetic conservation of ‘niche’ traits that allow species to partition resources. Therefore, undisturbed old field communities should exhibit higher phylogenetic dispersion than recently disturbed systems, which should be relatively ‘clustered’ with respect to phylogenetic relationships. Several measures of phylogenetic structure between plant communities were measured in recently plowed areas and nearby ‘undisturbed’ sites. There was no difference in the absolute values of these measures between disturbed and ‘undisturbed’ sites. However, there was a difference in the ‘expected’ phylogenetic structure between habitats, leading to significantly lower than expected phylogenetic diversity in disturbed plots, and no difference from random expectation in ‘undisturbed’ plots. This suggests that plant species characteristic of each habitat are fairly evenly distributed on the shared species pool phylogeny, but that once the initial sorting of species into the two habitat types has occurred, the processes operating on them affect each habitat differently. These results were consistent with an analysis of correlation between phylogenetic distance and co-occurrence indices of species pairs in the two habitat types. This study supports the notion that disturbed plots are more clustered than expected, rather than ‘undisturbed’ plots being more overdispersed, suggesting that disturbed plant communities are being more strongly influenced by environmental filtering of conserved niche traits.     

Conceptual learning by miniature brains

Concepts act as a cornerstone of human cognition. Humans and non-human primates learn conceptual relationships such as ‘same’, ‘different’, ‘larger than’, ‘better than’, among others. In all cases, the relationships have to be encoded by the brain independently of the physical nature of objects linked by the relation. Consequently, concepts are associated with high levels of cognitive sophistication and are not expected in an insect brain. Yet, various works have shown that the miniature brain of honeybees rapidly learns conceptual relationships involving visual stimuli. Concepts such as ‘same’, ‘different’, ‘above/below of’ or ‘left/right are well mastered by bees. We review here evidence about concept learning in honeybees and discuss both its potential adaptive advantage and its possible neural substrates. The results reviewed here challenge the traditional view attributing supremacy to larger brains when it comes to the elaboration of concepts and have wide implications for understanding how brains can form conceptual relations.     

Joint Effect of Education and Main Lifetime Occupation on Late Life Health: A Cross-Sectional Study of Older Adults in Xiamen,China

Background

The effects of education and occupation on health have been well documented individually, but little is known about their joint effect, especially their cumulative joint effect on late life health.

Methods

We enrolled 14,292 participants aged 60+ years by multistage sampling across 173 communities in Xiamen, China, in 2013. Heath status was assessed by the ability to perform six basic activities of daily life. Education was classified in four categories: ‘Illiterate’, ‘Primary’, ‘Junior high school’ and ‘Senior high school and beyond’. Main lifetime occupation was also four categorized: ‘Employed’, ‘Farmer’, ‘Jobless’ and ‘Others’. Odds ratios (ORs) were estimated by random-intercept multilevel models regressing health status on education and main lifetime occupation with or without their interactions, adjusting by some covariates.

Results

Totally, 13,880 participants had complete data, of whom 12.5% suffered from disability, and ‘Illiterate’ and ‘Farmer’ took up the greatest proportion (33.01% and 42.72%, respectively). Participants who were higher educated had better health status (ORs = 0.62, 0.46, and 0.44 for the ‘Primary’, ‘Junior high school’, and ‘Senior high school and beyond’, respectively, in comparison with ‘Illiterate’). Those who were long term jobless in early life had poorest heath (ORs = 1.88, 95% CI 1.47 to 2.40). Unexpectedly, for the farmers, the risk of poor health gradually increased in relation to higher education level (ORs = 1.26, 1.28, 1.40 and 2.24, respectively). For the ‘Employed’, similar ORs were obtained for the ‘Junior high school’ and ‘Senior high school and beyond’ educated (both ORs = 1.01). For the ‘Farmer’ and ‘Jobless’, participants who were ‘Illiterate’ and ‘Primary’ educated also showed similar ORs.

Conclusions

Both education and main lifetime occupation were associated with late life health. Higher education was observed to be associated with better health, but such educational advantage was mediated by main lifetime occupation.     

Red:far-red light conditions affect the emission of volatile organic compounds from barley (Hordeum vulgare), leading to altered biomass allocation in neighbouring plants

Background and Aims Volatile organic compounds (VOCs) play various roles in plant–plant interactions, and constitutively produced VOCs might act as a cue to sense neighbouring plants. Previous studies have shown that VOCs emitted from the barley (Hordeum vulgare) cultivar ‘Alva’ cause changes in biomass allocation in plants of the cultivar ‘Kara’. Other studies have shown that shading and the low red:far-red (R:FR) conditions that prevail at high plant densities can reduce the quantity and alter the composition of the VOCs emitted by Arabidopsis thaliana, but whether this affects plant–plant signalling remains unknown. This study therefore examines the effects of far-red light enrichment on VOC emissions and plant–plant signalling between ‘Alva’ and ‘Kara’.Methods The proximity of neighbouring plants was mimicked by supplemental far-red light treatment of VOC emitter plants of barley grown in growth chambers. Volatiles emitted by ‘Alva’ under control and far-red light-enriched conditions were analysed using gas chromatography–mass spectrometry (GC-MS). ‘Kara’ plants were exposed to the VOC blend emitted by the ‘Alva’ plants that were subjected to either of the light treatments. Dry matter partitioning, leaf area, stem and total root length were determined for ‘Kara’ plants exposed to ‘Alva’ VOCs, and also for ‘Alva’ plants exposed to either control or far-red-enriched light treatments.Key Results Total VOC emissions by ‘Alva’ were reduced under low R:FR conditions compared with control light conditions, although individual volatile compounds were found to be either suppressed, induced or not affected by R:FR. The altered composition of the VOC blend emitted by ‘Alva’ plants exposed to low R:FR was found to affect carbon allocation in receiver plants of ‘Kara’.Conclusions The results indicate that changes in R:FR light conditions influence the emissions of VOCs in barley, and that these altered emissions affect VOC-mediated plant–plant interactions.     

Fast,Simple and Accurate Handwritten Digit Classification by Training Shallow Neural Network Classifiers with the ‘Extreme Learning Machine’ Algorithm

Recent advances in training deep (multi-layer) architectures have inspired a renaissance in neural network use. For example, deep convolutional networks are becoming the default option for difficult tasks on large datasets, such as image and speech recognition. However, here we show that error rates below 1% on the MNIST handwritten digit benchmark can be replicated with shallow non-convolutional neural networks. This is achieved by training such networks using the ‘Extreme Learning Machine’ (ELM) approach, which also enables a very rapid training time (∼ 10 minutes). Adding distortions, as is common practise for MNIST, reduces error rates even further. Our methods are also shown to be capable of achieving less than 5.5% error rates on the NORB image database. To achieve these results, we introduce several enhancements to the standard ELM algorithm, which individually and in combination can significantly improve performance. The main innovation is to ensure each hidden-unit operates only on a randomly sized and positioned patch of each image. This form of random ‘receptive field’ sampling of the input ensures the input weight matrix is sparse, with about 90% of weights equal to zero. Furthermore, combining our methods with a small number of iterations of a single-batch backpropagation method can significantly reduce the number of hidden-units required to achieve a particular performance. Our close to state-of-the-art results for MNIST and NORB suggest that the ease of use and accuracy of the ELM algorithm for designing a single-hidden-layer neural network classifier should cause it to be given greater consideration either as a standalone method for simpler problems, or as the final classification stage in deep neural networks applied to more difficult problems.