Abstract: A computer-implemented method comprising: obtaining an output from each of a plurality of kernels in an extraction layer of a first trained convolutional neural network, wherein the first convolutional neural network is configured to identify one or more features in an image; aggregating the outputs corresponding to at least some input samples of a first domain to generate an aggregate map corresponding to that kernel; resizing the aggregate maps to a lower resolution to generate a plurality of region maps corresponding to the aggregate maps, respectively; clustering the region maps to generate clusters of region maps, each cluster comprising region maps having similar regions; and training, using input samples of a second domain, a second convolutional neural network with a kernel weight of at least one of the kernels which corresponds to at least one of the image regions of at least one of the clusters.

Abstract: A computer-implemented method comprising: obtaining an output from each of a plurality of kernels in an extraction layer of a first trained convolutional neural network, wherein the first convolutional neural network is configured to identify one or more features in an image; aggregating the outputs corresponding to at least some input samples of a first domain to generate an aggregate map corresponding to that kernel; resizing the aggregate maps to a lower resolution to generate a plurality of region maps corresponding to the aggregate maps, respectively; clustering the region maps to generate clusters of region maps, each cluster comprising region maps having similar regions; and training, using input samples of a second domain, a second convolutional neural network with a kernel weight of at least one of the kernels which corresponds to at least one of the image regions of at least one of the clusters.

Abstract: A computer-implemented method comprising: obtaining, based on an input image, a first activation map of a labelled filter of a first convolutional neural network, the first convolutional neural network being configured to identify one or more first features in the input image; obtaining, based on the input image, a second activation map of a filter of a second convolutional neural network, the second convolutional neural network being configured to identify one or more second features in the input image; calculating a similarity measure between the first activation map and the second activation map; and labelling, when the similarity measure is equal to or above a threshold similarity, the filter of the second convolutional neural network with a label of the labelled filter of the first convolutional neural network.

Abstract: Rules for explaining the output of an ANN are derived by: creating decision trees trained to approximate the ANN and optimize a defined criterion, a threshold value for the criterion being calculated to determine for which node of the ANN the input activations should be split between branches of the decision tree; obtaining threshold value combinations each comprising a threshold value obtained for respective nodes of the ANN; for each combination, using the combination to perform a rule extraction algorithm to extract a rule explaining the output of the ANN and to obtain a fidelity metric indicating the accuracy of the rule with respect to predictions of the ANN; determining which combination yields the best fidelity metric; and using the rule extraction algorithm with the combination of threshold values determined to yield the best fidelity metric to extract at least one rule for explaining the output of the ANN.

Abstract: An image classification method comprises: extracting a logic program from a CNN, trained to classify features in images, which is a symbolic approximation of outputs of kernels at an extraction layer of the CNN; deriving kernel-based classification rules; forward-propagating pairs of feature-labeled images through the logic program to obtain kernel activations at the extraction layer for features in the images, where the scene in one of the pair contains a particular feature and the other is of the same scene without the feature; and calculating the correlation between each kernel in the logic program and each feature in the feature-labeled images using the kernel activations and the features associated with the feature-labeled images, assigning to each kernel in the logic program the label of the feature with which the kernel has the highest correlation, and applying the assigned kernel labels to the kernels in the rules to obtain kernel-labeled rules.

Abstract: An image processing method to sample the image to generate patches. Feature vectors are extracted from the patches, and the extracted feature vectors are partitioned into clusters, where feature vectors in the same cluster share a common characteristic. A portion of interest in the image is segmented. An aggregate bounding region creation process is carried out by finding the largest segment and creating a bounding box around it; determining which cluster contains the most patches within the bounding box of the segment; and adding the patches of the determined cluster to an aggregate bounding region for the portion of interest. The aggregate bounding region creation process is repeated for each other segment in order of size. The resulting aggregate bounding region contains all the patches associated with the portion of interest. The patches which fall outside the resulting aggregate bounding region are then removed from the image.

Abstract: A computer-implemented method of automatically locating a portion of interest in image or matrix data derived from an item under consideration includes: identifying parts of the image or matrix data that satisfy a preset threshold as objects which are possibly parts of the portion of the interest; applying at least one preselected filter to the data corresponding to the objects to find a set of objects consisting of the objects most likely to be part of the portion of interest; sorting the objects of the set into clusters according to a predefined criterion; and using a known characteristic of the portion of interest to identify which one of the clusters corresponds to the portion of interest.

Abstract: An image processing method to sample the image to generate patches. Feature vectors are extracted from the patches, and the extracted feature vectors are partitioned into clusters, where feature vectors in the same cluster share a common characteristic. A portion of interest in the image is segmented. An aggregate bounding region creation process is carried out by finding the largest segment and creating a bounding box around it; determining which cluster contains the most patches within the bounding box of the segment; and adding the patches of the determined cluster to an aggregate bounding region for the portion of interest. The aggregate bounding region creation process is repeated for each other segment in order of size. The resulting aggregate bounding region contains all the patches associated with the portion of interest. The patches which fall outside the resulting aggregate bounding region are then removed from the image.

Abstract: A computer-implemented method of generating, from time-series data, a time-series of data sets for predictive analysis, comprises dividing the time-series data into evenly-sized overlapping segments of data, generating an image representing data for each segment, using the time-series data to determine a trend associated with each image, and storing each of the generated images and its associated trend as a data set. In some embodiments of the method the image from each stored data set is transformed into numerical vectors through a feature extraction process using a pre-trained convolutional neural network. The numerical vectors are stored in association with the data set, and the data sets and associated numerical vectors are used to predict the trend for a new time-series image which has been generated from any time-series data.

Abstract: A computer-implemented method of automatically locating a portion of interest in image or matrix data derived from an item under consideration includes: identifying parts of the image or matrix data that satisfy a preset threshold as objects which are possibly parts of the portion of the interest; applying at least one preselected filter to the data corresponding to the objects to find a set of objects consisting of the objects most likely to be part of the portion of interest; sorting the objects of the set into clusters according to a predefined criterion; and using a known characteristic of the portion of interest to identify which one of the clusters corresponds to the portion of interest.

Abstract: A computer-implemented method of generating, from time-series data, a time-series of data sets for predictive analysis, comprises dividing the time-series data into evenly-sized overlapping segments of data, generating an image representing data for each segment, using the time-series data to determine a trend associated with each image, and storing each of the generated images and its associated trend as a data set. In some embodiments of the method the image from each stored data set is transformed into numerical vectors through a feature extraction process using a pre-trained convolutional neural network. The numerical vectors are stored in association with the data set, and the data sets and associated numerical vectors are used to predict the trend for a new time-series image which has been generated from any time-series data.

Abstract: A vehicle roof rack for securing objects to the top of a vehicle having cross-bars mounted to the roof of the vehicle, has two pivot bars slidably connected to two load tubes. Each pivot bar is connected to a bracket via a hinge, and bracket is then mounted onto the cross-bars. Each of the load tubes has an interior cavity and the first ends of the pivot bars are slidably inserted within an end of the load tubes, so that the load tubes can be slid between an extended position and a closed position. Each load tube has a first portion with an open profile, and a second portion with a closed profile. In the extended position, the second portion of the load tube abuts the mounting bracket and acts as a stop to keep the load tube from freeing itself from the pivot bar.

Abstract: A vehicle roof rack for securing objects to the top of a vehicle having cross-bars mounted to the roof of the vehicle, has two pivot bars slidably connected to two load tubes. Each pivot bar is connected to a bracket via a hinge, and bracket is then mounted onto the cross-bars. Each of the load tubes has an interior cavity and the first ends of the pivot bars are slidably inserted within an end of the load tubes, so that the load tubes can be slid between an extended position and a closed position. Each load tube has a first portion with an open profile, and a second portion with a closed profile. In the extended position, the second portion of the load tube abuts the mounting bracket and acts as a stop to keep the load tube from freeing itself from the pivot bar.