TECH OFFERS

One use of AI, including deep learning, is in prediction tasks, such as image scene understanding and medical image diagnosis. As deep learning models are complex, heatmaps are often used to help explain the AI’s prediction by highlighting pixels that were salient to the prediction. While existing heatmaps are effective on clean images, real-world images are frequently degraded or ‘biased’-such as camera blur or colour distortion under low light. Images may also be deliberately blurred for privacy reasons. As the level of image clarity decreases, the performance of the heatmaps decreases. These heatmap explanations of degraded images therefore deviate from both reality and user expectations.  This novel technology-Debiased-CAM-describes a method of training a convolutional neural network (CNN) to produce accurate and relatable heatmaps for degraded images. By pinpointing relevant targets on the images that align with user expectations, Debiased-CAMs increase transparency and user trust in the AI’s predictions. Debiased-CAMs are effective in helping users identify relevant targets even on images affected by different clarity levels and multiple issues such as camera blur, poor lighting conditions and colour distortion. The AI’s prediction also becomes more accurate. As the model is trained using self-supervised learning, no additional data is needed to train it.  The training for Debiased-CAM is generalisable, and thus applicable to other types of degraded or corrupted data and other prediction tasks such as image captioning and human activity recognition. Used to train a convolutional neural network (CNN) to produce accurate and relatable heatmaps for degraded images. By pinpointing relevant targets on the images that align with user expectations, Debiased-CAMs increase transparency and user trust in the AI’s predictions. It also increases the ability of meeting regulatory standards to deploy CNN models in the following applications, where explainable AI is required. Healthcare, eg. Radiology Autonomous Vehicles   Produces accurate, robust and interpretable heatmaps for degraded images Works on images with multiple degradation levels and types such as blurring and improper white balance Agnostic to degradation level, so that enhancement can be applied even when the level is unknown Perceived by users to be more truthful and helpful as compared to current heatmaps distorted due to image degradation Method of training can be applied to other degradation types and prediction tasks Explainable AI Infocomm, Video/Image Analysis & Computer Vision, Artificial Intelligence

Lists are an indispensable part of the online experience, often used to show many results, such as products, web pages, and food dishes. These items can be neatly sorted by a desired attribute like price, relevance, or healthiness. Listed items often have multiple attributes. However, instead of being able to sort multiple attributes simultaneously, consumers are currently limited to sorting only one attribute at a time. This makes searching for the desired item tedious and confusing. Imma Sort supports interpretable and multi-attribute sorting. Sorting for two or more attributes is possible. In contrast to existing search technology, Imma Sort trades off the smoothness of the sorted trend for the main attribute to increase ease of prediction for other attributes, by sorting them more approximately. Results for specific attributes can be made smoother by setting higher importance weights. Provides intuitively sorted results sorted by two or more attributes to improve decision-making and user experience Results can be customised by allocating higher weightage for selected attributes Enables users to perform multi-attribute sorting in any existing list interface without requiring sophisticated spreadsheets or data visualisations Can be integrated into search and recommendation systems across a wide range of applications Can also be incorporated into various search and recommendation systems for more effective search results. Examples of possible applications: Food dishes can be sorted by healthiness and tastiness Hotels can be sorted by price and distance Sorting by price and rating would generate results that generally trend in one direction for both attributes. This makes it easy for users to anticipate the values of multiple attributes as they move down the list, without having to construct a mental list for the secondary attribute. By decreasing users’ mental effort, this will improve decision-making and increase satisfaction. Multi-Attribute Sorting, e-commerce, algorithm Infocomm, eCommerce & ePayment, Enterprise & Productivity