Abstract. Artificial intelligence (AI) techniques are starting to be used in IVF, in particular for selecting which embryos to transfer to the woman. AI has the

UNSW Engineering Professor Flora Salim has been honoured for her pioneering work in computing and machine learning by Women in AI, a global advocacy group for women in the artificial intelligence (AI) field. The 2022 Women in AI Awards Australia and New Zealand recognised women across various industries committed to excellence in AI. Finalists were judged on innovation, leadership and inspiring potential, global impact, and the ability of the AI solution to provide a social good for the community. Prof. Salim was recognised for her AI achievements in the Defence and Intelligence award category. The award acknowledged her research in the cross-cutting areas of ubiquitous computing and machine learning, with a focus on efficient, fair, and explainable machine learning for multi-dimensional sensor data, towards enabling situational and behaviour intelligence for multiple applications.

Deep learning methods have been successfully applied to remote sensing problems for several years. Among these methods, CNN based models have high accuracy in solving the land classification problem using satellite or aerial images. Although these models have high accuracy, this generally comes with large memory size requirements. On the other hand, it is desirable to have small-sized models for applications, such as the ones implemented on unmanned aerial vehicles, with low memory space. Unfortunately, small-sized CNN models do not provide high accuracy as with their large-sized versions. In this study, we propose a novel method to improve the accuracy of CNN models, especially the ones with small size, by injecting traditional features to them. To test the effectiveness of the proposed method, we applied it to the CNN models SqueezeNet, MobileNetV2, ShuffleNetV2, VGG16, and ResNet50V2 having size 0.5 MB to 528 MB. We used the sample mean, gray level co-occurrence matrix features, Hu moments, local binary patterns, histogram of oriented gradients, and color invariants as traditional features for injection. We tested the proposed method on the EuroSAT dataset to perform land classification. Our experimental results show that the proposed method significantly improves the land classification accuracy especially when applied to small-sized CNN models.

Many real-world ubiquitous applications, such as parking recommendations and air pollution monitoring, benefit significantly from accurate long-term spatio-temporal forecasting (LSTF). LSTF makes use of long-term dependency between spatial and temporal domains, contextual information, and inherent pattern in the data. Recent studies have revealed the potential of multi-graph neural networks (MGNNs) to improve prediction performance. However, existing MGNN methods cannot be directly applied to LSTF due to several issues: the low level of generality, insufficient use of contextual information, and the imbalanced graph fusion approach. To address these issues, we construct new graph models to represent the contextual information of each node and the long-term spatio-temporal data dependency structure. To fuse the information across multiple graphs, we propose a new dynamic multi-graph fusion module to characterize the correlations of nodes within a graph and the nodes across graphs via the spatial attention and graph attention mechanisms. Furthermore, we introduce a trainable weight tensor to indicate the importance of each node in different graphs. Extensive experiments on two large-scale datasets demonstrate that our proposed approaches significantly improve the performance of existing graph neural network models in LSTF prediction tasks.

Depression is a serious mental illness that impacts the way people communicate, especially through their emotions, and, allegedly, the way they interact with others. This work examines depression signals in dialogs, a less studied setting that suffers from data sparsity. We hypothesize that depression and emotion can inform each other, and we propose to explore the influence of dialog structure through topic and dialog act prediction. We investigate a Multi-Task Learning (MTL) approach, where all tasks mentioned above are learned jointly with dialog-tailored hierarchical modeling. We experiment on the DAIC and DailyDialog corpora-both contain dialogs in English-and show important improvements over state-ofthe-art on depression detection (at best 70.6% F 1), which demonstrates the correlation of depression with emotion and dialog organization and the power of MTL to leverage information from different sources.

We present a deep neural network for removing undesirable shading features from an unconstrained portrait image, recovering the underlying texture. Our training scheme incorporates three regularization strategies: masked loss, to emphasize high-frequency shading features; soft-shadow loss, which improves sensitivity to subtle changes in lighting; and shading-offset estimation, to supervise separation of shading and texture. Our method demonstrates improved delighting quality and generalization when compared with the state-of-the-art. We further demonstrate how our delighting method can enhance the performance of light-sensitive computer vision tasks such as face relighting and semantic parsing, allowing them to handle extreme lighting conditions.

Music signals are difficult to interpret from their low-level features, perhaps even more than images: e.g. highlighting part of a spectrogram or an image is often insufficient to convey high-level ideas that are genuinely relevant to humans. In computer vision, concept learning was therein proposed to adjust explanations to the right abstraction level (e.g. detect clinical concepts from radiographs). These methods have yet to be used for MIR. In this paper, we adapt concept learning to the realm of music, with its particularities. For instance, music concepts are typically non-independent and of mixed nature (e.g. genre, instruments, mood), unlike previous work that assumed disentangled concepts. We propose a method to learn numerous music concepts from audio and then automatically hierarchise them to expose their mutual relationships. We conduct experiments on datasets of playlists from a music streaming service, serving as a few annotated examples for diverse concepts. Evaluations show that the mined hierarchies are aligned with both ground-truth hierarchies of concepts -- when available -- and with proxy sources of concept similarity in the general case.

Multi-modal learning focuses on training models by equally combining multiple input data modalities during the prediction process. However, this equal combination can be detrimental to the prediction accuracy because different modalities are usually accompanied by varying levels of uncertainty. Using such uncertainty to combine modalities has been studied by a couple of approaches, but with limited success because these approaches are either designed to deal with specific classification or segmentation problems and cannot be easily translated into other tasks, or suffer from numerical instabilities. In this paper, we propose a new Uncertainty-aware Multi-modal Learner that estimates uncertainty by measuring feature density via Cross-modal Random Network Prediction (CRNP). CRNP is designed to require little adaptation to translate between different prediction tasks, while having a stable training process. From a technical point of view, CRNP is the first approach to explore random network prediction to estimate uncertainty and to combine multi-modal data. Experiments on two 3D multi-modal medical image segmentation tasks and three 2D multi-modal computer vision classification tasks show the effectiveness, adaptability and robustness of CRNP. Also, we provide an extensive discussion on different fusion functions and visualization to validate the proposed model.

We present a flow-based control strategy that enables resource-constrained marine robots to patrol gyre-like flow environments on an orbital trajectory with a periodicity in a given range. The controller does not require a detailed model of the flow field and relies only on the robot's location relative to the center of the gyre. Instead of precisely tracking a pre-defined trajectory, the robots are tasked to stay in between two bounding trajectories with known periodicity. Furthermore, the proposed strategy leverages the surrounding flow field to minimize control effort. We prove that the proposed strategy enables robots to cycle in the flow satisfying the desired periodicity requirements. Our method is tested and validated both in simulation and in experiments using a low-cost, underactuated, surface swimming robot, i.e. the Modboat.

The Modboat is a low-cost, underactuated, modular robot capable of surface swimming. It is able to swim individually, dock to other Modboats, and undock from them using only a single motor and two passive flippers. Undocking without additional actuation is achieved by causing intentional self-collision between the tails of neighboring modules; this becomes a challenge when group swimming as one connected component is desirable. In this work, we develop a control strategy to allow parallel lattices of Modboats to swim as a single unit, which conventionally requires holonomic modules. We show that the control strategy is guaranteed to avoid unintentional undocking and minimizes internal forces within the lattice. Experimental verification shows that the controller performs well and is consistent for lattices of various sizes. Controllability is maintained while swimming, but pure yaw control causes lateral movement that cannot be counteracted by the presented framework.