Language-independent tokenisation (LIT) methods that do not require labelled language resources or lexicons have recently gained popularity because of their applicability in resource-poor languages. Moreover, they compactly represent a language using a fixed size vocabulary and can efficiently handle unseen or rare words. On the other hand, language-specific tokenisation (LST) methods have a long and established history, and are developed using carefully created lexicons and training resources. Unlike subtokens produced by LIT methods, LST methods produce valid morphological subwords. Despite the contrasting tradeoffs between LIT vs. LST methods, their performance on downstream NLP tasks remain unclear. In this paper, we empirically compare the two approaches using semantic similarity measurement as an evaluation task across a diverse set of languages. Our experimental results covering eight languages show that LST consistently outperforms LIT when the vocabulary size is large, but LIT can produce comparable or better results than LST in many languages with comparatively smaller (i.e. less than 100K words) vocabulary sizes, encouraging the use of LIT when language-specific resources are unavailable, incomplete or a smaller model is required. Moreover, we find that smoothed inverse frequency (SIF) to be an accurate method to create word embeddings from subword embeddings for multilingual semantic similarity prediction tasks. Further analysis of the nearest neighbours of tokens show that semantically and syntactically related tokens are closely embedded in subword embedding spaces.

Cruise operators are increasingly offering bite-sized breaks along the coast of Australia for domestic travellers keen to explore the island continent.

Deep learning offers state of the art solutions for image recognition. However, deep models are vulnerable to adversarial perturbations in images that are subtle but significantly change the model's prediction. In a white-box attack, these perturbations are generally learned for deep models that operate on RGB images and, hence, the perturbations are equally distributed in the RGB color space. In this paper, we show that the adversarial perturbations prevail in the Y-channel of the YCbCr space. Our finding is motivated from the fact that the human vision and deep models are more responsive to shape and texture rather than color. Based on our finding, we propose a defense against adversarial images. Our defence, coined ResUpNet, removes perturbations only from the Y-channel by exploiting ResNet features in an upsampling framework without the need for a bottleneck. At the final stage, the untouched CbCr-channels are combined with the refined Y-channel to restore the clean image. Note that ResUpNet is model agnostic as it does not modify the DNN structure. ResUpNet is trained end-to-end in Pytorch and the results are compared to existing defence techniques in the input transformation category. Our results show that our approach achieves the best balance between defence against adversarial attacks such as FGSM, PGD and DDN and maintaining the original accuracies of VGG-16, ResNet50 and DenseNet121 on clean images. We perform another experiment to show that learning adversarial perturbations only for the Y-channel results in higher fooling rates for the same perturbation magnitude.

Graph representation learning embeds nodes in large graphs as low-dimensional vectors and is of great benefit to many downstream applications. Most embedding frameworks, however, are inherently transductive and unable to generalize to unseen nodes or learn representations across different graphs. Although inductive approaches can generalize to unseen nodes, they neglect different contexts of nodes and cannot learn node embeddings dually. In this paper, we present a context-aware unsupervised dual encoding framework, \textbf{CADE}, to generate representations of nodes by combining real-time neighborhoods with neighbor-attentioned representation, and preserving extra memory of known nodes. We exhibit that our approach is effective by comparing to state-of-the-art methods.

An open question in the Deep Learning community is why neural networks trained with Gradient Descent generalize well on real datasets even though they are capable of fitting random data. We propose an approach to answering this question based on a hypothesis about the dynamics of gradient descent that we call Coherent Gradients: Gradients from similar examples are similar and so the overall gradient is stronger in certain directions where these reinforce each other. Thus changes to the network parameters during training are biased towards those that (locally) simultaneously benefit many examples when such similarity exists. We support this hypothesis with heuristic arguments and perturbative experiments and outline how this can explain several common empirical observations about Deep Learning. Furthermore, our analysis is not just descriptive, but prescriptive. It suggests a natural modification to gradient descent that can greatly reduce overfitting.

The Dirichlet Belief Network~(DirBN) has been recently proposed as a promising approach in learning interpretable deep latent representations for objects. In this work, we leverage its interpretable modelling architecture and propose a deep dynamic probabilistic framework -- the Recurrent Dirichlet Belief Network~(Recurrent-DBN) -- to study interpretable hidden structures from dynamic relational data. The proposed Recurrent-DBN has the following merits: (1) it infers interpretable and organised hierarchical latent structures for objects within and across time steps; (2) it enables recurrent long-term temporal dependence modelling, which outperforms the one-order Markov descriptions in most of the dynamic probabilistic frameworks. In addition, we develop a new inference strategy, which first upward-and-backward propagates latent counts and then downward-and-forward samples variables, to enable efficient Gibbs sampling for the Recurrent-DBN. We apply the Recurrent-DBN to dynamic relational data problems. The extensive experiment results on real-world data validate the advantages of the Recurrent-DBN over the state-of-the-art models in interpretable latent structure discovery and improved link prediction performance.

The complexity of large-scale neural networks can lead to poor understanding of their internal details. We show that this opaqueness provides an opportunity for adversaries to embed unintended functionalities into the network in the form of Trojan horses. Our novel framework hides the existence of a Trojan network with arbitrary desired functionality within a benign transport network. We prove theoretically that the Trojan network's detection is computationally infeasible and demonstrate empirically that the transport network does not compromise its disguise. Our paper exposes an important, previously unknown loophole that could potentially undermine the security and trustworthiness of machine learning.

Causal effect estimation from observational data is a crucial but challenging task. Currently, only a limited number of data-driven causal effect estimation methods are available. These methods either only provide a bound estimation of the causal effect of a treatment on the outcome, or have impractical assumptions on the data or low efficiency although providing a unique estimation of the causal effect. In this paper, we identify a practical problem setting and propose an approach to achieving unique causal effect estimation from data with hidden variables under this setting. For the approach, we develop the theorems to support the discovery of the proper covariate sets for confounding adjustment (adjustment sets). Based on the theorems, two algorithms are presented for finding the proper adjustment sets from data with hidden variables to obtain unbiased and unique causal effect estimation. Experiments with benchmark Bayesian networks and real-world datasets have demonstrated the efficiency and effectiveness of the proposed algorithms, indicating the practicability of the identified problem setting and the potential of the approach in real-world applications.

In July this year, Holly Ransom (Fulbright Anne Wexler Scholar, CEO Emergent) interviewed former US Ambassador to Australia Jeffrey Bleich (Professorial Fellow at Flinders University) for the launch of JBC and the first instalment in the Jeff Bleich Series – a multimedia platform for the Centre's research, engagement, and education. In a wide-ranging interview, Ambassador Bleich outlines the JBC vision, its values, goals, and aspirations, and addresses the spectrum of challenges and opportunities Australia and the United States confront in the digital age. From the impacts of automation, Artificial Intelligence, 5G & Blockchain, to the need to reinvigorate good governance, democratic participation, civil society and industry collaboration, and community and individual level empowerment, this timely and important discussion is not to be missed.

Costa Group, one of Australia's largest horticulturist companies, has begun rolling out an artificial intelligence (AI) system to help the company better understand and manage the quantity and quality of its berry crops. The Sensing system, developed by Sydney-based company, The Yield, has been designed to measure 14 variables of a typical agriculture model such as rain, light, wind, temperature, and soil moisture in real time. The information is then ingested into an Internet of Things (IoT) platform and combined with existing data sets shared by Costa before AI is applied to create a localised prediction of each berry crop. "We literally describe the system like a maths robot because it's effectively crunching through data and selecting the most important feature sets, creating models, putting them into production, measuring the accuracy, feeding that back in, and continually adjusting," The Yield founder and managing director Ros Harvey told ZDNet. The system was recently installed within the polytunnels of Costa's eight berry farms in New South Wales, Queensland, and Tasmania.