One widely used approach towards understanding the inner workings of deep convolutional neural networks is to visualize unit responses via activation maximization. Feature visualizations via activation maximization are thought to provide humans with precise information about the image features that cause a unit to be activated. If this is indeed true, these synthetic images should enable humans to predict the effect of an intervention, such as whether occluding a certain patch of the image (say, a dog's head) changes a unit's activation. Here, we test this hypothesis by asking humans to predict which of two square occlusions causes a larger change to a unit's activation. Both a large-scale crowdsourced experiment and measurements with experts show that on average, the extremely activating feature visualizations by Olah et al. (2017) indeed help humans on this task ($67 \pm 4\%$ accuracy; baseline performance without any visualizations is $60 \pm 3\%$). However, they do not provide any significant advantage over other visualizations (such as e.g. dataset samples), which yield similar performance ($66 \pm 3\%$ to $67 \pm 3\%$ accuracy). Taken together, we propose an objective psychophysical task to quantify the benefit of unit-level interpretability methods for humans, and find no evidence that feature visualizations provide humans with better "causal understanding" than simple alternative visualizations.

The Australian Federal Court recently handed down its first-instance judgement in Thaler v Commissioner of Patents [2021] FCA 879 where the central issue considered was whether an artificial intelligence (AI) system could be an'inventor' for the purposes of the Australian Patents Act 1990 (Act) and its corresponding regulations. The Court found that an AI system can be an inventor – where'inventor' may be construed broadly to include a'person or thing that invents'1. This decision puts Australia in the spotlight as a favourable country to patent AI-created inventions – for now. Given the subject-matter and controversy generated by this decision, an appeal to the Full Federal Court is almost certain. This Federal Court decision is an appeal from a Patent Office hearing where the Office rejected Australian patent application no. Interestingly, the objection to inventorship was initially raised in a formalities objection issued within a few weeks after the application was filed, and not during examination which would be years later under normal circumstances.

Tim Baldwin is a Melbourne Laureate Professor in the School of Computing and Information Systems, The University of Melbourne, and also Director of the ARC Centre for Cognitive Computing in Medical Technologies and Vice President of the Association for Computational Linguistics. His primary research focus is on natural language processing (NLP), including social media analytics, deep learning, and computational social science. Tim completed a BSc(CS/Maths) and BA(Linguistics/Japanese) at The University of Melbourne in 1995, and an MEng(CS) and PhD(CS) at the Tokyo Institute of Technology in 1998 and 2001, respectively. Prior to joining The University of Melbourne in 2004, he was a Senior Research Engineer at the Center for the Study of Language and Information, Stanford University (2001-2004). His research has been funded by organisations including the Australia Research Council, Google, Microsoft, Xerox, ByteDance, SEEK, NTT, and Fujitsu, and has been featured in MIT Tech Review, IEEE Spectrum, The Times, ABC News, The Age/SMH, Australian Financial Review, and The Australian.

In a landmark decision, an Australian court has set a groundbreaking precedent, deciding artificial intelligence (AI) systems can be legally recognised as an inventor in patent applications. That might not sound like a big deal, but it challenges a fundamental assumption in the law: that only human beings can be inventors. The AI machine called DABUS is an "artificial neural system" and its designs have set off a string of debates and court battles across the globe. On Friday, Australia's Federal Court made the historic finding that "the inventor can be non-human". It came just days after South Africa became the first country to defy the status quo and award a patent recognising DABUS as an inventor. AI pioneer and creator of DABUS, Stephen Thaler, and his legal team have been waging a ferocious global campaign to have DABUS recognised as an inventor for more than two years.

Today, advances in agronomy combined with smart agriculture technology have improved crop yields and sustainability. Enhancements in animal husbandry technology, improved breeding, nutrition and disease management help ensure optimal growth and performance of livestock. In spite of these innovations, the agricultural industry still faces significant challenges in producing enough food and getting it safely to market. These include changing weather patterns, water shortages, urbanization, population growth, complex environmental regulations, and dwindling available agricultural land, among others. In addition, food waste is a significant drain on the global food supply.

This paper presents a new insight into improving the performance of Stochastic Neighbour Embedding (t-SNE) by using Isolation kernel instead of Gaussian kernel. Isolation kernel outperforms Gaussian kernel in two aspects. First, the use of Isolation kernel in t-SNE overcomes the drawback of misrepresenting some structures in the data, which often occurs when Gaussian kernel is applied in t-SNE. This is because Gaussian kernel determines each local bandwidth based on one local point only, while Isolation kernel is derived directly from the data based on space partitioning. Second, the use of Isolation kernel yields a more efficient similarity computation because data-dependent Isolation kernel has only one parameter that needs to be tuned. In contrast, the use of data-independent Gaussian kernel increases the computational cost by determining n bandwidths for a dataset of n points. As the root cause of these deficiencies in t-SNE is Gaussian kernel, we show that simply replacing Gaussian kernel with Isolation kernel in t-SNE significantly improves the quality of the final visualisation output (without creating misrepresented structures) and removes one key obstacle that prevents t-SNE from processing large datasets. Moreover, Isolation kernel enables t-SNE to deal with large-scale datasets in less runtime without trading off accuracy, unlike existing methods in speeding up t-SNE.

The null space of the $k$-th order Laplacian $\mathbf{\mathcal L}_k$, known as the {\em $k$-th homology vector space}, encodes the non-trivial topology of a manifold or a network. Understanding the structure of the homology embedding can thus disclose geometric or topological information from the data. The study of the null space embedding of the graph Laplacian $\mathbf{\mathcal L}_0$ has spurred new research and applications, such as spectral clustering algorithms with theoretical guarantees and estimators of the Stochastic Block Model. In this work, we investigate the geometry of the $k$-th homology embedding and focus on cases reminiscent of spectral clustering. Namely, we analyze the {\em connected sum} of manifolds as a perturbation to the direct sum of their homology embeddings. We propose an algorithm to factorize the homology embedding into subspaces corresponding to a manifold's simplest topological components. The proposed framework is applied to the {\em shortest homologous loop detection} problem, a problem known to be NP-hard in general. Our spectral loop detection algorithm scales better than existing methods and is effective on diverse data such as point clouds and images.

The study of complex systems through the lens of category theory consistently proves to be a powerful approach. We propose that cognition deserves the same category-theoretic treatment. We show that by considering a highly-compact cognitive system, there are fundamental physical trade-offs resulting in a utility problem. We then examine how to do this systematically, and propose some requirements for "cognitive categories", before investigating the phenomenona of topological defects in gauge fields over conceptual spaces.

These days, due to the increasing amount of information generated on the web, most web service providers try to personalize their services. Users also interact with web-based systems in multiple ways and state their interests and preferences by rating the provided items. This paper proposes a framework to predict users' demographic based on ratings registered by users in a system. To the best of our knowledge, this is the first time that the item ratings are employed for users' demographic prediction problems, which have extensively been studied in recommendation systems and service personalization. We apply the framework to the Movielens dataset's ratings and predict users' age and gender. The experimental results show that using all ratings registered by users improves the prediction accuracy by at least 16% compared with previously studied models. Moreover, by classifying the items as popular and unpopular, we eliminate ratings that belong to 95% of items and still reach an acceptable level of accuracy. This significantly reduces update costs in a time-varying environment. Besides this classification, we propose other methods to reduce data volume while keeping the predictions accurate.

Natural language to SQL (NL2SQL) aims to parse a natural language with a given database into a SQL query, which widely appears in practical Internet applications. Jointly encode database schema and question utterance is a difficult but important task in NL2SQL. One solution is to treat the input as a heterogeneous graph. However, it failed to learn good word representation in question utterance. Learning better word representation is important for constructing a well-designed NL2SQL system. To solve the challenging task, we present a Relation aware Semi-autogressive Semantic Parsing (\MODN) ~framework, which is more adaptable for NL2SQL. It first learns relation embedding over the schema entities and question words with predefined schema relations with ELECTRA and relation aware transformer layer as backbone. Then we decode the query SQL with a semi-autoregressive parser and predefined SQL syntax. From empirical results and case study, our model shows its effectiveness in learning better word representation in NL2SQL.