Machine Learning has seen tremendous growth recently, which has led to a larger adoption of ML systems for educational assessments, credit risk, healthcare, employment, criminal justice, to name a few. Trustworthiness of ML and NLP systems is a crucial aspect and requires guarantee that the decisions they make are fair and robust. Aligned with this, we propose a framework GYC, to generate a set of counterfactual text samples, which are crucial for testing these ML systems. Our main contributions include a) We introduce GYC, a framework to generate counterfactual samples such that the generation is plausible, diverse, goal-oriented, and effective, b) We generate counterfactual samples, that can direct the generation towards a corresponding condition such as named-entity tag, semantic role label, or sentiment. Our experimental results on various domains show that GYC generates counterfactual text samples exhibiting the above four properties. %The generated counterfactuals can then be fed complementary to the existing data augmentation for improving the debiasing algorithms performance as compared to existing counterfactuals generated by token substitution. GYC generates counterfactuals that can act as test cases to evaluate a model and any text debiasing algorithm.

We put forward a video dataset with 5k+ facial bounding box annotations across a troop of 7 western lowland gorillas at Bristol Zoo Gardens. Training on this dataset, we implement and evaluate a standard deep learning pipeline on the task of facially recognising individual gorillas in a zoo environment. We show that a basic YOLOv3-powered application is able to perform identifications at 92% mAP when utilising single frames only. Tracking-by-detection-association and identity voting across short tracklets yields an improved robust performance of 97% mAP. To facilitate easy utilisation for enriching the research capabilities of zoo environments, we publish the code, video dataset, weights, and ground-truth annotations at data.bris.ac.uk.

Recent advances in self-supervised learning (SSL) have largely closed the gap with supervised ImageNet pretraining. Despite their success these methods have been primarily applied to unlabeled ImageNet images, and show marginal gains when trained on larger sets of uncurated images. We hypothesize that current SSL methods perform best on iconic images, and struggle on complex scene images with many objects. Analyzing contrastive SSL methods shows that they have poor visual grounding and receive poor supervisory signal when trained on scene images. We propose Contrastive Attention-Supervised Tuning(CAST) to overcome these limitations. CAST uses unsupervised saliency maps to intelligently sample crops, and to provide grounding supervision via a Grad-CAM attention loss. Experiments on COCO show that CAST significantly improves the features learned by SSL methods on scene images, and further experiments show that CAST-trained models are more robust to changes in backgrounds.

In 2016, the Defense Advanced Research Projects Agency (DARPA) hosted the Cyber Grand Challenge (Song & Alves-Foss, 2015), a competition which invited participating finalist teams to develop automated cyber defense systems that can self-discover, prove, and correct software vulnerabilities at real-time - without human intervention. For the first time, the world witnessed hackers being automated at scale, i.e. cyber autonomy (Brumley, 2018). As the competition progressed, the systems were not only able to auto-detect and correct their software, but also able to attack other systems (other participants' machines) in the network. Even though the competition did not catch much mainstream media attention, the DARPA Cyber Grand Challenge proved the feasibility of cyber autonomy, stretched the imagination of the national and cyber security industries and created a mix of perceptions ranging from hope to fear - the hope of increasingly secure computing systems at scale, and the fear of current jobs such as penetration testing being automated.

To counter the challenge associated with noise labels, the learning strategy of deep neural networks must be differentiated over the learning period during the training process. Therefore, we propose a novel two-phase learning method, MORPH, which automatically transitions its learning phase at the point when the network begins to rapidly memorize false-labeled samples. In the first phase, MORPH starts to update the network for all the training samples before the transition point. Without any supervision, the learning phase is converted to the next phase on the basis of the estimated best transition point. Subsequently, MORPH resumes the training of the network only for a maximal safe set, which maintains the collection of almost certainly true-labeled samples at each epoch. Owing to its two-phase learning, MORPH realizes noise-free training for any type of label noise for practical use. Moreover, extensive experiments using six datasets verify that MORPH significantly outperforms five state-of-the art methods in terms of test error and training time.

Most standard learning approaches lead to fragile models which are prone to drift when sequentially trained on samples of a different nature - the well-known "catastrophic forgetting" issue. In particular, when a model consecutively learns from different visual domains, it tends to forget the past ones in favor of the most recent. In this context, we show that one way to learn models that are inherently more robust against forgetting is domain randomization - for vision tasks, randomizing the current domain's distribution with heavy image manipulations. Building on this result, we devise a meta-learning strategy where a regularizer explicitly penalizes any loss associated with transferring the model from the current domain to different "auxiliary" meta-domains, while also easing adaptation to them. Such meta-domains, are also generated through randomized image manipulations. We empirically demonstrate in a variety of experiments - spanning from classification to semantic segmentation - that our approach results in models that are less prone to catastrophic forgetting when transferred to new domains.

Recent advances in Artificial Intelligence and deep learning have revived the interest in studying the gap between the reasoning capabilities of humans and machines. In this ongoing work, we introduce CRAFT, a new visual question answering dataset that requires causal reasoning about physical forces and object interactions. It contains 38K video and question pairs that are generated from 3K videos from 10 different virtual environments, containing different number of objects in motion that interact with each other. Two question categories from CRAFT include previously studied descriptive and counterfactual questions. Besides, inspired by the theory of force dynamics from the field of human cognitive psychology, we introduce new question categories that involve understanding the intentions of objects through the notions of cause, enable, and prevent. Our preliminary results demonstrate that even though these tasks are very intuitive for humans, the implemented baselines could not cope with the underlying challenges.

Recent advances in Artificial Intelligence (AI), especially in Machine Learning (ML), have introduced various practical applications (e.g., virtual personal assistants and autonomous cars) that enhance the experience of everyday users. However, modern ML technologies like Deep Learning require considerable technical expertise and resources to develop, train and deploy such models, making effective reuse of the ML models a necessity. Such discovery and reuse by practitioners and researchers are being addressed by public ML package repositories, which bundle up pre-trained models into packages for publication. Since such repositories are a recent phenomenon, there is no empirical data on their current state and challenges. Hence, this paper conducts an exploratory study that analyzes the structure and contents of two popular ML package repositories, TFHub and PyTorch Hub, comparing their information elements (features and policies), package organization, package manager functionalities and usage contexts against popular software package repositories (npm, PyPI, and CRAN). Through these studies, we have identified unique SE practices and challenges for sharing ML packages. These findings and implications would be useful for data scientists, researchers and software developers who intend to use these shared ML packages.

This paper proposes a end-to-end deep network to recognize kinds of accents under the same language, where we develop and transfer the deep architecture in speaker-recognition area to accent classification task for learning utterance-level accent representation. Compared with the individual-level feature in speaker-recognition, accent recognition throws a more challenging issue in acquiring compact group-level features for the speakers with the same accent, hence a good discriminative accent feature space is desired. Our deep framework adopts multitask-learning mechanism and mainly consists of three modules: a shared CNNs and RNNs based front-end encoder, a core accent recognition branch, and an auxiliary speech recognition branch, where we take speech spectrogram as input. More specifically, with the sequential descriptors learned from a shared encoder, the accent recognition branch first condenses all descriptors into an embedding vector, and then explores different discriminative loss functions which are popular in face recognition domain to enhance embedding discrimination. Additionally, due to the accent is a speaking-related timbre, adding speech recognition branch effectively curbs the over-fitting phenomenon in accent recognition during training. We show that our network without any data-augment preproccessings is significantly ahead of the baseline system on the accent classification track in the Accented English Speech Recognition Challenge 2020 (AESRC2020), where the state-of-the-art loss function Circle-Loss achieves the best discriminative optimization for accent representation.

Downscaling is necessary to generate high-resolution observation data to validate the climate model forecast or monitor rainfall at the micro-regional level operationally. Dynamical and statistical downscaling models are often used to get information at high-resolution gridded data over larger domains. As rainfall variability is dependent on the complex Spatio-temporal process leading to non-linear or chaotic Spatio-temporal variations, no single downscaling method can be considered efficient enough. In data with complex topographies, quasi-periodicities, and non-linearities, deep Learning (DL) based methods provide an efficient solution in downscaling rainfall data for regional climate forecasting and real-time rainfall observation data at high spatial resolutions. In this work, we employed three deep learning-based algorithms derived from the super-resolution convolutional neural network (SRCNN) methods, to precipitation data, in particular, IMD and TRMM data to produce 4x-times high-resolution downscaled rainfall data during the summer monsoon season. Among the three algorithms, namely SRCNN, stacked SRCNN, and DeepSD, employed here, the best spatial distribution of rainfall amplitude and minimum root-mean-square error is produced by DeepSD based downscaling. Hence, the use of the DeepSD algorithm is advocated for future use. We found that spatial discontinuity in amplitude and intensity rainfall patterns is the main obstacle in the downscaling of precipitation. Furthermore, we applied these methods for model data postprocessing, in particular, ERA5 data. Downscaled ERA5 rainfall data show a much better distribution of spatial covariance and temporal variance when compared with observation.