Today: Accelerating Medical and Drug Discovery To be sure, the quality of these tools and their promise varies, but given there is already reliance on AI to accelerate disease insights and drug development, we should be accelerating work on guardrails too. Among the bigger governance challenges is assessing the reliability, safety, and fairness of such tools. In many instances, a well-designed audit framework would enable leaders to evaluate whether the system is trustworthy but this can be challenging in practice. Tomorrow: Enabling Population Management and Limiting Disease Spread Tracking COVID-19 patients and their contacts is widely understood to be central to the effort to contain spread and eventually, loosen the restrictions on our mobility and economic activity. The AI/ML-enabled contact tracing tools that were so effective in Singapore and South Korea are now, in some form, likely coming to the EU and US.

From where I'm standing, at the intersection between AI and eCommerce, I can see opportunities for businesses to better serve their customers – and themselves – in ways we never thought possible. The road to them, though, is not a straight line to success. For those willing to patiently make incremental progress, there's an $80 billion reward at the end. Here's where you can find these opportunities for growth. Worldwide eCommerce sales topped $3.5 trillion USD in 2019, an increase of approximately 18% from the year before.

How well do you know the greatest people who contributed to the field of Artificial Intelligence? This is the focus of our 10th crossword in the series. This one focuses on your knowledge of the greatest minds who contributed to the field of AI. Please use the full name of the person without any space. Do give it a try and let us know your feedback.

We study the problem of learning exploration-exploitation strategies that effectively adapt to dynamic environments, where the task may change over time. While RNN-based policies could in principle represent such strategies, in practice their training time is prohibitive and the learning process often converges to poor solutions. In this paper, we consider the case where the agent has access to a description of the task (e.g., a task id or task parameters) at training time, but not at test time. We propose a novel algorithm that regularizes the training of an RNN-based policy using informed policies trained to maximize the reward in each task. This dramatically reduces the sample complexity of training RNN-based policies, without losing their representational power. As a result, our method learns exploration strategies that efficiently balance between gathering information about the unknown and changing task and maximizing the reward over time. We test the performance of our algorithm in a variety of environments where tasks may vary within each episode.

At present, recognition of the Bangla handwriting compound character has been an essential issue for many years. In recent years there have been application-based researches in machine learning, and deep learning, which is gained interest, and most notably is handwriting recognition because it has a tremendous application such as Bangla OCR. MatrriVasha, the project which can recognize Bangla, handwritten several compound characters. Currently, compound character recognition is an important topic due to its variant application, and helps to create old forms, and information digitization with reliability. But unfortunately, there is a lack of a comprehensive dataset that can categorize all types of Bangla compound characters. MatrriVasha is an attempt to align compound character, and it's challenging because each person has a unique style of writing shapes. After all, MatrriVasha has proposed a dataset that intends to recognize Bangla 120(one hundred twenty) compound characters that consist of 2552(two thousand five hundred fifty-two) isolated handwritten characters written unique writers which were collected from within Bangladesh. This dataset faced problems in terms of the district, age, and gender-based written related research because the samples were collected that includes a verity of the district, age group, and the equal number of males, and females. As of now, our proposed dataset is so far the most extensive dataset for Bangla compound characters. It is intended to frame the acknowledgment technique for handwritten Bangla compound character. In the future, this dataset will be made publicly available to help to widen the research.

Current explanation techniques towards a transparent Convolutional Neural Network (CNN) mainly focuses on building connections between the human-understandable input features with models' prediction, overlooking an alternative representation of the input, the frequency components decomposition. In this work, we present an analysis of the connection between the distribution of frequency components in the input dataset and the reasoning process the model learns from the data. We further provide quantification analysis about the contribution of different frequency components toward the model's prediction. We show that the vulnerability of the model against tiny distortions is a result of the model is relying on the high-frequency features, the target features of the adversarial (black and white-box) attackers, to make the prediction. We further show that if the model develops stronger association between the low-frequency component with true labels, the model is more robust, which is the explanation of why adversarially trained models are more robust against tiny distortions.

This paper proposes a detailed and extensive comparison of the Trust Region Policy Optimization and DeepQ-Network with Normalized Advantage Functions with respect to other state of the art algorithms, namely Deep Deterministic Policy Gradient and Vanilla Policy Gradient. Comparisons demonstrate that the former have better performances then the latter when asking robotic arms to accomplish manipulation tasks such as reaching a random target pose and pick &placing an object. Both simulated and real-world experiments are provided. Simulation lets us show the procedures that we adopted to precisely estimate the algorithms hyper-parameters and to correctly design good policies. Real-world experiments let show that our polices, if correctly trained on simulation, can be transferred and executed in a real environment with almost no changes.

Reducing domain divergence is a key step in transfer learning problems. Existing works focus on the minimization of global domain divergence. However, two domains may consist of several shared subdomains, and differ from each other in each subdomain. In this paper, we take the local divergence of subdomains into account in transfer. Specifically, we propose to use low-dimensional manifold to represent subdomain, and align the local data distribution discrepancy in each manifold across domains. A Manifold Maximum Mean Discrepancy (M3D) is developed to measure the local distribution discrepancy in each manifold. We then propose a general framework, called Transfer with Manifolds Discrepancy Alignment (TMDA), to couple the discovery of data manifolds with the minimization of M3D. We instantiate TMDA in the subspace learning case considering both the linear and nonlinear mappings. We also instantiate TMDA in the deep learning framework. Extensive experimental studies demonstrate that TMDA is a promising method for various transfer learning tasks.

Model Stealing (MS) attacks allow an adversary with black-box access to a Machine Learning model to replicate its functionality, compromising the confidentiality of the model. Such attacks train a clone model by using the predictions of the target model for different inputs. The effectiveness of such attacks relies heavily on the availability of data necessary to query the target model. Existing attacks either assume partial access to the dataset of the target model or availability of an alternate dataset with semantic similarities. This paper proposes MAZE -- a data-free model stealing attack using zeroth-order gradient estimation. In contrast to prior works, MAZE does not require any data and instead creates synthetic data using a generative model. Inspired by recent works in data-free Knowledge Distillation (KD), we train the generative model using a disagreement objective to produce inputs that maximize disagreement between the clone and the target model. However, unlike the white-box setting of KD, where the gradient information is available, training a generator for model stealing requires performing black-box optimization, as it involves accessing the target model under attack. MAZE relies on zeroth-order gradient estimation to perform this optimization and enables a highly accurate MS attack. Our evaluation with four datasets shows that MAZE provides a normalized clone accuracy in the range of 0.91x to 0.99x, and outperforms even the recent attacks that rely on partial data (JBDA, clone accuracy 0.13x to 0.69x) and surrogate data (KnockoffNets, clone accuracy 0.52x to 0.97x). We also study an extension of MAZE in the partial-data setting and develop MAZE-PD, which generates synthetic data closer to the target distribution. MAZE-PD further improves the clone accuracy (0.97x to 1.0x) and reduces the query required for the attack by 2x-24x.

Federated learning (FL) is a well established method for performing machine learning tasks over massively distributed data. However in settings where data is distributed in a non-iid (not independent and identically distributed) fashion -- as is typical in real world situations -- the joint model produced by FL suffers in terms of test set accuracy and/or communication costs compared to training on iid data. We show that learning a single joint model is often not optimal in the presence of certain types of non-iid data. In this work we present a modification to FL by introducing a hierarchical clustering step (FL+HC) to separate clusters of clients by the similarity of their local updates to the global joint model. Once separated, the clusters are trained independently and in parallel on specialised models. We present a robust empirical analysis of the hyperparameters for FL+HC for several iid and non-iid settings. We show how FL+HC allows model training to converge in fewer communication rounds (significantly so under some non-iid settings) compared to FL without clustering. Additionally, FL+HC allows for a greater percentage of clients to reach a target accuracy compared to standard FL. Finally we make suggestions for good default hyperparameters to promote superior performing specialised models without modifying the the underlying federated learning communication protocol.