Modern statistical applications often involve minimizing an objective function that may be nonsmooth and/or nonconvex. This paper focuses on a broad Bregman-surrogate algorithm framework including the local linear approximation, mirror descent, iterative thresholding, DC programming and many others as particular instances. The recharacterization via generalized Bregman functions enables us to construct suitable error measures and establish global convergence rates for nonconvex and nonsmooth objectives in possibly high dimensions. For sparse learning problems with a composite objective, under some regularity conditions, the obtained estimators as the surrogate's fixed points, though not necessarily local minimizers, enjoy provable statistical guarantees, and the sequence of iterates can be shown to approach the statistical truth within the desired accuracy geometrically fast. The paper also studies how to design adaptive momentum based accelerations without assuming convexity or smoothness by carefully controlling stepsize and relaxation parameters.

Data augmentation has been an important ingredient for boosting performances of learned models. Prior data augmentation methods for few-shot text classification have led to great performance boosts. However, they have not been designed to capture the intricate compositional structure of natural language. As a result, they fail to generate samples with plausible and diverse sentence structures. Motivated by this, we present the data Augmentation using Lexicalized Probabilistic context-free grammars (ALP) that generates augmented samples with diverse syntactic structures with plausible grammar. The lexicalized PCFG parse trees consider both the constituents and dependencies to produce a syntactic frame that maximizes a variety of word choices in a syntactically preservable manner without specific domain experts. Experiments on few-shot text classification tasks demonstrate that ALP enhances many state-of-the-art classification methods. As a second contribution, we delve into the train-val splitting methodologies when a data augmentation method comes into play. We argue empirically that the traditional splitting of training and validation sets is sub-optimal compared to our novel augmentation-based splitting strategies that further expand the training split with the same number of labeled data. Taken together, our contributions on the data augmentation strategies yield a strong training recipe for few-shot text classification tasks.

Explanation of an AI agent requires knowledge of its design and operation. An open question is how to identify, access and use this design knowledge for generating explanations. Many AI agents used in practice, such as intelligent tutoring systems fielded in educational contexts, typically come with a User Guide that explains what the agent does, how it works and how to use the agent. However, few humans actually read the User Guide in detail. Instead, most users seek answers to their questions on demand. In this paper, we describe a question answering agent (AskJill) that uses the User Guide for an interactive learning environment (VERA) to automatically answer questions and thereby explains the domain, functioning, and operation of VERA. We present a preliminary assessment of AskJill in VERA.

Few-shot classification tasks aim to classify images in query sets based on only a few labeled examples in support sets. Most studies usually assume that each image in a task has a single and unique class association. Under these assumptions, these algorithms may not be able to identify the proper class assignment when there is no exact matching between support and query classes. For example, given a few images of lions, bikes, and apples to classify a tiger. However, in a more general setting, we could consider the higher-level concept of large carnivores to match the tiger to the lion for semantic classification. Existing studies rarely considered this situation due to the incompatibility of label-based supervision with complex conception relationships. In this work, we advanced the few-shot learning towards this more challenging scenario, the semantic-based few-shot learning, and proposed a method to address the paradigm by capturing the inner semantic relationships using interactive psychometric learning. We evaluate our method on the CIFAR-100 dataset. The results show the merits of our proposed method.

The lifelong learning paradigm in machine learning is an attractive alternative to the more prominent isolated learning scheme not only due to its resemblance to biological learning, but also its potential to reduce energy waste by obviating excessive model re-training. A key challenge to this paradigm is the phenomenon of catastrophic forgetting. With the increasing popularity and success of pre-trained models in machine learning, we pose the question: What role does pre-training play in lifelong learning, specifically with respect to catastrophic forgetting? We investigate existing methods in the context of large, pre-trained models and evaluate their performance on a variety of text and image classification tasks, including a large-scale study using a novel dataset of 15 diverse NLP tasks. Across all settings, we observe that generic pre-training implicitly alleviates the effects of catastrophic forgetting when learning multiple tasks sequentially compared to randomly initialized models. We then further investigate why pre-training alleviates forgetting in this setting. We study this phenomenon by analyzing the loss landscape, finding that pre-trained weights appear to ease forgetting by leading to wider minima. Based on this insight, we propose jointly optimizing for current task loss and loss basin sharpness in order to explicitly encourage wider basins during sequential fine-tuning. We show that this optimization approach leads to performance comparable to the state-of-the-art in task-sequential continual learning across multiple settings, without retaining a memory that scales in size with the number of tasks. The contemporary machine learning paradigm concentrates on isolated learning (Chen & Liu, 2018) i.e., learning a model from scratch for every new task. In contrast, the lifelong learning (LL) paradigm (Thrun, 1996) defines a biologically-inspired learning approach where models learn tasks in sequence, ideally preserving past knowledge and leveraging it to efficiently learn new tasks. LL has the added benefit of avoiding periodical re-training of models from scratch to learn novel tasks or adapt to new data, with the potential to reduce both computational and energy requirements (Hazelwood et al., 2018; Strubell et al., 2019; Schwartz et al., 2020). In the context of modern machine learning where state-of-the-art models are powered by deep neural networks, catastrophic forgetting has been identified as a key challenge to implementing successful LL systems (McCloskey & Cohen, 1989; French, 1999). Catastrophic forgetting happens when the model forgets knowledge learned in previous tasks as information relevant to the current task is incorporated.

The use of deep neural networks as function approximators has led to striking progress for reinforcement learning algorithms and applications. Yet the knowledge we have on decision boundary geometry and the loss landscape of neural policies is still quite limited. In this paper we propose a framework to investigate the decision boundary and loss landscape similarities across states and across MDPs. We conduct experiments in various games from Arcade Learning Environment, and discover that high sensitivity directions for neural policies are correlated across MDPs. We argue that these high sensitivity directions support the hypothesis that non-robust features are shared across training environments of reinforcement learning agents. We believe our results reveal fundamental properties of the environments used in deep reinforcement learning training, and represent a tangible step towards building robust and reliable deep reinforcement learning agents.

To what extent do language models (LMs) build "mental models" of a scene when answering situated questions (e.g., questions about a specific ethical dilemma)? While cognitive science has shown that mental models play a fundamental role in human problem-solving, it is unclear whether the high question-answering performance of existing LMs is backed by similar model building - and if not, whether that can explain their well-known catastrophic failures. We observed that Macaw, an existing T5-based LM, when probed provides somewhat useful but inadequate mental models for situational questions (estimated accuracy=43%, usefulness=21%, consistency=42%). We propose DREAM, a model that takes a situational question as input to produce a mental model elaborating the situation, without any additional task specific training data for mental models. It inherits its social commonsense through distant supervision from existing NLP resources. Our analysis shows that DREAM can produce significantly better mental models (estimated accuracy=67%, usefulness=37%, consistency=71%) compared to Macaw. Finally, mental models generated by DREAM can be used as additional context for situational QA tasks. This additional context improves the answer accuracy of a Macaw zero-shot model by between +1% and +4% (absolute) on three different datasets.

Chinese TV will introduce the first AI sign language presenter in time for the 2022 Winter Olympics in Beijing. China Central Television (CCTV) and Baidu AI Cloud said the launch of the AI sign language presenter represents a huge leap forwards in'overcoming the barrier of sound with technology'. Nearly 28 million people in China are hearing impaired and about 430 million around the world also suffer from hearing loss. The launch of the AI presenter will allow the state broadcaster to include sign language services for viewers around the clock, and will start by giving updates of the Winter Olympics in Beijing early next year. The presenter achieves high-level sign language expression thanks to Baidu's natural action engine and their sign language translation engine.

Why do we mistake computer generated poems as the work of humans? From Wordsworth's "spontaneous overflow of powerful feelings" to Eavan Boland's "figure in which secret things confide," poetry is often defined by -- and extolled for -- its ability to convey human emotion. What, then, does it mean that we can not distinguish poems penned by humans from those generated by machine? Indeed, researchers Nils Köbis and Luca D.Mossink at the University of Amsterdam have found that humans cannot tell AI-generated poems from those written by amateur poets, or by well known professionals, provided a human selects the best poem from a set of machine-generated verses to compare. Have machines become as talented as our poets?

Having machines turn text into speech is nothing new. Professor Stephen Hawking communicated with a computerized voice for many years, and by now, we're used to our GPS devices or smart speakers asking questions and responding to our queries. What is different these days is that the quality of synthesized speech is improving, thanks to several companies using AI to create voice skins for enterprise companies and content creators that give more options for turning text into speech. LOVO, an AI voice and synthetic speech startup company, uses a voiceover API to turn text into speech in real-time using 200 human-like voices in 33 languages using their "voice library." Users also can clone their own voices to create their own skins, simply by reading 15 minutes of a script.