Automatic Melody Harmonization via Reinforcement Learning by Exploring Structured Representations for Melody Sequences

We present a novel reinforcement learning architecture that learns a structured representation for use in symbolic melody harmonization. Probabilistic models are predominant in melody harmonization tasks, most of which only treat melody notes as independent observations and do not take note of substructures in the melodic sequence. To fill this gap, we add substructure discovery as a crucial step in automatic chord generation. The proposed method consists of a structured representation module that generates hierarchical structures for the symbolic melodies, a policy module that learns to break a melody into segments (whose boundaries concur with chord changes) and phrases (the subunits in segments), and a harmonization module that generates chord sequences for each segment. We formulate the structure discovery process as a sequential decision problem with a policy gradient RL method selecting the boundary of each segment or phrase to obtain an optimized structure. We conduct experiments on our preprocessed HookTheory Lead Sheet Dataset, which has 17,979 melody/chord pairs. The results demonstrate that our proposed method can learn task-specific representations and, thus, yield competitive results compared with state-of-the-art baselines.

Masked Contrastive Learning for Anomaly Detection

Detecting anomalies is one fundamental aspect of a safety-critical software system, however, it remains a long-standing problem. Numerous branches of works have been proposed to alleviate the complication and have shown promising results. In particular, self-supervised learning based methods are spurring interest due to their capability of learning diverse representations without additional labels. Among self-supervised learning tactics, contrastive learning is one specific framework showing pronounced results in various fields including anomaly detection. However, the primary objective of contrastive learning is to learn task-agnostic features without any labels, which is not entirely suited to discern anomalies. In this paper, we propose a task-specific variant of contrastive learning named masked contrastive learning, which is more befitted for anomaly detection. Moreover, we propose a new inference method dubbed self-ensemble inference that further boosts performance by leveraging the ability learned through auxiliary self-supervision tasks. By combining our models, we can outperform previous state-of-the-art methods by a significant margin on various benchmark datasets.

The New Normal

Adult learners are different from younger learners. Many have taken Knowles' ideas to work with adult learners as if they all are the same. Knowles described adult learners as more self-directed, willing to be responsible for what they do, unwilling to have teachers impose arbitrary information on them, ready to learn, task oriented, and experienced. Prather adds many adults have more immediate goals for their lives and careers. Kenner and Weinerman find adults want more collaborative relationships with professors. Adult students are unique and come to classes from a variety of backgrounds and circumstances, with different needs, wants, and learning preferences. Many students are prepared for college, while others are petrified. In this chapter, the authors argue that instructors should understand their students in order to help them be successful. Students are not alternative students; they are normal, intelligent people who can and will learn. Thus, the goal should be student-centered online learning.

Variational Metric Scaling for Metric-Based Meta-Learning

Metric-based meta-learning has attracted a lot of attention due to its effectiveness and efficiency in few-shot learning. Recent studies show that metric scaling plays a crucial role in the performance of metric-based meta-learning algorithms. However, there still lacks a principled method for learning the metric scaling parameter automatically. In this paper, we recast metric-based meta-learning from a Bayesian perspective and develop a variational metric scaling framework for learning a proper metric scaling parameter. Firstly, we propose a stochastic variational method to learn a single global scaling parameter. To better fit the embedding space to a given data distribution, we extend our method to learn a dimensional scaling vector to transform the embedding space. Furthermore, to learn task-specific embeddings, we generate task-dependent dimensional scaling vectors with amortized variational inference. Our method is end-to-end without any pre-training and can be used as a simple plug-and-play module for existing metric-based meta-algorithms. Experiments on miniImageNet show that our methods can be used to consistently improve the performance of existing metric-based meta-algorithms including prototypical networks and TADAM.

Learning and Information Use in an Intergroup Context

When faced with uncertainty, human observers maximize performance by effectively integrating sensory information with learned task-relevant regularities. Does this behavior similarly occur in social, intergroup settings? Here, by providing participants with repeated exposure to a probabilistic decision-making task involving social information, we characterized their ability to learn task-relevant regularities and explored how social information interacted with learning over time. In particular, we examined how a behavior typically attributed to implicit intergroup bias––taking group membership into account when irrelevant or inappropriate––might be influenced by learning mechanisms that promote efficacy in the face of uncertainty. Across two experiments we show that observers learned and utilized task-relevant regularities to inform their decisions and maximize rewards. Notably, social information was utilized only when doing so conferred reward gains. Furthermore, learning about the utility of social information had a long-term influence on observers’ ability to subsequently learn and utilize other sources of information. Taken together, our findings highlight the powerful influence of learning in intergroup contexts: human observers are sensitive to–and quickly learn–the task-relevance of social information, and are able to use this learned knowledge to flexibly reweight available sources of information.

Young Children are Sensitive to Task Context When Learning Predictive Patterns

Previous research suggests that young children often fail to preferentially learn task-relevant over task-irrelevant information, suggesting they may learn the same thing regardless of the task context. To investigate this hypothesis, two hundred and ninety-five children ranging from four- to eight-years old learned to predict patterns of features (e.g., eyes, noses) of novel faces in four task contexts. Results demonstrate that young children do in fact tailor what they are learning to specific task demands. When tasks required participants to learn a single predictive pattern, they learned that one pattern well but not other equally reliable patterns (e.g., pick mouths given eyes, but not noses given hats). However, when tasks allowed or required attention to multiple patterns, only older children showed evidence of learning any of the patterns. Thus, for young children, focusing on one thing compromises their ability to learn other things, but trying to learn too much at once may mean learning nothing. These results demonstrate tradeoffs between broad and narrow learning, that may be especially severe for younger children.

The New Normal

Adult learners are different from younger learners. Many have taken Knowles' ideas to work with adult learners as if they all are the same. Knowles described adult learners as more self-directed, willing to be responsible for what they do, unwilling to have teachers impose arbitrary information on them, ready to learn, task oriented, and experienced. Prather adds many adults have more immediate goals for their lives and careers. Kenner and Weinerman find adults want more collaborative relationships with professors. Adult students are unique and come to classes from a variety of backgrounds and circumstances, with different needs, wants, and learning preferences. Many students are prepared for college, while others are petrified. In this chapter, the authors argue that instructors should understand their students in order to help them be successful. Students are not alternative students; they are normal, intelligent people who can and will learn. Thus, the goal should be student-centered online learning.