Many resource management problems require sequential decision-making under uncertainty, where the only uncertainty affecting the decision outcomes are exogenous variables outside the control of the decision-maker. We model these problems as Exo-MDPs (Markov Decision Processes with Exogenous Inputs) and design a class of data-efficient algorithms for them termed Hindsight Learning (HL). Our HL algorithms achieve data efficiency by leveraging a key insight: having samples of the exogenous variables, past decisions can be revisited in hindsight to infer counterfactual consequences that can accelerate policy improvements. We compare HL against classic baselines in the multi-secretary and airline revenue management problems. We also scale our algorithms to a business-critical cloud resource management problem -- allocating Virtual Machines (VMs) to physical machines, and simulate their performance with real datasets from a large public cloud provider. We find that HL algorithms outperform domain-specific heuristics, as well as state-of-the-art reinforcement learning methods.
相關內容
In computational social choice, the distortion of a voting rule quantifies the degree to which the rule overcomes limited preference information to select a socially desirable outcome. This concept has been investigated extensively, but only through a worst-case lens. Instead, we study the expected distortion of voting rules with respect to an underlying distribution over voter utilities. Our main contribution is the design and analysis of a novel and intuitive rule, binomial voting, which provides strong distribution-independent guarantees for both expected distortion and expected welfare.
We address the problem of parameter estimation for degenerate diffusion processes defined via the solution of Stochastic Differential Equations (SDEs) with diffusion matrix that is not full-rank. For this class of hypo-elliptic diffusions recent works have proposed contrast estimators that are asymptotically normal, provided that the step-size in-between observations $\Delta=\Delta_n$ and their total number $n$ satisfy $n \to \infty$, $n \Delta_n \to \infty$, $\Delta_n \to 0$, and additionally $\Delta_n = o (n^{-1/2})$. This latter restriction places a requirement for a so-called `rapidly increasing experimental design'. In this paper, we overcome this limitation and develop a general contrast estimator satisfying asymptotic normality under the weaker design condition $\Delta_n = o(n^{-1/p})$ for general $p \ge 2$. Such a result has been obtained for elliptic SDEs in the literature, but its derivation in a hypo-elliptic setting is highly non-trivial. We provide numerical results to illustrate the advantages of the developed theory.
We consider the problem of quantifying uncertainty over expected cumulative rewards in model-based reinforcement learning. In particular, we focus on characterizing the variance over values induced by a distribution over MDPs. Previous work upper bounds the posterior variance over values by solving a so-called uncertainty Bellman equation (UBE), but the over-approximation may result in inefficient exploration. We propose a new UBE whose solution converges to the true posterior variance over values and leads to lower regret in tabular exploration problems. We identify challenges to apply the UBE theory beyond tabular problems and propose a suitable approximation. Based on this approximation, we introduce a general-purpose policy optimization algorithm, Q-Uncertainty Soft Actor-Critic (QU-SAC), that can be applied for either risk-seeking or risk-averse policy optimization with minimal changes. Experiments in both online and offline RL demonstrate improved performance compared to other uncertainty estimation methods.
In the mentor-student communication process, students often struggle to receive prompt and clear guidance from their mentors, making it challenging to determine their next steps. When mentors don't respond promptly, it can lead to student confusion, as they may be uncertain whether their message has been acknowledged without resulting action. Instead of the binary options of "read" and "unread," there's a pressing need for more nuanced descriptions of message states. To tackle this ambiguity, we've developed a set of symbols to precisely represent the cognitive states associated with messages in transit. Through experimentation, this design not only assists mentors and students in effectively labeling their responses but also mitigates unnecessary misunderstandings. By utilizing symbols for accurate information and understanding state marking, we've enhanced communication efficiency between mentors and students, thereby improving the quality and efficacy of communication in mentor-student relationships.
We consider a distributed coding for computing problem with constant decoding locality, i.e. with a vanishing error probability, any single sample of the function can be approximately recovered by probing only constant number of compressed bits. We establish an achievable rate region by designing an efficient coding scheme. The scheme reduces the required rate by introducing auxiliary random variables and supports local decoding at the same time. Then we show the rate region is optimal under mild regularity conditions on source distributions. A coding for computing problem with side information is analogously studied. These results indicate that more rate has to be taken in order to achieve lower coding complexity in distributed computing settings. Moreover, useful graph characterizations are developed to simplify the computation of the achievable rate region.
There has been a lot of work in question generation where different methods to provide target answers as input, have been employed. This experimentation has been mostly carried out for RNN based models. We use three different methods and their combinations for incorporating answer information and explore their effect on several automatic evaluation metrics. The methods that are used are answer prompting, using a custom product method using answer embeddings and encoder outputs, choosing sentences from the input paragraph that have answer related information, and using a separate cross-attention attention block in the decoder which attends to the answer. We observe that answer prompting without any additional modes obtains the best scores across rouge, meteor scores. Additionally, we use a custom metric to calculate how many of the generated questions have the same answer, as the answer which is used to generate them.
AI recommender systems are sought for decision support by providing suggestions to operators responsible for making final decisions. However, these systems are typically considered black boxes, and are often presented without any context or insight into the underlying algorithm. As a result, recommender systems can lead to miscalibrated user reliance and decreased situation awareness. Recent work has focused on improving the transparency of recommender systems in various ways such as improving the recommender's analysis and visualization of the figures of merit, providing explanations for the recommender's decision, as well as improving user training or calibrating user trust. In this paper, we introduce an alternative transparency technique of structuring the order in which contextual information and the recommender's decision are shown to the human operator. This technique is designed to improve the operator's situation awareness and therefore the shared situation awareness between the operator and the recommender system. This paper presents the results of a two-phase between-subjects study in which participants and a recommender system jointly make a high-stakes decision. We varied the amount of contextual information the participant had, the assessment technique of the figures of merit, and the reliability of the recommender system. We found that providing contextual information upfront improves the team's shared situation awareness by improving the human decision maker's initial and final judgment, as well as their ability to discern the recommender's error boundary. Additionally, this technique accurately calibrated the human operator's trust in the recommender. This work proposes and validates a way to provide model-agnostic transparency into AI systems that can support the human decision maker and lead to improved team performance.
Emotion recognition in conversation (ERC) aims to detect the emotion label for each utterance. Motivated by recent studies which have proven that feeding training examples in a meaningful order rather than considering them randomly can boost the performance of models, we propose an ERC-oriented hybrid curriculum learning framework. Our framework consists of two curricula: (1) conversation-level curriculum (CC); and (2) utterance-level curriculum (UC). In CC, we construct a difficulty measurer based on "emotion shift" frequency within a conversation, then the conversations are scheduled in an "easy to hard" schema according to the difficulty score returned by the difficulty measurer. For UC, it is implemented from an emotion-similarity perspective, which progressively strengthens the model's ability in identifying the confusing emotions. With the proposed model-agnostic hybrid curriculum learning strategy, we observe significant performance boosts over a wide range of existing ERC models and we are able to achieve new state-of-the-art results on four public ERC datasets.
Multi-relation Question Answering is a challenging task, due to the requirement of elaborated analysis on questions and reasoning over multiple fact triples in knowledge base. In this paper, we present a novel model called Interpretable Reasoning Network that employs an interpretable, hop-by-hop reasoning process for question answering. The model dynamically decides which part of an input question should be analyzed at each hop; predicts a relation that corresponds to the current parsed results; utilizes the predicted relation to update the question representation and the state of the reasoning process; and then drives the next-hop reasoning. Experiments show that our model yields state-of-the-art results on two datasets. More interestingly, the model can offer traceable and observable intermediate predictions for reasoning analysis and failure diagnosis.
Detecting carried objects is one of the requirements for developing systems to reason about activities involving people and objects. We present an approach to detect carried objects from a single video frame with a novel method that incorporates features from multiple scales. Initially, a foreground mask in a video frame is segmented into multi-scale superpixels. Then the human-like regions in the segmented area are identified by matching a set of extracted features from superpixels against learned features in a codebook. A carried object probability map is generated using the complement of the matching probabilities of superpixels to human-like regions and background information. A group of superpixels with high carried object probability and strong edge support is then merged to obtain the shape of the carried object. We applied our method to two challenging datasets, and results show that our method is competitive with or better than the state-of-the-art.
小貼士
登錄享
注冊地址： 北京市海淀區羊坊店路18號2幢3層301-191