In recent years, researchers have proposed many deep learning (DL) methods for various tasks, and particularly face recognition (FR) made an enormous leap using these techniques. Deep FR systems benefit from the hierarchical architecture of the DL methods to learn discriminative face representation. Therefore, DL techniques significantly improve state-of-the-art performance on FR systems and encourage diverse and efficient real-world applications. In this paper, we present a comprehensive analysis of various FR systems that leverage the different types of DL techniques, and for the study, we summarize 168 recent contributions from this area. We discuss the papers related to different algorithms, architectures, loss functions, activation functions, datasets, challenges, improvement ideas, current and future trends of DL-based FR systems. We provide a detailed discussion of various DL methods to understand the current state-of-the-art, and then we discuss various activation and loss functions for the methods. Additionally, we summarize different datasets used widely for FR tasks and discuss challenges related to illumination, expression, pose variations, and occlusion. Finally, we discuss improvement ideas, current and future trends of FR tasks.

There hardly exists a general solver that is efficient for scheduling problems due to their diversity and complexity. In this study, we develop a two-stage framework, in which reinforcement learning (RL) and traditional operations research (OR) algorithms are combined together to efficiently deal with complex scheduling problems. The scheduling problem is solved in two stages, including a finite Markov decision process (MDP) and a mixed-integer programming process, respectively. This offers a novel and general paradigm that combines RL with OR approaches to solving scheduling problems, which leverages the respective strengths of RL and OR: The MDP narrows down the search space of the original problem through an RL method, while the mixed-integer programming process is settled by an OR algorithm. These two stages are performed iteratively and interactively until the termination criterion has been met. Under this idea, two implementation versions of the combination methods of RL and OR are put forward. The agile Earth observation satellite scheduling problem is selected as an example to demonstrate the effectiveness of the proposed scheduling framework and methods. The convergence and generalization capability of the methods are verified by the performance of training scenarios, while the efficiency and accuracy are tested in 50 untrained scenarios. The results show that the proposed algorithms could stably and efficiently obtain satisfactory scheduling schemes for agile Earth observation satellite scheduling problems. In addition, it can be found that RL-based optimization algorithms have stronger scalability than non-learning algorithms. This work reveals the advantage of combining reinforcement learning methods with heuristic methods or mathematical programming methods for solving complex combinatorial optimization problems.

Policy gradient methods estimate the gradient of a policy objective solely based on either the likelihood ratio (LR) estimator or the reparameterization (RP) estimator for estimating gradients. Many policy gradient methods based on the LR estimator can be unified under the policy gradient theorem (Sutton et al., 2000). However, such a unifying theorem does not exist for policy gradient methods based on the RP estimator. Moreover, no existing method requires and uses both estimators beyond a trivial interpolation between them. In this paper, we provide a theoretical framework that unifies several existing policy gradient methods based on the RP estimator. Utilizing our framework, we introduce a novel strategy to compute the policy gradient that, for the first time, incorporates both the LR and RP estimators and can be unbiased only when both estimators are present. Based on this strategy, we develop a new on-policy algorithm called the Reward Policy Gradient algorithm, which is the first model-free policy gradient method to utilize reward gradients. Using an idealized environment, we show that policy gradient solely based on the RP estimator for rewards are biased even with true rewards whereas our combined estimator is not. Finally, we show that our method either performs comparably with or outperforms Proximal Policy Optimization -- an LR-based on-policy method -- on several continuous control tasks.

Natural Language Processing algorithms have made incredible progress recently, but they still struggle when applied to out-of-distribution examples. In this paper, we address a very challenging and previously underexplored version of this domain adaptation problem. In our setup an algorithm is trained on several source domains, and then applied to examples from an unseen domain that is unknown at training time. Particularly, no examples, labeled or unlabeled, or any other knowledge about the target domain are available to the algorithm at training time. We present PADA: A Prompt-based Autoregressive Domain Adaptation algorithm, based on the T5 model. Given a test example, PADA first generates a unique prompt and then, conditioned on this prompt, labels the example with respect to the NLP task. The prompt is a sequence of unrestricted length, consisting of pre-defined Domain Related Features (DRFs) that characterize each of the source domains. Intuitively, the prompt is a unique signature that maps the test example to the semantic space spanned by the source domains. In experiments with two tasks: Rumour Detection and Multi-Genre Natural Language Inference (MNLI), for a total of 10 multi-source adaptation scenarios, PADA strongly outperforms state-of-the-art approaches and additional strong baselines.

Multi-label classification tasks such as OCR and multi-object recognition are a major focus of the growing machine learning as a service industry. While many multi-label prediction APIs are available, it is challenging for users to decide which API to use for their own data and budget, due to the heterogeneity in those APIs' price and performance. Recent work shows how to select from single-label prediction APIs. However the computation complexity of the previous approach is exponential in the number of labels and hence is not suitable for settings like OCR. In this work, we propose FrugalMCT, a principled framework that adaptively selects the APIs to use for different data in an online fashion while respecting user's budget. The API selection problem is cast as an integer linear program, which we show has a special structure that we leverage to develop an efficient online API selector with strong performance guarantees. We conduct systematic experiments using ML APIs from Google, Microsoft, Amazon, IBM, Tencent and other providers for tasks including multi-label image classification, scene text recognition and named entity recognition. Across diverse tasks, FrugalMCT can achieve over 90% cost reduction while matching the accuracy of the best single API, or up to 8% better accuracy while matching the best API's cost.

Catastrophic forgetting remains a severe hindrance to the broad application of artificial neural networks (ANNs), however, it continues to be a poorly understood phenomenon. Despite the extensive amount of work on catastrophic forgetting, we argue that it is still unclear how exactly the phenomenon should be quantified, and, moreover, to what degree all of the choices we make when designing learning systems affect the amount of catastrophic forgetting. We use various testbeds from the reinforcement learning and supervised learning literature to (1) provide evidence that the choice of which modern gradient-based optimization algorithm is used to train an ANN has a significant impact on the amount of catastrophic forgetting and show that--surprisingly--in many instances classical algorithms such as vanilla SGD experience less catastrophic forgetting than the more modern algorithms such as Adam. We empirically compare four different existing metrics for quantifying catastrophic forgetting and (2) show that the degree to which the learning systems experience catastrophic forgetting is sufficiently sensitive to the metric used that a change from one principled metric to another is enough to change the conclusions of a study dramatically. Our results suggest that a much more rigorous experimental methodology is required when looking at catastrophic forgetting. Based on our results, we recommend inter-task forgetting in supervised learning must be measured with both retention and relearning metrics concurrently, and intra-task forgetting in reinforcement learning must--at the very least--be measured with pairwise interference.

Hindsight rationality is an approach to playing multi-agent, general-sum games that prescribes no-regret learning dynamics and describes jointly rational behavior with mediated equilibria. We explore the space of deviation types in extensive-form games (EFGs) and discover powerful types that are efficient to compute in games with moderate lengths. Specifically, we identify four new types of deviations that subsume previously studied types within a broader class we call partial sequence deviations. Integrating the idea of time selection regret minimization into counterfactual regret minimization (CFR), we introduce the extensive-form regret minimization (EFR) algorithm that is hindsight rational for a general and natural class of deviations in EFGs. We provide instantiations and regret bounds for EFR that correspond to each partial sequence deviation type. In addition, we present a thorough empirical analysis of EFR's performance with different deviation types in common benchmark games. As theory suggests, instantiating EFR with stronger deviations leads to behavior that tends to outperform that of weaker deviations.

AI systems often rely on two key components: a specified goal or reward function and an optimization algorithm to compute the optimal behavior for that goal. This approach is intended to provide value for a principal: the user on whose behalf the agent acts. The objectives given to these agents often refer to a partial specification of the principal's goals. We consider the cost of this incompleteness by analyzing a model of a principal and an agent in a resource constrained world where the $L$ attributes of the state correspond to different sources of utility for the principal. We assume that the reward function given to the agent only has support on $J < L$ attributes. The contributions of our paper are as follows: 1) we propose a novel model of an incomplete principal-agent problem from artificial intelligence; 2) we provide necessary and sufficient conditions under which indefinitely optimizing for any incomplete proxy objective leads to arbitrarily low overall utility; and 3) we show how modifying the setup to allow reward functions that reference the full state or allowing the principal to update the proxy objective over time can lead to higher utility solutions. The results in this paper argue that we should view the design of reward functions as an interactive and dynamic process and identifies a theoretical scenario where some degree of interactivity is desirable.

We consider the problem of local planning in fixed-horizon Markov Decision Processes (MDPs) with a generative model under the assumption that the optimal value function lies in the span of a feature map that is accessible through the generative model. As opposed to previous work where linear realizability of all policies was assumed, we consider the significantly relaxed assumption of a single linearly realizable (deterministic) policy. A recent lower bound established that the related problem when the action-value function of the optimal policy is linearly realizable requires an exponential number of queries, either in H (the horizon of the MDP) or d (the dimension of the feature mapping). Their construction crucially relies on having an exponentially large action set. In contrast, in this work, we establish that poly$(H, d)$ learning is possible (with state value function realizability) whenever the action set is small (i.e. O(1)). In particular, we present the TensorPlan algorithm which uses poly$((dH/\delta)^A)$ queries to find a $\delta$-optimal policy relative to any deterministic policy for which the value function is linearly realizable with a parameter from a fixed radius ball around zero. This is the first algorithm to give a polynomial query complexity guarantee using only linear-realizability of a single competing value function. Whether the computation cost is similarly bounded remains an interesting open question. The upper bound is complemented by a lower bound which proves that in the infinite-horizon episodic setting, planners that achieve constant suboptimality need exponentially many queries, either in the dimension or the number of actions.

Cooperative multi-agent reinforcement learning (MARL) has achieved significant results, most notably by leveraging the representation learning abilities of deep neural networks. However, large centralized approaches quickly become infeasible as the number of agents scale, and fully decentralized approaches can miss important opportunities for information sharing and coordination. Furthermore, not all agents are equal - in some cases, individual agents may not even have the ability to send communication to other agents or explicitly model other agents. This paper considers the case where there is a single, powerful, central agent that can observe the entire observation space, and there are multiple, low powered, local agents that can only receive local observations and cannot communicate with each other. The job of the central agent is to learn what message to send to different local agents, based on the global observations, not by centrally solving the entire problem and sending action commands, but by determining what additional information an individual agent should receive so that it can make a better decision. After explaining our MARL algorithm, hammer, and where it would be most applicable, we implement it in the cooperative navigation and multi-agent walker domains. Empirical results show that 1) learned communication does indeed improve system performance, 2) results generalize to multiple numbers of agents, and 3) results generalize to different reward structures.