Several approaches have been developed to answer specific questions that a user may have about an AI system that can plan and act. However, the problems of identifying which questions to ask and that of computing a user-interpretable symbolic descrip
tion of the overall capabilities of the system have remained largely unaddressed. This paper presents an approach for addressing these problems by learning user-interpretable symbolic descriptions of the limits and capabilities of a black-box AI system using low-level simulators. It uses a hierarchical active querying paradigm to generate questions and to learn a user-interpretable model of the AI system based on its responses. In contrast to prior work, we consider settings where imprecision of the users conceptual vocabulary precludes a direct expression of the agents capabilities. Furthermore, our approach does not require assumptions about the internal design of the target AI system or about the methods that it may use to compute or learn task solutions. Empirical evaluation on several game-based simulator domains shows that this approach can efficiently learn symbolic models of AI systems that use a deterministic black-box policy in fully observable scenarios.
In humans and animals, curriculum learning -- presenting data in a curated order - is critical to rapid learning and effective pedagogy. Yet in machine learning, curricula are not widely used and empirically often yield only moderate benefits. This s
tark difference in the importance of curriculum raises a fundamental theoretical question: when and why does curriculum learning help? In this work, we analyse a prototypical neural network model of curriculum learning in the high-dimensional limit, employing statistical physics methods. Curricula could in principle change both the learning speed and asymptotic performance of a model. To study the former, we provide an exact description of the online learning setting, confirming the long-standing experimental observation that curricula can modestly speed up learning. To study the latter, we derive performance in a batch learning setting, in which a network trains to convergence in successive phases of learning on dataset slices of varying difficulty. With standard training losses, curriculum does not provide generalisation benefit, in line with empirical observations. However, we show that by connecting different learning phases through simple Gaussian priors, curriculum can yield a large improvement in test performance. Taken together, our reduced analytical descriptions help reconcile apparently conflicting empirical results and trace regimes where curriculum learning yields the largest gains. More broadly, our results suggest that fully exploiting a curriculum may require explicit changes to the loss function at curriculum boundaries.
Transfer learning can significantly improve the sample efficiency of neural networks, by exploiting the relatedness between a data-scarce target task and a data-abundant source task. Despite years of successful applications, transfer learning practic
e often relies on ad-hoc solutions, while theoretical understanding of these procedures is still limited. In the present work, we re-think a solvable model of synthetic data as a framework for modeling correlation between data-sets. This setup allows for an analytic characterization of the generalization performance obtained when transferring the learned feature map from the source to the target task. Focusing on the problem of training two-layer networks in a binary classification setting, we show that our model can capture a range of salient features of transfer learning with real data. Moreover, by exploiting parametric control over the correlation between the two data-sets, we systematically investigate under which conditions the transfer of features is beneficial for generalization.
When optimizing over loss functions it is common practice to use momentum-based accelerated methods rather than vanilla gradient-based method. Despite widely applied to arbitrary loss function, their behaviour in generically non-convex, high dimensio
nal landscapes is poorly understood. In this work we used dynamical mean field theory techniques to describe analytically the average behaviour of these methods in a prototypical non-convex model: the (spiked) matrix-tensor model. We derive a closed set of equations that describe the behaviours of several algorithms including heavy-ball momentum and Nesterov acceleration. Additionally we characterize the evolution of a mathematically equivalent physical system of massive particles relaxing toward the bottom of an energetic landscape. Under the correct mapping the two dynamics are equivalent and it can be noticed that having a large mass increases the effective time step of the heavy ball dynamics leading to a speed up.
We study the dynamics of optimization and the generalization properties of one-hidden layer neural networks with quadratic activation function in the over-parametrized regime where the layer width $m$ is larger than the input dimension $d$. We cons
ider a teacher-student scenario where the teacher has the same structure as the student with a hidden layer of smaller width $m^*le m$. We describe how the empirical loss landscape is affected by the number $n$ of data samples and the width $m^*$ of the teacher network. In particular we determine how the probability that there be no spurious minima on the empirical loss depends on $n$, $d$, and $m^*$, thereby establishing conditions under which the neural network can in principle recover the teacher. We also show that under the same conditions gradient descent dynamics on the empirical loss converges and leads to small generalization error, i.e. it enables recovery in practice. Finally we characterize the time-convergence rate of gradient descent in the limit of a large number of samples. These results are confirmed by numerical experiments.
In this paper we consider the epidemic competition between two generic diffusion processes, where each competing side is represented by a different state of a stochastic process. For this setting, we present the Generalized Largest Reduction in Infec
tious Edges (gLRIE) dynamic resource allocation strategy to advantage the preferred state against the other. Motivated by social epidemics, we apply this method to a generic continuous-time SIS-like diffusion model where we allow for: i) arbitrary node transition rate functions that describe the dynamics of propagation depending on the network state, and ii) competition between the healthy (positive) and infected (negative) states, which are both diffusive at the same time, yet mutually exclusive on each node. Finally we use simulations to compare empirically the proposed gLRIE against competitive approaches from literature.
Despite the widespread use of gradient-based algorithms for optimizing high-dimensional non-convex functions, understanding their ability of finding good minima instead of being trapped in spurious ones remains to a large extent an open problem. Here
we focus on gradient flow dynamics for phase retrieval from random measurements. When the ratio of the number of measurements over the input dimension is small the dynamics remains trapped in spurious minima with large basins of attraction. We find analytically that above a critical ratio those critical points become unstable developing a negative direction toward the signal. By numerical experiments we show that in this regime the gradient flow algorithm is not trapped; it drifts away from the spurious critical points along the unstable direction and succeeds in finding the global minimum. Using tools from statistical physics we characterize this phenomenon, which is related to a BBP-type transition in the Hessian of the spurious minima.
Prior work on generating explanations in a planning and decision-making context has focused on providing the rationale behind an AI agents decision making. While these methods provide the right explanations from the explainers perspective, they fail
to heed the cognitive requirement of understanding an explanation from the explainees (the humans) perspective. In this work, we set out to address this issue by first considering the influence of information order in an explanation, or the progressiveness of explanations. Intuitively, progression builds later concepts on previous ones and is known to contribute to better learning. In this work, we aim to investigate similar effects during explanation generation when an explanation is broken into multiple parts that are communicated sequentially. The challenge here lies in modeling the humans preferences for information order in receiving such explanations to assist understanding. Given this sequential process, a formulation based on goal-based MDP for generating progressive explanations is presented. The reward function of this MDP is learned via inverse reinforcement learning based on explanations that are retrieved via human subject studies. We first evaluated our approach on a scavenger-hunt domain to demonstrate its effectively in capturing the humans preferences. Upon analyzing the results, it revealed something more fundamental: the preferences arise strongly from both domain dependent and independence features. The correlation with domain independent features pushed us to verify this result further in an escape room domain. Results confirmed our hypothesis that the process of understanding an explanation was a dynamic process. The human preference that reflected this aspect corresponded exactly to the progression for knowledge assimilation hidden deeper in our cognitive process.
Traditional slot filling in natural language understanding (NLU) predicts a one-hot vector for each word. This form of label representation lacks semantic correlation modelling, which leads to severe data sparsity problem, especially when adapting an
NLU model to a new domain. To address this issue, a novel label embedding based slot filling framework is proposed in this paper. Here, distributed label embedding is constructed for each slot using prior knowledge. Three encoding methods are investigated to incorporate different kinds of prior knowledge about slots: atomic concepts, slot descriptions, and slot exemplars. The proposed label embeddings tend to share text patterns and reuses data with different slot labels. This makes it useful for adaptive NLU with limited data. Also, since label embedding is independent of NLU model, it is compatible with almost all deep learning based slot filling models. The proposed approaches are evaluated on three datasets. Experiments on single domain and domain adaptation tasks show that label embedding achieves significant performance improvement over traditional one-hot label representation as well as advanced zero-shot approaches.
We review recent works on analyzing the dynamics of gradient-based algorithms in a prototypical statistical inference problem. Using methods and insights from the physics of glassy systems, these works showed how to understand quantitatively and qual
itatively the performance of gradient-based algorithms. Here we review the key results and their interpretation in non-technical terms accessible to a wide audience of physicists in the context of related works.
