Binary Neural Networks (BNNs) have received significant attention due to their promising efficiency. Currently, most BNN studies directly adopt widely-used CNN architectures, which can be suboptimal for BNNs. This paper proposes a novel Binary ARchitecture Search (BARS) flow to discover superior binary architecture in a large design space. Specifically, we design a two-level (Macro \& Micro) search space tailored for BNNs and apply a differentiable neural architecture search (NAS) to explore this search space efficiently. The macro-level search space includes depth and width decisions, which is required for better balancing the model performance and capacity. And we also make modifications to the micro-level search space to strengthen the information flow for BNN. A notable challenge of BNN architecture search lies in that binary operations exacerbate the "collapse" problem of differentiable NAS, and we incorporate various search and derive strategies to stabilize the search process. On CIFAR-10, \method achieves $1.5\%$ higher accuracy with $2/3$ binary Ops and $1/10$ floating-point Ops. On ImageNet, with similar resource consumption, \method-discovered architecture achieves $3\%$ accuracy gain than hand-crafted architectures, while removing the full-precision downsample layer.

The assumption of complete domain knowledge is not warranted for robot planning and decision-making in the real world. It could be due to design flaws or arise from domain ramifications or qualifications. In such cases, existing planning and learning algorithms could produce highly undesirable behaviors. This problem is more challenging than partial observability in the sense that the agent is unaware of certain knowledge, in contrast to it being partially observable: the difference between known unknowns and unknown unknowns. In this work, we formulate it as the problem of Domain Concretization, an inverse problem to domain abstraction. Based on an incomplete domain model provided by the designer and teacher traces from human users, our algorithm searches for a candidate model set under a minimalistic model assumption. It then generates a robust plan with the maximum probability of success under the set of candidate models. In addition to a standard search formulation in the model-space, we propose a sample-based search method and also an online version of it to improve search time. We tested our approach on IPC domains and a simulated robotics domain where incompleteness was introduced by removing domain features from the complete model. Results show that our planning algorithm increases the plan success rate without impacting the cost much.

Sensory information can only be used meaningfully in the brain when integrated with and compared with internally generated top-down signals. However, we know little about the brainwide afferents that convey such top-down signals, their information content, and learning-related plasticity. Pardi et al. identified the higher-order thalamus as a major source of top-down input to mouse auditory cortex and investigated a circuit in cortical layer 1 that facilitates plastic changes and flexible responses. These results demonstrate how top-down feedback information can reach cortical areas through a noncortical structure that has received little attention despite its widespread connections with the cortex. Science , this issue p. [844][1] The sensory neocortex is a critical substrate for memory. Despite its strong connection with the thalamus, the role of direct thalamocortical communication in memory remains elusive. We performed chronic in vivo two-photon calcium imaging of thalamic synapses in mouse auditory cortex layer 1, a major locus of cortical associations. Combined with optogenetics, viral tracing, whole-cell recording, and computational modeling, we find that the higher-order thalamus is required for associative learning and transmits memory-related information that closely correlates with acquired behavioral relevance. In turn, these signals are tightly and dynamically controlled by local presynaptic inhibition. Our results not only identify the higher-order thalamus as a highly plastic source of cortical top-down information but also reveal a level of computational flexibility in layer 1 that goes far beyond hard-wired connectivity. [1]: /lookup/doi/10.1126/science.abc2399

The first three entries in my "2020: A consumer electronics forecast for the year(s) ahead" piece, published back in January, all had to do with deep learning. Why? Here's part of what I wrote back then: The ability to pattern-match and extrapolate from already-identified data ("training") to not-yet-identified data ("inference") has transformed the means by which many algorithms are developed nowadays, with impact on numerous applications. This transformation is already well underway, as even a casual perusal of the titles and coverage topics of content published at EDN, EE Times, and elsewhere will make clear. Don't panic: there's still time to "catch the wave," especially if your focus is on resource-constrained implementations. But you don't want to wait too long lest you end up stuck bobbing around in the water while more foresighted colleagues are already at the beach enjoying the AI "party."

In this comprehensive review, we describe a new mathematical problem in artificial intelligence (AI) from a mathematical modeling perspective, following the philosophy stated by Rudolf E. Kalman that "Once you get the physics right, the rest is mathematics". The new problem is called "Generalized Constraints (GCs)", and we adopt GCs as a general term to describe any type of prior information in modelings. To understand better about GCs to be a general problem, we compare them with the conventional constraints (CCs) and list their extra challenges over CCs. In the construction of AI machines, we basically encounter more often GCs for modeling, rather than CCs with well-defined forms. Furthermore, we discuss the ultimate goals of AI and redefine transparent, interpretable, and explainable AI in terms of comprehension levels about machines. We review the studies in relation to the GC problems although most of them do not take the notion of GCs. We demonstrate that if AI machines are simplified by a coupling with both knowledge-driven submodel and data-driven submodel, GCs will play a critical role in a knowledge-driven submodel as well as in the coupling form between the two submodels. Examples are given to show that the studies in view of a generalized constraint problem will help us perceive and explore novel subjects in AI, or even in mathematics, such as generalized constraint learning (GCL).

Classical conditioning is a psychological process through which animals or humans pair desired or unpleasant stimuli (e.g., food or a painful experiences) with a seemingly neutral stimulus (e.g., the sound of a bell, the flash of a light, etc.) after these two stimuli are repeatedly presented together. Russian psychologist Ivan Pavlov studied classical conditioning in great depth and introduced the idea of "associative memory," which entails building strong associations between the pleasant/unpleasant and neutral stimuli. Pavlov is renowned for his studies on dogs, in which he gave the animals food after they heard a specific sound for several trials. Interestingly, he observed that the dogs would eventually start salivating (i.e., anticipating the food) after hearing the sound, even if the food had not yet been presented to them. This suggests that they had learned to associate the sound with the arrival of food.

Reasoning is a fundamental capability for harnessing valuable insight, knowledge and patterns from knowledge graphs. Existing work has primarily been focusing on point-wise reasoning, including search, link predication, entity prediction, subgraph matching and so on. This paper introduces comparative reasoning over knowledge graphs, which aims to infer the commonality and inconsistency with respect to multiple pieces of clues. We envision that the comparative reasoning will complement and expand the existing point-wise reasoning over knowledge graphs. In detail, we develop KompaRe, the first of its kind prototype system that provides comparative reasoning capability over large knowledge graphs. We present both the system architecture and its core algorithms, including knowledge segment extraction, pairwise reasoning and collective reasoning. Empirical evaluations demonstrate the efficacy of the proposed KompaRe.

In order to facilitate natural language understanding, the key is to engage commonsense or background knowledge. However, how to engage commonsense effectively in question answering systems is still under exploration in both research academia and industry. In this paper, we propose a novel question-answering method by integrating multiple knowledge sources, i.e. ConceptNet, Wikipedia, and the Cambridge Dictionary, to boost the performance. More concretely, we first introduce a novel graph-based iterative knowledge retrieval module, which iteratively retrieves concepts and entities related to the given question and its choices from multiple knowledge sources. Afterward, we use a pre-trained language model to encode the question, retrieved knowledge and choices, and propose an answer choice-aware attention mechanism to fuse all hidden representations of the previous modules. Finally, the linear classifier for specific tasks is used to predict the answer. Experimental results on the CommonsenseQA dataset show that our method significantly outperforms other competitive methods and achieves the new state-of-the-art. In addition, further ablation studies demonstrate the effectiveness of our graph-based iterative knowledge retrieval module and the answer choice-aware attention module in retrieving and synthesizing background knowledge from multiple knowledge sources.

We describe a framework for research and evaluation in Embodied AI. Our proposal is based on a canonical task: Rearrangement. A standard task can focus the development of new techniques and serve as a source of trained models that can be transferred to other settings. In the rearrangement task, the goal is to bring a given physical environment into a specified state. The goal state can be specified by object poses, by images, by a description in language, or by letting the agent experience the environment in the goal state. We characterize rearrangement scenarios along different axes and describe metrics for benchmarking rearrangement performance. To facilitate research and exploration, we present experimental testbeds of rearrangement scenarios in four different simulation environments. We anticipate that other datasets will be released and new simulation platforms will be built to support training of rearrangement agents and their deployment on physical systems.

Bayesian optimisation is a popular method for efficient optimisation of expensive black-box functions. Traditionally, BO assumes that the search space is known. However, in many problems, this assumption does not hold. To this end, we propose a novel BO algorithm which expands (and shifts) the search space over iterations based on controlling the expansion rate thought a hyperharmonic series. Further, we propose another variant of our algorithm that scales to high dimensions. We show theoretically that for both our algorithms, the cumulative regret grows at sub-linear rates. Our experiments with synthetic and real-world optimisation tasks demonstrate the superiority of our algorithms over the current state-of-the-art methods for Bayesian optimisation in unknown search space.