Recently, deep reinforcement learning (DRL) methods have achieved impressive performance on tasks in a variety of domains. However, neural network policies produced with DRL methods are not human-interpretable and often have difficulty generalizing t
o novel scenarios. To address these issues, prior works explore learning programmatic policies that are more interpretable and structured for generalization. Yet, these works either employ limited policy representations (e.g. decision trees, state machines, or predefined program templates) or require stronger supervision (e.g. input/output state pairs or expert demonstrations). We present a framework that instead learns to synthesize a program, which details the procedure to solve a task in a flexible and expressive manner, solely from reward signals. To alleviate the difficulty of learning to compose programs to induce the desired agent behavior from scratch, we propose to first learn a program embedding space that continuously parameterizes diverse behaviors in an unsupervised manner and then search over the learned program embedding space to yield a program that maximizes the return for a given task. Experimental results demonstrate that the proposed framework not only learns to reliably synthesize task-solving programs but also outperforms DRL and program synthesis baselines while producing interpretable and more generalizable policies. We also justify the necessity of the proposed two-stage learning scheme as well as analyze various methods for learning the program embedding.
In this paper, we study the representations of the Hopf-Ore extensions $kG(chi^{-1}, a, 0)$ of group algebra $kG$, where $k$ is an algebraically closed field. We classify all finite dimensional simple $kG(chi^{-1}, a, 0)$-modules under the assumption
$|chi|=infty$ and $|chi|=|chi(a)|<infty$ respectively, and all finite dimensional indecomposable $kG(chi^{-1}, a, 0)$-modules under the assumption that $kG$ is finite dimensional and semisimple, and $|chi|=|chi(a)|$. Moreover, we investigate the decomposition rules for the tensor product modules over $kG(chi^{-1}, a, 0)$ when char$(k)$=0. Finally, we consider the representations of some Hopf-Ore extension of the dihedral group algebra $kD_n$, where $n=2m$, $m>1$ odd, and char$(k)$=0. The Grothendieck ring and the Green ring of the Hopf-Ore extension are described respectively in terms of generators and relations.
Rare information on photodisintegration reactions of nuclei with mass numbers $A approx 160$ at astrophysical conditions impedes our understanding of the origin of $p$-nuclei. Experimental determination of the key ($p,gamma$) cross sections has been
playing an important role to verify nuclear reaction models and to provide rates of relevant ($gamma,p$) reactions in $gamma$-process. In this paper we report the first cross section measurements of $^{160}$Dy($p,gamma$)$^{161}$Ho and $^{161}$Dy($p,n$)$^{161}$Ho in the beam energy range of 3.4 - 7.0 MeV, partially covering the Gamow window. Such determinations are possible by using two targets with various isotopic fractions. The cross section data can put a strong constraint on the nuclear level densities and gamma strength functions for $A approx$ 160 in the Hauser-Feshbach statistical model. Furthermore, we find the best parameters for TALYS that reproduce the A $thicksim$ 160 data available, $^{160}$Dy($p,gamma$)$^{161}$Ho and $^{162}$Er($p,gamma$)$^{163}$Tm, and recommend the constrained $^{161}$Ho($gamma,p$)$^{160}$Dy reaction rates over a wide temperature range for $gamma$-process network calculations. Although the determined $^{161}$Ho($gamma$, p) stellar reaction rates at the temperature of 1 to 2 GK can differ by up to one order of magnitude from the NON-SMOKER predictions, it has a minor effect on the yields of $^{160}$Dy and accordingly the $p$-nuclei, $^{156,158}$Dy. A sensitivity study confirms that the cross section of $^{160}$Dy($p$, $gamma$)$^{161}$Ho is measured precisely enough to predict yields of $p$-nuclei in the $gamma$-process.
Object detection is widely used on embedded devices. With the wide availability of CNN (Convolutional Neural Networks) accelerator chips, the object detection applications are expected to run with low power consumption, and high inference speed. In a
ddition, the CPU load is expected to be as low as possible for a CNN accelerator chip working as a co-processor with a host CPU. In this paper, we optimize the object detection model on the CNN accelerator chip by minimizing the CPU load. The resulting model is called GnetDet. The experimental result shows that the GnetDet model running on a 224mW chip achieves the speed of 106FPS with excellent accuracy.
In the robust secure aggregation problem, a server wishes to learn and only learn the sum of the inputs of a number of users while some users may drop out (i.e., may not respond). The identity of the dropped users is not known a priori and the server
needs to securely recover the sum of the remaining surviving users. We consider the following minimal two-round model of secure aggregation. Over the first round, any set of no fewer than $U$ users out of $K$ users respond to the server and the server wants to learn the sum of the inputs of all responding users. The remaining users are viewed as dropped. Over the second round, any set of no fewer than $U$ users of the surviving users respond (i.e., dropouts are still possible over the second round) and from the information obtained from the surviving users over the two rounds, the server can decode the desired sum. The security constraint is that even if the server colludes with any $T$ users and the messages from the dropped users are received by the server (e.g., delayed packets), the server is not able to infer any additional information beyond the sum in the information theoretic sense. For this information theoretic secure aggregation problem, we characterize the optimal communication cost. When $U leq T$, secure aggregation is not feasible, and when $U > T$, to securely compute one symbol of the sum, the minimum number of symbols sent from each user to the server is $1$ over the first round, and $1/(U-T)$ over the second round.
In the conventional robust $T$-colluding private information retrieval (PIR) system, the user needs to retrieve one of the possible messages while keeping the identity of the requested message private from any $T$ colluding servers. Motivated by the
possible heterogeneous privacy requirements for different messages, we consider the $(N, T_1:K_1, T_2:K_2)$ two-level PIR system, where $K_1$ messages need to be retrieved privately against $T_1$ colluding servers, and all the messages need to be retrieved privately against $T_2$ colluding servers where $T_2leq T_1$. We obtain a lower bound to the capacity by proposing two novel coding schemes, namely the non-uniform successive cancellation scheme and the non-uniform block cancellation scheme. A capacity upper bound is also derived. The gap between the upper bound and the lower bounds is analyzed, and shown to vanish when $T_1=T_2$. Lastly, we show that the upper bound is in general not tight by providing a stronger bound for a special setting.
Semantic segmentation is the task to cluster pixels on an image belonging to the same class. It is widely used in the real-world applications including autonomous driving, medical imaging analysis, industrial inspection, smartphone camera for person
segmentation and so on. Accelerating the semantic segmentation models on the mobile and edge devices are practical needs for the industry. Recent years have witnessed the wide availability of CNN (Convolutional Neural Networks) accelerators. They have the advantages on power efficiency, inference speed, which are ideal for accelerating the semantic segmentation models on the edge devices. However, the CNN accelerator chips also have the limitations on flexibility and memory. In addition, the CPU load is very critical because the CNN accelerator chip works as a co-processor with a host CPU. In this paper, we optimize the semantic segmentation model in order to fully utilize the limited memory and the supported operators on the CNN accelerator chips, and at the same time reduce the CPU load of the CNN model to zero. The resulting model is called GnetSeg. Furthermore, we propose the integer encoding for the mask of the GnetSeg model, which minimizes the latency of data transfer between the CNN accelerator and the host CPU. The experimental result shows that the model running on the 224mW chip achieves the speed of 318FPS with excellent accuracy for applications such as person segmentation.
Timeliness is an emerging requirement for many Internet of Things (IoT) applications. In IoT networks, where a large-number of nodes are distributed, severe interference may incur during the transmission phase which causes age of information (AoI) de
gradation. It is therefore important to study the performance limit of AoI as well as how to achieve such limit. In this paper, we aim to optimize the AoI in random access Poisson networks. By taking into account the spatio-temporal interactions amongst the transmitters, an expression of the peak AoI is derived, based on explicit expressions of the optimal peak AoI and the corresponding optimal system parameters including the packet arrival rate and the channel access probability are further derived. It is shown that with a given packet arrival rate (resp. a given channel access probability), the optimal channel access probability (resp. the optimal packet arrival rate), is equal to one under a small node deployment density, and decrease monotonically as the spatial deployment density increases due to the severe interference caused by spatio-temproal coupling between transmitters. When joint tuning of the packet arrival rate and channel access probability is performed, the optimal channel access probability is always set to be one. Moreover, with the sole tuning of the channel access probability, it is found that the optimal peak AoI performance can be improved with a smaller packet arrival rate only when the node deployment density is high, which is contrast to the case of the sole tuning of the packet arrival rate, where a higher channel access probability always leads to better optimal peak AoI regardless of the node deployment density. In all the cases of optimal tuning of system parameters, the optimal peak AoI linearly grows with the node deployment density as opposed to an exponential growth with fixed system parameters.
Optical Character Recognition (OCR) has many real world applications. The existing methods normally detect where the characters are, and then recognize the character for each detected location. Thus the accuracy of characters recognition is impacted
by the performance of characters detection. In this paper, we propose a method for recognizing characters without detecting the location of each character. This is done by converting the OCR task into an image captioning task. One advantage of the proposed method is that the labeled bounding boxes for the characters are not needed during training. The experimental results show the proposed method outperforms the existing methods on both the license plate recognition and the watermeter character recognition tasks. The proposed method is also deployed into a low-power (300mW) CNN accelerator chip connected to a Raspberry Pi 3 for on-device applications.
Lithium (Li) is the simplest metal and the lightest solid element. Here we report the first demonstration of controlled growth of two-dimensional (2D) ultrathin Li nanosheets with large lateral dimensions up to several hundreds of nanometres and thic
kness limited to just a few nanometres by in-situ transmission electron microscopy (TEM). The nanoscale dynamics of nanosheets growth were unravelled by real-time TEM imaging, which, in combination with density function theory (DFT) calculations indicates that the growth of bcc structured Li into 2D nanosheets is a consequence of kinetic control as mediated by preferential oxidization of the (111) surfaces due to the trace amount of O2 (~10-6 Pa) within TEM chamber. The plasmonic optical properties of the as-grown Li nanosheets were probed by cathodoluminescence (CL) spectroscopy equipped within TEM, and a broadband visible emission was observed that contains contributions of both in-plane and out-of-plane plasmon resonance modes.
