Functional verification of power gated designs by compositional reasoning

Power gating is a technique for low power design in which whole sections of the chip are powered off when they are not needed, and powered back on when they are. Functional correctness of power gating is usually checked at the system level, where the most widely used technique is simulation using pseudo-random stimuli. This normally entails running extremely expensive ternary simulations, in order to model the memory loss that occurs as a result of a memory element being powered off. We propose instead a methodology in which we prove sequential equivalence between the power gated design and a simplified version of itself, then use the simplified version in a binary simulation. We use a compositional approach that looks for partial equivalence of each unit under a suitable set of assumptions, guaranteed by the neighboring units. The partial equivalences are then composed into total equivalence on the whole chip. Our method is applicable to any power gated design, no matter the side effects (e.g., in timing of events across the interfaces) caused by the particular implementation of power gating.     

Hybrid systems: from verification to falsification by combining motion planning and discrete search

We propose HyDICE, Hybrid Discrete Continuous Exploration, a multi-layered approach for hybrid-system falsification that combines motion planning with discrete search and discovers safety violations by computing witness trajectories to unsafe states. The discrete search uses discrete transitions and a state-space decomposition to guide the motion planner during the search for witness trajectories. Experiments on a nonlinear hybrid robotic system with over one million modes and experiments with an aircraft conflict-resolution protocol with high-dimensional continuous state spaces demonstrate the effectiveness of HyDICE. Comparisons to related work show computational speedups of up to two orders of magnitude. Work supported in part by NSF CNS 0615328 (EP, LK, MV), NSF 0713623 (EP, LK), a Sloan Fellowship (LK), and NSF CCF 0613889 (MV). Equipment supported by NSF CNS 0454333 and NSF CNS 0421109 in partnership with Rice, AMD, and Cray. A preliminary version of this work was published in the Proceedings of the 19th International Conference on Computer Aided Verification (CAV 2007). Lecture Notes in Computer Science, eds. W. Damm and H. Hermanns, vol. 4590, pp. 468–481. This work contains substantial improvements to the overall computational method introduced in the preliminary work and new experiments that were not included in the preliminary work.     

Synthetic speech in foreign language learning: an evaluation by learners

Can synthetic speech be utilized in foreign language learning as natural speech? In this paper, we evaluated synthetic speech from the viewpoint of learners in order to find out an answer. The results pointed out that learners do not recognize remarkable differences between synthetic voices and natural voices for the words with short vowels and long vowels when they try to understand the meanings of the sounds. The data explicates that synthetic voice utterances of sentences are easier to understand and more acceptable by learners compared to synthetic voice utterances of words. In addition, the ratings on both synthetic voices and natural voices strongly depend upon the learners’ listening comprehension abilities. We conclude that some synthetic speech with specific pronunciations of vowels may be suitable for listening materials and suggest that evaluating TTS systems by comparing synthetic speech with natural speech and building a lexical database of synthetic speech that closely approximates natural speech will be helpful for teachers to readily use many existing CALL tools.     

Formal verification of real-time systems with preemptive scheduling

In this paper, we propose a method for the verification of timed properties for real-time systems featuring a preemptive scheduling policy: the system, modeled as a scheduling time Petri net, is first translated into a linear hybrid automaton to which it is time-bisimilar. Timed properties can then be verified using HyTech. The efficiency of this approach leans on two major points: first, the translation features a minimization of the number of variables (clocks) of the resulting automaton, which is a critical parameter for the efficiency of the ensuing verification. Second, the translation is performed by an over-approximating algorithm, which is based on Difference Bound Matrix and therefore efficient, that nonetheless produces a time-bisimilar automaton despite the over-approximation. The proposed modeling and verification method are generic enough to account for many scheduling policies. In this paper, we specifically show how to deal with Fixed Priority and Earliest Deadline First policies, with the possibility of using Round-Robin for tasks with the same priority. We have implemented the method and give some experimental results illustrating its efficiency.

Bisimulation conversion and verification procedure for goal-based control systems

Fault tolerance and safety verification of control systems are essential for the success of autonomous robotic systems. A control architecture called Mission Data System (MDS), developed at the Jet Propulsion Laboratory, addresses these needs with a goal-based control approach. In this paper, a software algorithm for converting goal network control systems into linear hybrid systems is described. The conversion process is a bisimulation; the resulting linear hybrid system can be verified for safety in the presence of failures using existing symbolic model checkers, and thus the original goal network is verified. A moderately complex example goal network control system is converted to a linear hybrid system using the automatic conversion software that is based on the bisimulation and then is verified.     

Automatic generation of temporal planning domains for e-learning problems

AI Planning & Scheduling techniques are being widely used to adapt learning paths to the special features and needs of students both in distance learning and lifelong learning environments. However, instructors strongly rely on Planning & Scheduling experts to encode and review the domains for the planner/scheduler to work. This paper presents an approach to automatically extract a fully operational HTN planning domain and problem from a learning objects repository without requiring the intervention of any planning expert, and thus enabling an easier adoption of this technology in practice. The results of a real experiment with a small group of students within an e-Learning private company in Spain are also shown.     

Pivoting based manipulation by a humanoid robot

In this paper we address whole-body manipulation of bulky objects by a humanoid robot. We adopt a “pivoting” manipulation method that allows the humanoid to displace an object without lifting, but by the support of the ground contact. First, the small-time controllability of pivoting is demonstrated. On its basis, an algorithm for collision-free pivoting motion planning is established taking into account the naturalness of motion as nonholonomic constraints. Finally, we present a whole-body motion generation method by a humanoid robot, which is verified by experiments.     

Timed verification of the generic architecture of a memory circuit using parametric timed automata

Using a variant of Clariso-Cortadella’s parametric method for verifying asynchronous circuits, we analyse some crucial timing behaviors of the architecture of SPSMALL memory, a commercial product of STMicroelectronics. Using the model of parametric timed automata and model checker HYTECH, we formally derive a set of linear constraints that ensure the correctness of the response times of the memory. We are also able to infer the constraints characterizing the optimal setup timings of input signals. We have checked, for two different implementations of this architecture, that the values given by our model match remarkably with the values obtained by the designer through electrical simulation. Partially supported by project MEDEA+ Blueberries. A preliminary version appeared in the Proceedings of 4th International Conference on Formal Modelling and Analysis of Timed Systems (FORMATS’06), Sept. 2006.     

Automatic Georeferencing of Images Acquired by UAV’s

This paper implements and evaluates experimentally a procedure for automatically georeferencing images acquired by unmanned aerial vehicles (UAV’s) in the sense that ground control points (GCP) are not necessary. Since the camera model is necessary for georeferencing, this paper also proposes a completely automatic procedure for collecting corner pixels in the model plane image to solve the camera calibration problem, i.e., to estimate the camera and the lens distortion parameters. The performance of the complete georeferencing system is evaluated with real flight data obtained by a typical UAV.     

Distance learning success – a perspective from socio-technical systems theory

With widespread adoption of computer-based distance education as a mission-critical component of the institution's educational program, the need for evaluation has emerged. In this research, we aim to expand on the systems approach by offering a model for evaluation based on socio-technical systems theory addressing a stated need in the literature for comprehensive models for evaluating e-learning environments (Holsapple, C.W. and Lee-Post, A., 2006 Holsapple, C. W. and Lee-Post, A. 2006. Defining, assessing, and promoting e-learning success: an information systems perspective. Decision Sciences Journal of Innovative Education, 4(1): 67–85. [Crossref] , [Google Scholar]. Defining, assessing, and promoting e-learning success: an information systems perspective. Decision Sciences Journal of Innovative Education, 4(1), 67–85). The proposed systems model evaluates distance learning success from the instructor's perspective. It defines and develops measures for course quality, system quality and corresponding impacts. The model is tested based on the data collected from 548 instructors of seven universities in the Midwest region of the USA. The results suggest that the proposed multi-dimensional system flexibility scale is reliable. The course quality significantly affects both system flexibility and faculty perceived impacts of distance education. The system flexibility also significantly affects both course quality and faculty perceived impacts.     

A novel unsupervised classification approach for network anomaly detection by k-Means clustering and ID3 decision tree learning methods

This paper presents a novel host-based combinatorial method based on k-Means clustering and ID3 decision tree learning algorithms for unsupervised classification of anomalous and normal activities in computer network ARP traffic. The k-Means clustering method is first applied to the normal training instances to partition it into k clusters using Euclidean distance similarity. An ID3 decision tree is constructed on each cluster. Anomaly scores from the k-Means clustering algorithm and decisions of the ID3 decision trees are extracted. A special algorithm is used to combine results of the two algorithms and obtain final anomaly score values. The threshold rule is applied for making the decision on the test instance normality. Experiments are performed on captured network ARP traffic. Some anomaly criteria has been defined and applied to the captured ARP traffic to generate normal training instances. Performance of the proposed approach is evaluated using five defined measures and empirically compared with the performance of individual k-Means clustering and ID3 decision tree classification algorithms and the other proposed approaches based on Markovian chains and stochastic learning automata. Experimental results show that the proposed approach has specificity and positive predictive value of as high as 96 and 98%, respectively.     

Large scale image annotation: learning to rank with joint word-image embeddings

Image annotation datasets are becoming larger and larger, with tens of millions of images and tens of thousands of possible annotations. We propose a strongly performing method that scales to such datasets by simultaneously learning to optimize precision at k of the ranked list of annotations for a given image and learning a low-dimensional joint embedding space for both images and annotations. Our method both outperforms several baseline methods and, in comparison to them, is faster and consumes less memory. We also demonstrate how our method learns an interpretable model, where annotations with alternate spellings or even languages are close in the embedding space. Hence, even when our model does not predict the exact annotation given by a human labeler, it often predicts similar annotations, a fact that we try to quantify by measuring the newly introduced “sibling” precision metric, where our method also obtains excellent results.     

Knows what it knows: a framework for self-aware learning

We introduce a learning framework that combines elements of the well-known PAC and mistake-bound models. The KWIK (knows what it knows) framework was designed particularly for its utility in learning settings where active exploration can impact the training examples the learner is exposed to, as is true in reinforcement-learning and active-learning problems. We catalog several KWIK-learnable classes as well as open problems, and demonstrate their applications in experience-efficient reinforcement learning.     

Automatic tuning of iterative computation on heterogeneous multiprocessors with ADITHE

This work studies the problem of balancing the workload of iterative algorithms on heterogeneous multiprocessors. An approach, called ADITHE, is proposed and evaluated. Its main features are: (1) using a homogeneous distribution of the workload on the heterogeneous system, the speed of every node is estimated during the first iterations of the algorithm; (2) according to the speed of every node, a new workload distribution is carried out; (3) the remaining iterations of the algorithm are executed. The result of this workload redistribution is that the execution times for every iteration at every node are similar and, consequently, the penalties due to synchronization between nodes at every iteration are mostly eliminated. This approach is appropriate for iterative algorithms with similar workload at every iteration, and with a relevant number of iterations. The high portability of ADITHE is guaranteed because the estimation of speed of nodes is included in the execution of the parallel algorithm. There is a wide variety of iterative algorithms related to science and engineering which can take advantage of ADITHE. An example of this kind of algorithms (morphological processing of hyperspectral images) is considered in this work to evaluate its performance when ADITHE is applied. The analysis of the results shows that ADITHE significantly improves the performance of parallel iterative algorithms on heterogeneous platforms.     

A new formalism for mathematical description and verification of component-based systems

In this paper, we introduce a formal approach for composing software components into a distributed system. We describe the system as a hierarchical composition of some components, which can be distributed on a wide variety of hardware platforms and executed in parallel. We represent each component by a mathematical model and specify the abstract communication protocols of the components using Interface Automata (IAs). To model hierarchical systems, besides the basic components’ model, we will present other components, called nodes. A node consists of a set of subnodes interacting under the supervision of a controller. Each subnode, in turn, is a node or discrete event component. By considering a subnode as a node we can make hierarchical nodes/components. The entire system, therefore, forms the root of the hierarchy. A controller, in turn, is a set of subcontrollers/interface automata that specifies interaction protocol of the components inside a node. We have also presented an example demonstrating the model by illustrating nodes, subnodes, controllers, and subcontrollers. To address the state space explosion problem in system verification, we utilize the controller as a contract for independent analysis of the components and their interactions. Therefore, a node will not be analyzed directly; instead, we will analyze the controller.     

A co-classification approach to learning from multilingual corpora

We address the problem of learning text categorization from a corpus of multilingual documents. We propose a multiview learning, co-regularization approach, in which we consider each language as a separate source, and minimize a joint loss that combines monolingual classification losses in each language while ensuring consistency of the categorization across languages. We derive training algorithms for logistic regression and boosting, and show that the resulting categorizers outperform models trained independently on each language, and even, most of the times, models trained on the joint bilingual data. Experiments are carried out on a multilingual extension of the RCV2 corpus, which is available for benchmarking.     

Single machine scheduling models with deterioration and learning: handling precedence constraints via priority generation

We consider various single machine scheduling problems in which the processing time of a job depends either on its position in a processing sequence or on its start time. We focus on problems of minimizing the makespan or the sum of (weighted) completion times of the jobs. In many situations we show that the objective function is priority-generating, and therefore the corresponding scheduling problem under series-parallel precedence constraints is polynomially solvable. In other situations we provide counter-examples that show that the objective function is not priority-generating.     

Imperative-program transformation by instrumented-interpreter specialization

We describe how to implement strength reduction, loop-invariant code motion and loop quasi-invariant code motion by specializing instrumented interpreters. To curb code duplication intrinsic to such specialization, we introduce a new program transformation, rewinding, which uses Moore-automata minimization to remove duplicated code.     

Ad hoc teamwork by learning teammates’ task

This paper addresses the problem of ad hoc teamwork, where a learning agent engages in a cooperative task with other (unknown) agents. The agent must effectively coordinate with the other agents towards completion of the intended task, not relying on any pre-defined coordination strategy. We contribute a new perspective on the ad hoc teamwork problem and propose that, in general, the learning agent should not only identify (and coordinate with) the teammates’ strategy but also identify the task to be completed. In our approach to the ad hoc teamwork problem, we represent tasks as fully cooperative matrix games. Relying exclusively on observations of the behavior of the teammates, the learning agent must identify the task at hand (namely, the corresponding payoff function) from a set of possible tasks and adapt to the teammates’ behavior. Teammates are assumed to follow a bounded-rationality best-response model and thus also adapt their behavior to that of the learning agent. We formalize the ad hoc teamwork problem as a sequential decision problem and propose two novel approaches to address it. In particular, we propose (i) the use of an online learning approach that considers the different tasks depending on their ability to predict the behavior of the teammate; and (ii) a decision-theoretic approach that models the ad hoc teamwork problem as a partially observable Markov decision problem. We provide theoretical bounds of the performance of both approaches and evaluate their performance in several domains of different complexity.