Abstract: In an example, a method for decision-making by an Artificial Intelligence (AI) agent based on risk attitude includes processing an explored state space of an environment to identify one or more potential outcomes for each of a plurality of potential decisions; assigning, based on a utility function, a utility value to each of the one or more potential outcomes for each of the plurality of potential decisions, wherein the utility function depends on a risk parameter indicative of a specified risk preference of the AI agent; determining, based on a predefined sigma algebra defining a set of events that may occur in the environment, a probability of each of the one or more potential outcomes occurring for each of the plurality of potential decisions; selecting a decision from the plurality of potential decisions based on the utility values and the probabilities; and outputting an indication of the decision.

Abstract: One embodiment provides a system which facilitates reasoning about classifiers. During operation, the system determines a plurality of neural networks. The system derives, from a respective neural network, a linear model, wherein the linear model is constructed based on an output of a penultimate layer of the respective neural network. The system trains the linear model based on activations of the penultimate layer. The system maps parameters of the trained linear model into a version space.

Abstract: The present disclosure provides techniques for processing a three-dimensional (3D) object file or object model in a privacy-preserving manner. An example method includes receiving, from a remote computing device, an encrypted object file comprising a specification of a 3D printable object and receiving a request to process the encrypted object file to identify a characteristic of the 3D printable object. The method also includes obtaining an encrypted comparison file and computing an encrypted Minkowski sum of the encrypted object file and the encrypted comparison file to generate an encrypted result file that comprises information about the characteristic. Computing the encrypted Minkowski sum is performed without decrypting the encrypted object file. The method also includes sending the encrypted result file to the remote computing device.

Abstract: The present disclosure provides techniques for processing a three-dimensional (3D) object file in a privacy-preserving manner. An example method includes obtaining an object file that comprises a specification of a 3D printable object, encrypting the object file using a public key to generate an encrypted object file, and sending, to a remote computing system, the encrypted object file and a request to process the encrypted object file to identify a characteristic of the 3D printable object. The method also includes receiving an encrypted result file from the remote computing system, wherein the encrypted result file comprises an encrypted Minkowski sum of the encrypted object file and an encrypted comparison file. The method also includes decrypting the encrypted result file using a private key corresponding with the public key to generate an unencrypted result file and processing the unencrypted result file to determine the characteristic of the 3D printable object.

Abstract: The present disclosure provides techniques for identifying regions of a three-dimensional (3D) printable object that are accessible by a tool. An example method includes receiving, from a remote computing device, an encrypted object file that includes a specification of a 3D printable object and receiving a request to process the encrypted object file to identify regions of the 3D printable object that are accessible by a tool. The method also includes obtaining a tool specification, computing a complement of the tool specification, and encrypting the complement of the tool specification to generate an encrypted comparison file. The method also includes computing an encrypted Minkowski sum of the encrypted object file and the encrypted comparison file to generate an encrypted result file that describes the regions of the 3D printable object that are accessible by the tool. The encrypted Minkowski sum is performed without decrypting the encrypted object file.

Abstract: The present disclosure provides techniques for finding non-printable features in a three-dimensional (3D) printable object. An example method includes receiving a first encrypted file and a request to process the first encrypted file to identify non-printable features of a 3D printable object, wherein the first encrypted file is an encrypted complement of an object file that comprises a specification of the 3D printable object. The method also includes computing a first encrypted Minkowski sum of the first encrypted file and an encrypted minimum feature file to generate an encrypted intermediate file and sending the encrypted intermediate file. The method also includes receiving a second encrypted file that comprises an encrypted complement of the encrypted intermediate file and computing a second encrypted Minkowski sum of the second encrypted file and the encrypted minimum feature file to generate an encrypted result file that describes non-printable features of the 3D printable object.

Abstract: One embodiment provides a system which facilitates construction of an ensemble of machine learning models. During operation, the system determines a training set of data objects, wherein each data object is associated with one of a plurality of classes. The system divides the training set of data objects into a number of partitions. The system generates a respective machine learning model for each respective partition using a universal kernel function, which processes the data objects divided into a respective partition to obtain the ensemble of machine learning models. The system trains the machine learning models based on the data objects of the training set. The system predicts an outcome for a testing data object based on the ensemble of machine learning models and an ensemble decision rule.

Abstract: Embodiments described herein provide a design architecture for co-designing a controller and a watermarking signal for a cyber-physical system. During operation, the architecture can determine, in conjunction with each other, respective values of a first set of parameters indicating operations of the controller and a second set of parameters representing the watermarking signal. Here, the watermarking signal is combinable with a control signal from the controller for monitoring an output signal of the cyber-physical system for detecting malicious data at different time instances. Subsequently, the architecture can determine a state manager for determining the states of the cyber-physical system from the monitored output signal based on the first and second sets of parameters. The architecture can also determine a detector capable of identifying presence of an attack from the states of the cyber-physical system at a plurality of time instances using the watermarking signal.

Abstract: A building environmental sensor includes a sensing element for collecting measurements of environmental parameters such as temperature, humidity, light, sound or the absence or presence of gas. The sensor will: (a) detect that a data collection device is within a communication range of the sensor; (b) generate a data stream that includes the data that the sensor collected; (c) transmit the data stream to the first data collection device; (d) determine that a communication link between the sensor and the first data collection device was lost before the first data stream was fully transmitted; (e) detect that a second data collection device is within the communication range of the sensor; (f) generate a second data stream that includes the remaining data; and (g) transmit the second data stream to the second data collection device.

Abstract: Embodiments provide a system and method for constructing a graph-based model for optimizing the security posture of a composed system. During operation, the system constructs a multi-layer graph for a system with a plurality of components, wherein the multi-layer graph comprises a configuration subgraph, a vulnerability subgraph, and a dependency subgraph. The system constructs the multi-layer graph by the following. The system generates nodes in the configuration subgraph, including: nodes in a first class which encode information associated with a configuration parameter for a respective component, wherein the encoded information includes a name, a default value, a range of values, and a data type; and nodes in a second class which encode value assignments for configuration parameters and relationships between configuration parameters. The system generates nodes in the vulnerability subgraph based on known vulnerabilities associated with a component, bad security practices, and best security practices.

Abstract: A system and a method for secure control of a physical system are described. During operation, the system can obtain measurement of one or more sensors associated with the physical system. The system then estimates a state of the physical system based on the measurement of the sensors. Subsequently, the system generates a feedback control signal based on the estimated state and generates a watermarking signal based on a stored estimated state of the physical system at a previous time instant. The system then generates a watermarked control signal by combining the feedback control signal and the watermarking signal and applies the watermarked control signal to the physical system to regulate the state of the physical system, thereby facilitating secure control of the physical system.

Abstract: Embodiments described herein provide a supervisor for fault management at a production system. During operation, the supervisor can obtain a set of sensor readings and a state of the production system. A respective sensor reading is an output of a sensor in the production system. The supervisor can then determine, using an artificial intelligence (AI) model, whether the set of sensor readings accommodates a fault associated with a corresponding sensor. Subsequently, the supervisor can determine an action that mitigates an effect of the fault and modify the set of sensor readings based on the action. Here, the modified set of sensor readings is used by a controller that controls the production system.

Abstract: One embodiment provides a system which facilitates construction of an ensemble of neural network-based classifiers that optimize a diversity metric. During operation, the system defines a diversity metric based on pairwise angles between decision boundaries of three or more affine classifiers. The system includes the diversity metric as a regularization term in a loss function optimization for designing a pair of mutually orthogonal affine classifiers of the three or more affine classifiers. The system trains one or more neural networks such that parameters of the one or more neural networks are consistent with parameters of the affine classifiers to obtain an ensemble of neural network-based classifiers which optimize the diversity metric. The system predicts an outcome for a testing data object based on the obtained ensemble of neural-network based classifiers which optimize the diversity metric.

Abstract: One embodiment provides a method and system which facilitates optimizing a pair of affine classifiers based on a diversity metric. During operation, the system defines a diversity metric based on an angle between decision boundaries of a pair of affine classifiers. The system includes the diversity metric as a regularization term in a loss function optimization for designing the pair of affine classifiers, wherein the designed pair of affine classifiers are mutually orthogonal. The system predicts an outcome for a testing data object based on the designed pair of mutually orthogonal affine classifiers.

Abstract: Embodiments described herein provide a design architecture for co-designing a controller and a watermarking signal for a cyber-physical system. During operation, the architecture can determine, in conjunction with each other, respective values of a first set of parameters indicating operations of the controller and a second set of parameters representing the watermarking signal. Here, the watermarking signal is combinable with a control signal from the controller for monitoring an output signal of the cyber-physical system for detecting malicious data at different time instances. Subsequently, the architecture can determine a state manager for determining the states of the cyber-physical system from the monitored output signal based on the first and second sets of parameters. The architecture can also determine a detector capable of identifying presence of an attack from the states of the cyber-physical system at a plurality of time instances using the watermarking signal.

Abstract: Embodiments described herein provide a supervisor for fault management at a production system. During operation, the supervisor can obtain a set of sensor readings and a state of the production system. A respective sensor reading is an output of a sensor in the production system. The supervisor can then determine, using an artificial intelligence (AI) model, whether the set of sensor readings accommodates a fault associated with a corresponding sensor. Subsequently, the supervisor can determine an action that mitigates an effect of the fault and modify the set of sensor readings based on the action. Here, the modified set of sensor readings is used by a controller that controls the production system.

Abstract: Systems and methods for indexing blockchain data in a blockchain system. These systems and methods receive a set of transactions from one or more transaction blocks of a blockchain, wherein the transactions in the set have been validated by one or more peer systems of the blockchain. The systems and methods further generate an index to one or more fields of one or more transactions in the set of transactions of the transaction block generate an index representative of at least one field in the set of transactions of the transaction block and provide the generated index for validation by a peer system of the blockchain. After receiving verification from at least a threshold number of peer systems that the generated index has been validated by the peer system, the generated index is stored as an index block in the blockchain.

Abstract: Embodiments provide a system and method for extracting configuration-related information for reasoning about the security and functionality of a composed system. During operation, the system determines, by a computing device, information sources associated with hardware and software components of a system, wherein the information sources include at least specification sheets, standard operating procedures, user manuals, and vulnerability databases. The system selects a set of categories of vulnerabilities in a vulnerability database, and ingests the information sources to obtain data in a normalized format. The system extracts, from the ingested information sources, configuration information, vulnerability information, dependency information, and functionality requirements to create a model for the system.

Abstract: Systems and methods for indexing blockchain data in a blockchain system. These systems and methods receive a set of transactions from one or more transaction blocks of a blockchain, wherein the transactions in the set have been validated by one or more peer systems of the blockchain. The systems and methods further generate an index to one or more fields of one or more transactions in the set of transactions of the transaction block generate an index representative of at least one field in the set of transactions of the transaction block and provide the generated index for validation by a peer system of the blockchain. After receiving verification from at least a threshold number of peer systems that the generated index has been validated by the peer system, the generated index is stored as an index block in the blockchain.

Abstract: The following relates generally to defense mechanisms and security systems. Broadly, systems and methods are disclosed that detect an anomaly in an Embedded Mission Specific Device (EMSD). Disclosed approaches include a meta-material antenna configured to receive a radio frequency signal from the EMSD, and a central reader configured to receive a signal from the meta-material antenna. The central reader may be configured to: build a finite state machine model of the EMSD based on the signal received from the meta-material antenna; and detect if an anomaly exists in the EMSD based on the built finite state machine model.