21 innovations from Bar-Ilan University, available for licensing, co-investment, or spin-out through BIRAD.
Peer Avraham
The discovery describes an innovative source of broad-band, highly-coherent quantum light and high-efficiency photon-pair generation with low pump power. Additionally, a novel method for self-measurement of the quantum coherence is described - the source itself can be used to measure its own performance. The source is based on a nonlinear crystal with polished and coated end facets to create a monolithic broad-band Optical Parametric Oscillator (OPO). Its advantages over common sources of nonlinear crystals in single-pass include: 1. A monolithic OPO provides ideal coherence quality due to minimizing internal losses to a minimum. 2. The required pump power for a given photon flux is low. The threshold for lasing in such a source can be low (less than 5 watts, sometimes down to hundreds of milliwatts, depending on design). 3. Integral dispersion compensation in the crystal mirrors ensures a maximal bandwidth of tens of nanometers and above, generating many pairs of squeezed photons and allowing for a very high flux of entangled photons - up to terahertz pairs per second, which is a significant advantage for quantum communication applications in wavelength-division multiplexing. 4. Mode spacing convenient for telecom (around 10 GHz) in the telecommunication range (1550 nanometers) allows for the construction of separate channels, which is crucial for communication applications. 5. The monolithic design of the OPO ensures passive stability, which facilitates the feedback loop for stabilizing the pump laser to the resonator frequency. Moreover, the concentric design of the resonator ensures stability and resistance to spatial misalignments. Furthermore, the discovery includes a method for self-measurement of the generated coherence using parametric homodyne detection within the crystal itself. Specifically, by operating the monolithic source in a ring resonator configuration, the source can be used in one direction (with the clock) to generate quantum light, and in the opposite direction for measurement, enabling wide-bandwidth homodyne-based measurement, as described in the accompanying documents. Together, the source and measurement method provide a foundation for various applications of quantum technology, such as secure quantum communication (QKD) and wide-bandwidth quantum sensing.
Klein Shmuel Tomi
The invention relates to a coding system that simultaneously provides both compression and encryption. Systems combining compression and encryption do exist, the innovation here is to devise a single process the output of which is both reduced in space and secure against unauthorized decoding. Contrary to our previous work, in which this was achieved by controlling the coding model by means of a secret key, the present work controls the code generation itself.
Pinkas Binyamin
The invention is about handling data that is shared between several servers such that no server learns the contents of the data. The new invention is a method for verifying that two sets of data are equal. The goal of this comparison is to ensure that none of the servers modified the data. This verification is critical when applying different database operations to the data, in the context of sharing a database of sensitive data between the servers such that none of them learns what’s in the database, and yet the servers are able to run database queries on this shared data.
Lindell Yehuda
High Throughput Secure Multi-Party Computation With Identifiable Abort
Lindell Yehuda
Impersonation Detection
Shor Joseph
A new concept for Physical Unclonable Functions (PUFs), the Mirror PUF, is proposed. The Mirror PUF can be applied to existing preselected PUF circuits and has the potential to double the number of effective bits, with no additional area at the bit-cell level. The Mirror PUF utilizes a preselection test, which measures the amount of mismatch within the PUF cell. This data is generally used to determine the mask for a conventional PUF. Here, it is used as an entropy source that is not correlated to the original response of the PUF. Bit-cells with low mismatch are considered as a ‘0’ and cells with high mismatch as a ‘1’. A systematic method is shown for identifying unstable bits in the Mirror PUF response. It classifies the cells to ‘low’, ‘medium’ and ‘high’ mismatch, such that medium mismatch bits are considered as unstable and masked from the response. The concept was applied to a 65nm Si implementation of the Capacitive Tilt PUF [1]. All of the unstable cells of the Mirror PUF, except for 0.03%, were identified and masked, with only 0.0003% erroneous responses. The application of the Mirror PUF to the Capacitive tilt PUF increased the number of stable response bits by 63%. No observable correlation was found between the responses of the original PUF and the Mirror PUF.
Lindell Yehuda
Multiparty Computation Method
Asharov, Gilad
The present disclosure relates to non-interactive zero-knowledge proof systems, and more particularly to non-interactive zero-knowledge proof systems that provide meaningful notions of privacy and soundness even when the common reference string is maliciously generated.
Cohen Eliahu
Systems and methods are described for secure communication to facilitate encrypted transmission of data between a transmitting device (encoder) and a receiving device (decoder), leveraging quandle algebra. An example system includes an encoder, a decoder, and a communication channel. The encoder may generate a ciphertext (c) based on a message (x), an encoding variable (y), and a public encryption key (e), wherein, c=x▹y. The cipher text (c) is then transmitted, via the communication channel, to the decoder. The decoder may receive the ciphertext (c) via the communication channel and generate a deciphered form (x’) of the message (x) based on the ciphertext (c), the encoding variable (y), and a private encryption key (f), wherein, x^'=c◃y, and ▹and ◃ are binary operations that satisfy axioms of a quandle and/or a rack. The proposed methods inherits the strength of the famous RSA scheme and improves upon it using a multi-step process involving rational numbers instead of integers.
Singer Gonen
Transformers play a central role in modern artificial intelligence, yet their susceptibility to adversarial perturbations raises serious reliability concerns. Current defenses, such as adversarial training, are often computationally expensive and tailored to specific attack vectors, while formal verification methods remain difficult to scale. In this work, we propose RAHP (Robustness-Aware Head Pruning), a framework that enhances the intrinsic robustness of Transformers by selectively removing attention heads that contribute to model adversarial vulnerability. Unlike standard pruning, which often degrades robustness, RAHP guides the pruning process using a composite score of two complementary signals: (i) Fisher Information, which preserves task accuracy, and (ii) CLEVER, a sensitivity-based proxy derived from the local Lipschitz constant that estimates the model’s vulnerability to perturbations. By pruning attention heads that exhibit high adversarial sensitivity, RAHP steers the model toward a more stable decision boundary without the need for costly adversarial retraining. Extensive experiments demonstrate that RAHP yields compact models that are not only efficient but also more resistant to a wide variety of attacks compared to standard pruning and regularization baselines. These results suggest that incorporating local stability criteria into the pruning process provides a scalable and attack-agnostic pathway toward robust and efficient Transformer models.
Keshet Joseph
There is provided a system for computing a secure statistical classifier, comprising: at least one hardware processor executing a code for: accessing code instructions of an untrained statistical classifier, accessing a training dataset, accessing a plurality of cryptographic keys, creating a plurality of instances of the untrained statistical classifier, creating a plurality of trained sub-classifiers by training each of the plurality of instances of the untrained statistical classifier by iteratively adjusting adjustable classification parameters of the respective instance of the untrained statistical classifier according to a portion of the training data serving as input and a corresponding ground truth label, and at least one unique cryptographic key of the plurality of cryptographic keys, wherein the adjustable classification parameters of each trained sub-classifier have unique values computed according to corresponding at least one unique cryptographic key, and providing the statistical classifier, wherein the statistical classifier includes the plurality of trained sub-classifiers.
Cohen Eliahu
The proposed method enables to infer, with various degrees of disturbance, the information content of quantum channels. As a possible hacking and security analysis method it exposes and utilizes physical imperfections within common quantum key distribution protocols. It is based on noise injection followed by quantum weak measurements which allow to infer, to some extent, the bits of a distributed secret key. The method can be used for detecting and quantifying the severity of security vulnerabilities within quantum channels. As another application, the proposed method can monitor the performance and error rate of quantum computers.
