Find research datasets worth reusing
Search datasets from major research repositories and use ShareScore to quickly assess how well each record supports discovery, access, and reuse.
74
datasets available to search
ShareScore release 0.9.0
Dataset results
74 results for “hardware”
Power, performance and system measures of HPC benchmarks on multiple hardware
Open the record for dataset details and reuse information.
Example results for use-cases of DAVOS toolkit (dependability assessment, verification, optimisation and selection of hardware models)
<p>This dataset exemplifies the results that can be obtained for several basic experimentation scenarios by means of DAVOS toolkit, available under MIT licence at https://github.com/IlyaTuzov/DAVOS</p> <p>Particular experimentation scenarios are:</p> <p>- Dependability assessment (LEON3 processor core)<br> - Dependability benchmarking of implementation alternatives (MC8051 processor core)<br> - Dependability-aware design space exploration (when implementing PIC core by Xilinx ISE toolkit)</p> <p><br> <strong>Installation steps.</strong></p> <p>1. Ensure to have python ver. 2.x installed. Type in terminal (cmd console in Windows): “python --version” – if the output looks like > Python 2.x.x – python is installed. <br> Otherwise download and install 2.x.x distribution: https://www.python.org/<br> Add python installation path to environment path variable.</p> <p><br> 2. Ensure to have Web-Server installed (Apache preferable). For instance, XAMPP: https://www.apachefriends.org/index.html</p> <p> <br> 3. Ensure that Web-server is configured to execute CGI scripts, particularly python-scripts:<br> In the 'httpd.conf' file (XAMMP control panel – button config in front of apache module):<br> </p> <p>– search for line Options Indexes FollowSymLinks and add ExecCGI, so the resulting line looks like this: <br> Options Indexes FollowSymLinks ExecCGI<br> – search for #AddHandler cgi-script .cgi, uncomment (remove #), and append “.py” to this line, so the results looks like:<br> AddHandler cgi-script .cgi .pl .asp .py </p> <p>4. Unpack the contents of *.zip package into the folder on the Web Server. <br> For instance into 'Web-server root folder'/Dataset.<br> The Web-Server root can be configured in the ‘httpd.conf’ file in the DocumentRoot section, for instance: <br> DocumentRoot "F:/HTWEB"<br> <Directory "F:/HTWEB"><br> ...</p> <p>5. In the web-browser navigate to the root directory of extracted package:<br> http://localhost/Dataset/index.html<br> </p>
Phytoplankton Dataset: Supporting HAB monitoring based on cost-effective open-hardware solutions
<p>ASTRAL Phytoplankton monitoring dataset which has supported technology validation based on images from single-specie phytoplankton sample and prototype deployment within IMTA labs</p>
Microplastic Dataset: Supporting microplastic monitoring based on cost-effective open-hardware solutions
<p>ASTRAL Dataset related to microplastic monitoring based on the open-hardware initiative PlanktoScope</p>
Hardware perceptron and ReLU measurements and results
<p>Recently, in-memory analog computing through memristive crossbar arrays attracted a lot of attention due to its efficient power consumption, area, and computing throughput. Using this computing method, different types of neural networks can be implemented for different applications. In such neural networks, memristors represent the synapses. However, in previous work, digital processors have been used to implement the activation functions or neurons. Implementing neurons using analog-based hardware further improves the power consumption, area, and throughput by removing unnecessary data conversions and communication. </p> <p>In this study, we designed a ReLU activation function and built a fully hardware-based two-layer fully connected perceptron using memristive arrays, and verified the operation by classifying downsampled MNIST images. We measured the DC and AC characteristics of our designed ReLU, forming, set, and reset behavior of the memristive arrays, and the perceptron behavior during training and inference in the classification task. We also studied the non-idealities related to both the ReLU design and memristors which is significantly critical in future integrated designs. Moreover, the downsampled 8*8 MNIST images that we generated from the original MNIST dataset are included in the data which can be used in future studies with the limited size of the network.</p>
Secure Video Streaming Using Dedicated Hardware
<p>Demo video of Secure Video Streaming Using Dedicated Hardware</p>
Data for "Relaxing Hardware Requirements for Surface Code Circuits using Time-dynamics"
<p>Includes the circuit files for each benchmarked stim circuit (circuits.zip), the raw statistics for each memory experiment (stats.csv), the 'fused' stats combining the X and Z error rates to represent the liklihood that neither logical observable fails (fused_stats.csv), the svg files for the main figures in the manuscript (svg_figures.zip), and a snapshot of the full code repo (midout.zip)</p>
Transfer-Path-Based Hardware-Reuse Strong PUF Achieving Modeling Attack Resilience With >200 Million Training CRPs [DATASET]
<p>This is the CRP dataset for the IEEE TIFS paper titled "Transfer-Path-Based Hardware-Reuse Strong PUF Achieving Modeling Attack Resilience With >200 Million Training CRPs". It contains 2 packages with details as below:</p> <p>Package 1: "CRPs_14k_2048TPs_bit5_bit6.zip"<br> Description: It consists of 2048*3 files, which contains all the CRPs extracted from 2048 TPs with DOA disable. Using these data, we can train each ANN model for each TP(i,j), and the corresponding prediction results are shown in Fig. 14 and Fig. 15 in the paper.</p> <p>- "tdc_stim_i_j.csv" is the 64-bit challenges of TP(i,j).<br> - "tdc_resp_F1_i_j_bit5.csv" is the responses of TP(i,j) with bit(5) of TDC used as PUF response<br> - "tdc_resp_F1_i_j_bit6.csv" is the responses of TP(i,j) with bit(6) of TDC used as PUF response</p> <p>Package 2: "CRPs_DOA_330818x1024_bit5_bit6.zip"<br> Description: It consists of 3 files. Using the dataset, we trained each ANN model for each bit in RSobf, and the prediction results of all the ANN models are shown in Fig.17. Notice that the length of RS is dependent on the specific challenge. Here, only those RS consisting of ≥1024 responses are selected and only the first 1024 bits in the selected RS are used for authentication accordingly.</p> <p>- "chal_330818.csv" is the 330818 64-bit challenges.<br> - "resp_330818x1024_bit5.csv" is the 330818 1024-bit obfuscated response streams (RSobf) extracted from 2048 TPs using bit(5) of TDC.<br> - "resp_330818x1024_bit6.csv" is the 330818 1024-bit RSobf extracted from 2048 TPs using bit(6) of TDC.<br> </p>
[DATASET] Supporting Open Science Hardware in Academia: Policy Recommendations for Science Funders and University Managers
<p>Raw data from a real-time Delphi exercise used to write the policy brief: <a href="https://zenodo.org/record/8030029">Supporting Open Science Hardware in Academia: Policy Recommendations for Science Funders and University Managers</a></p> <p>Access to the Real-Time Delphi platform was kindly provided by <a href="https://4strat.de">4strat</a>. See tab "meta - start here" for description of the dataset.</p>
Hardware perceptron and ReLU measurements and results
Open the record for dataset details and reuse information.
Leveraging Hardware Probes and Acceleration for Testing Heterogeneous Applications
<p>The part of data and code for HFuzz: Leveraging Hardware Probes and Acceleration for Testing Heterogeneous Applications</p>
CNF Encoded Isomorphic and Optimized Miters from Hardware Model Checking Competition 2012 Models
<p>These miter benchmarks are used in a paper on congruence closure in 2024.</p> <p>The miter AIGs in the [`aig`](aig) directory are the same as those used to<br>generate CNFs of miter benchmarks submitted to the SAT Challenge in 2012.<br>Some of them have regularly been used in the SAT Competitions since then.</p> <p>The AIGs are separated into two sets:</p> <p>- [`aig/iso`](aig/iso) miters of isomorphic circuits<br>- [`aig/opt`](aig/opt) miters of optimized versus original circuit</p> <p>The first set checks equivalence of one circuit with itself in the<br>[`aig/iso`](aig/iso) (isomorphic) sub-directory and second set contains<br>equivalence checking problems (aka miters) comparing the original circuit<br>with an optimized version in the [`aig/opt`](aig/opt) (optimized)<br>sub-directory. The optimized versions of the circuits were obtained in 2012<br>by the `dc2` script for [ABC](https://people.eecs.berkeley.edu/~alanmi/abc).<br>We did not re-run ABC with a newer version in order to make sure we have as<br>base-line in our experiments the same CNFs, as they are well-known, i.e., <br>they have been used in the SAT competitions for several years.</p> <p>Note that the original AIGER models have state elements (called "latches"<br>in AIGER terminology), but both the optimization with ABC as well as our<br>equivalence checking is purely combinational, by treating latches as<br>additional pseudo inputs, and next-state functions as output-functions. The<br>miters were accordingly generated (by the 2011 version) of `aigmiter` using<br>the `-c` option. For the same reason as for ABC explained above, we did not<br>rerun `aigmiter` either.</p> <p>We then encoded the two sets of AIGs into CNF in two different ways.<br>The first encoding just uses a simple Tseitin encoding, without<br>Plaisted-Greenbaum optimization, i.e., using the `--no-pg` option, as in 2012.<br>Even though we use a new version of `aigtocnf` in `src/aiger/aigtocnf.c` we<br>made sure that identical CNFs are generated for this first encoding variant<br>by checking that the MD5 sum of the (actually compressed) CNF files match.<br>This required to disable a newly introduced cone-of-influence (COI)<br>optimization with `--no-coi` which skips unreachable gates even when the<br>Plaisted-Greenbaum optimization is disabled.</p> <p>The second variant of the encoding detects XOR and ITE gates by pattern<br>matching and produces more concise CNFs by skipping internal AND gates<br>of detected XOR and ITE gates producing a direct encoding instead.</p> <p>This gives the following sets of CNFs;</p> <p>- [`cnf/ands/iso`](cnf/ands/iso) isomorphic circuits with ANDs only<br>- [`cnf/ands/opt`](cnf/ands/opt) optimized versus original with ANDs only<br>- [`cnf/xits/iso`](cnf/xits/isor) isomorphic circuits with ANDs, XORs, ITEs<br>- [`cnf/xits/opt`](cnf/xits/opt) optimized versus original with ANDs, XORs, ITEs</p> <p>All the CNFs in these directories have the same original file name of the<br>AIGER model to simplify run-time comparison. In `cnf/all` we further<br>provide links with qualified names to distinguish them.</p>
Performance Measurement Datasets of the HPC Benchmarks LAMMPS, MiniFE, LULESH for Hardware Counter Variance Analysis
Open the record for dataset details and reuse information.
Data Artifact for "Gaze into the Pattern: Characterizing Spatial Patterns with Internal Temporal Correlations for Hardware Prefetching"
<p>This dataset contains the Ligra, PARSEC, GAP, and QMM traces used in our paper "Gaze into the Pattern: Characterizing Spatial Patterns with Internal Temporal Correlations for Hardware Prefetching", which is accepted by HPCA'25. These traces are provided as part of the AE proces.</p>
Stim circuits for 'Accommodating Fabrication Defects on Floquet Codes with Minimal Hardware Requirements' manuscript
<p>Example Stim circuits for Honeycomb quantum memory experiments with defective qubits.</p> <p>Because accommodating each sample of fabrication defects requires a separate Stim circuit, we only provide example circuits rather than all Stim circuits used in simulations.</p> <p>Please note that the example circuits presented here are constructed by sampling defective qubits according to an iid distribution, with a single parameter defining the probability of an individual qubit being defective. In general a circuit which has a higher probability of each qubit being defective will perform worse than a circuit which has a lower probability of each qubit being defective. However, this does not necessarily mean that the specific circuits presented here will see that behaviour, as these are just individual samples from a large distribution.</p> <p>V2: Uploaded more example circuits with a higher target distance.</p>
Datasets for experiments in support of "HADA: an Automated Tool for Hardware Dimensioning of AI Applications"
<p>The zip archive contains three datasets used during the experimental phase of the paper:</p> <ul> <li><em>ANTICIPATE_trainDataset.csv</em>: used in order to train the ML models for the ANTICIPATE algorithm (Section 4.1);</li> <li><em>CONTINGENCY_trainDataset.csv</em>: used in order to train the ML models for the CONTINGENCY algorithm (Section 4.1);</li> <li><em>EmpiricalValidationSet.csv</em>: used for validating the EML optimization model (Section 4.2)</li> </ul>
Quantum Hardware-Enabled Molecular Dynamics via Transfer Learning
<p>This zip file contains datasets for training Behler-Parrinello neural networks using the n2p2 software. Both DFT and VQE training sets are given for both the water monomer and dimer dataset.</p>
CNF Encoded Isomorphic and Optimized Miters from Hardware Model Checking Competition 2020 Models
<p>From the Hardware Model Checking Competition 2020 we have collected 324<br>sequential model checking problems and for each generated an isomorphic and<br>an optimized miter. The isomorphic miters just compares two identical<br>copies while for the optimized miter one copy went through optimization<br>with ABC using the `dc2` command.</p> <p>The CNFs are generated with a new version of `aigtocnf` which detects<br>XOR and ITE gates in the AIGER circuit and if detected uses a more compact<br>encoding (4 clauses clauses instead of 9) for each detected gate.</p>
Covert Communication Channels based on Hardware Trojans: open-source dataset
<h1>Dataset of Hardware-Trojan (HT) based Covert Channels (HT-CCs) for the IEEE 802.11 (WiFi) standard</h1> <div> <h2>Datasets arhitecture</h2> <ul> <li>The dataset has 80 dataset elements organized in 5 classes, namely CC-free (HT0-CC) and CC-infected (HTX-CC, X={1, · · · , 4}), each class has 16 elements.</li> <li>Those 16 elements are 8 acquisitions for different SNR values ranging from 1dB to 29dB with a step of 4dB, 2 acquisitions for each SNR value. The second acquisition was obtained immediately after the first one, with no other parameters changing between one acquisition and the other. The second acquisition is marked with an underscore in the naming of the file.</li> <li>Each dataset element has 2000 frames and it is the concatenation of ten 200-frames sub-acquisitions.</li> </ul> </div> <div> <h3>Naming convention</h3> </div> <p>Files are named according to the following convention:</p> <p>First acquisition:</p> <div> <pre><code>rxSig_<SNR value>dB_HT<HT-CC attack>.mat </code></pre> <div> </div> </div> <p>Second acquisition:</p> <div> <pre><code>rxSig_<SNR value>dB_HT<HT-CC attack>_.mat </code></pre> <div> </div> </div> <ul> <li>SNR value: ranging from 1dB to 29dB with a step of 4dB, i.e., <1>=SNR of 1db, <2>=SNR of 5db, <3>=SNR of 9dB, <4>=SNR of 13dB, <5>=SNR of 17dB, <6>=SNR of 21dB, <7>=SNR of 25dB, <8>=SNR of 29dB.</li> <li>HT-CC attack: CC-free (HT0-CC) and CC-infected (HTX-CC, X={1, · · · , 4}).</li> </ul> <div> <h3>HT0-CC</h3> </div> <p>The HT0-CC is the benchmark CC-free transmission according to the WiFi standard.</p> <div> <h3>HT1-CC</h3> </div> <p>The HT1-CC is the Amplitude Modulation (AM) Short Training Sequence (STS) HT attak based on the paper</p> <blockquote> <p>A. R. Díaz-Rizo, H. Aboushady and H.-G. Stratigopoulos, "Leaking Wireless ICs via Hardware Trojan-Infected Synchronization," in IEEE Transactions on Dependable and Secure Computing, vol. 20, no. 5, pp. 3845-3859, 1 Sept.-Oct. 2023, doi: 10.1109/TDSC.2022.3218507.</p> </blockquote> <p>The amplitude modulation value alpha is set to 10%.</p> <div> <h3>HT2-CC</h3> </div> <p>The HT2-CC is the PSK STF HT attack based on the paper</p> <blockquote> <p>J. Classen, M. Schulz and M. Hollick, "Practical covert channels for WiFi systems," 2015 IEEE Conference on Communications and Network Security (CNS), Florence, Italy, 2015, pp. 209-217, doi: 10.1109/CNS.2015.7346830.</p> </blockquote> <p>The implemented version is leaking 8 bits per OFDM PPDU.</p> <div> <h3>HT3-CC</h3> </div> <p>The HT3-CC is the Dirty Constellation attack based on paper</p> <blockquote> <p>A. Dutta, D. Saha, D. Grunwald and D. Sicker, "Secret Agent Radio: Covert Communication through Dirty Constellations," in Proceedings of the 14th international conference on Information Hiding (IH'12). Springer-Verlag, Berlin, Heidelberg, 160–175, doi: 10.1007/978-3-642-36373-3_11.</p> </blockquote> <p>The implemented version of this attack has an embedding frequency of 5 subcarriers per OFDM symbol, where each OFDM symbol comprises 48 data subcarriers. Each dirty subcarrier is leaking 2 bits; therefore, we are leaking 10 bits per OFDM symbol,</p> <div> <h3>HT4-CC</h3> </div> <p>The HT4-CC is the AM analog/RF attack based on paper</p> <blockquote> <p>K. S. Subramani, N. Helal, A. Antonopoulos, A. Nosratinia and Y. Makris, "Amplitude-Modulating Analog/RF Hardware Trojans in Wireless Networks: Risks and Remedies," in IEEE Transactions on Information Forensics and Security, vol. 15, pp. 3497-3510, 2020, doi: 10.1109/TIFS.2020.2990792.</p> </blockquote> <p>Due to SPI limitations, the implemented version of the attack is a digitally-emulated version of the AM analog/RF attack. It leaks 8 bits per transmitted frame.</p> <div> <h2>Hardware Platform</h2> </div> <p>Acquisitions are received downconverted IQ samples using a Software Defined Radio (SDR) board bladeRF xA9.</p>
Common in Research Software, Hardware and Data: the experience of the SEEKCommons project (recording)
<p>Luis Felipe R. Murillo, professor of anthropology at University of Notre Dame, delivered this keynote address on August 1st, 2024, at the Ethical Open Science for Past Global Change Data 2024 Symposium, in Keshena, Wisconsin, on the lands of the Menominee Nation. </p>
ScienceDex guides
Understand access before you commit
These curated guides explain access requirements, typical timelines, costs, and reuse considerations for widely used research datasets.
Allen Brain Atlas
Allen Brain Atlas is an Allen Institute collection of brain map atlases, datasets, APIs, and analysis tools covering mouse, human, and non-human primate brain resources.
Annotated Behaviour and Observability Dataset (ABODe)
ABODe is a University of Edinburgh DataShare dataset for behavior classification in group-housed mice using home-cage video, identities, bounding boxes, ground-plate positions, and annotator labels.
DANDI Archive for NWB datasets
DANDI is a BRAIN Initiative archive for publishing and sharing neurophysiology data, including electrophysiology, optophysiology, and behavioral data packaged as NWB and related standards.
International Brain Laboratory public data
The International Brain Laboratory public data releases expose standardized mouse decision-making experiments, including Neuropixels recordings, widefield calcium imaging, behavior, and session metadata accessed through the ONE API.
OpenNeuro
OpenNeuro is a free, open platform for sharing neuroimaging datasets, with public search, dataset pages, and download paths for web, S3, DataLad, and the OpenNeuro CLI.