Pith. sign in

REVIEW 3 major objections 5 minor 202 references

Edge-Optimized Deep Learning & Pattern Recognition Techniques for Non-Intrusive Load Monitoring of Energy Time Series

T0 review · 3 major / 5 minor · reviewed 2026-08-15 · deepseek-v4-flash

Pith's one-line read Pruning before training finds subnetworks of 5% of the parameters that match full NILM model accuracy, a new Greek 10-second dataset fills the Mediterranean data gap, and a graph-based encoder beats sequential baselines on washing machines.

desk verdict Plegma dataset is a solid regional contribution; the GCN result is invalid as NILM due to test-time leakage, and pruning claims rest on in-sample threshold selection. read the letter →

arxiv 2505.06289 v1 pith:VL6L4JU7 submitted 2025-05-07 cs.LG eess.SPstat.ML

classification cs.LGeess.SPstat.ML
keywords non-intrusiveloadmonitoringenergydisaggregationPlegmadatasetiterativemagnitudepruninggraphneuralnetworksedgedeploymentMediterraneanhouseholdtransformerdecoder
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This thesis sets out to make non-intrusive load monitoring (NILM), inferring which appliance is using what power from a single household meter, practical outside the lab. It argues that real deployment is blocked by two gaps: a data gap, because public datasets come mostly from the USA and UK and miss Mediterranean staples like air conditioners and electric water boilers, and a compute gap, because deep disaggregation models are trained and run on cloud infrastructure, raising cost and privacy concerns. To close the first gap it introduces the Plegma dataset, one year of 10-second readings from 13 Greek households with roughly 218 million measurements. To close the second it claims that iterative magnitude pruning before full training can isolate subnetworks with only 5% of the original parameters that disaggregate as well as the full model, and that a graph-based architecture with a transformer decoder outperforms sequential baselines on multi-state appliances. If these claims hold, appliance-level feedback could run on low-cost devices in the home, including in regions whose consumption patterns current models have never seen.

What carries the argument

The argument runs on two mechanisms. The first is iterative magnitude pruning before training, in the lottery-ticket style: the network is trained for a single epoch, the weights with the smallest L1 norm below a threshold $p_t$ are masked out via a binary mask $\mu \in \{0,1\}^N$ so that $\hat{w} = \mu \odot w$, and the surviving weights are reset to their initial values $w_0$; repeating this for a few rounds $R \ll K$ yields a sparse subnetwork that is then fully trained. This is what lets the scheme claim compute savings in both the training and the inference phase, since the expensive full model is never trained. The second is the entropy-based graph construction for the GCN encoder: aggregate windows are assigned to clusters according to the sample entropy of the target appliance's consumption windows, cluster means seed the node embeddings, Markov-chain transition probabilities between clusters define the directed edge weights, and stacked graph convolutional layers propagate information across nodes before a transformer decoder produces the appliance signal.

What would settle it

Train the GCN-encoder model on UK-DALE houses 1, 3, 4, and 5, then evaluate on house 2 without computing cluster assignments from house 2's appliance-level signal, for instance by reusing the training clusters or clustering the aggregate alone. If the washing-machine F1 collapses from the reported 0.7805 toward the CNN baseline of 0.7261, the claim's test-time dependence on knowing the appliance's entropy is confirmed as load-bearing; if the F1 holds, the graph encoder generalizes without that oracle.

Watch

Extended reading notes

Core claim

The thesis's central claim is that the two obstacles standing between NILM research and real deployment, missing data for Mediterranean consumption patterns and the high compute cost of deep disaggregation models, are both removable. On the data side, it presents the Plegma dataset: a year-long, 10-second-resolution record of aggregate and appliance-level power from 13 Greek households, roughly 218 million readings, covering air conditioners (18 units) and electric water boilers (12 units) that rarely appear in other public datasets, alongside environmental and sociodemographic metadata. On the compute side, it argues that iterative L1-magnitude pruning performed before full training, with surviving weights reset to their initialization, isolates subnetworks containing 5% of the original parameters whose disaggregation performance matches the full model on the UK-DALE benchmark, and that the same idea in structured and dependency-graph form cuts model size up to 90% when evaluated on Plegma. It further claims that a sequence-to-sequence model with a graph convolutional encoder and a transformer decoder is the first graph-based NILM, recovering multi-state appliances like the washing machine better than CNN, LSTM, and GRU baselines (F1 0.7805 versus 0.7261 for the CNN baseline).

Load-bearing premise

The load-bearing premise, disclosed by the thesis itself, is that the graph encoder knows the target appliance's consumption entropy on the test data; in a genuine deployment, where the appliance signal is exactly what is being inferred, that information is not available and the graph cannot be built.

Editorial extensions

If this is right

  • Training NILM models on the edge becomes plausible: if subnetworks found before training match full models, the resource-heavy full model never has to be trained centrally, and the reported 5% parameter subnetworks match full-model performance on UK-DALE.
  • The Plegma dataset gives researchers a Mediterranean benchmark with 10-second granularity, enabling NILM evaluation on air conditioners and electric water boilers and comparisons of cross-region transfer against UK-DALE and REFIT.
  • Structured and dependency-graph pruning variants, validated on Plegma, reach up to 90% model-size reduction and a 10x compute saving, a level suited to Raspberry-Pi-class devices.
  • For multi-state appliances such as washing machines, graph-based encoding with a transformer decoder beats sequential baselines on activation detection, with F1 of 0.7805 compared with 0.7261 for the CNN baseline.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The mask-reset pruning recipe is demonstrated on one CNN architecture; a direct extension is testing whether the same schedule holds for transformer or GCN backbones, whose per-layer sparsity tolerances likely differ.
  • Plegma's combination of UK-DALE-comparable granularity and Mediterranean appliance mix makes a cross-dataset transfer experiment the natural next test: train on UK-DALE, test on Plegma, and quantify how much air-conditioner and boiler patterns degrade models that never saw them.
  • The disclosed test-time-entropy limitation suggests a concrete fix the thesis does not pursue: learning the cluster assignment from the aggregate signal alone, which would make the graph encoder truly non-intrusive.
  • The questionnaire metadata (occupancy, income, heating type) enables conditioning disaggregation on household context, a direction the dataset opens but the thesis leaves unexplored.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 5 minor

Summary. This PhD dissertation combines an interoperable IoT data-collection framework and a new residential energy dataset, the Plegma Dataset from Greece, with a set of deep-learning contributions for non-intrusive load monitoring (NILM). The dataset chapter reports 13 households, 10-second aggregate and appliance-level measurements, roughly 218 million readings, environmental data, and sociodemographic/building metadata, together with explicit cleaning rules (Algorithms 1-2). The modeling chapters claim three advances: a GCN-encoder/transformer-decoder as the first graph-based NILM model, an iterative pre-training pruning scheme ('OPT-NILM') that finds subnetworks matching full-model performance at up to 95% sparsity, and structured dependency-graph pruning evaluated on the Plegma dataset. The pruning chapters report large complexity reductions on UK-DALE and Plegma, while the GCN chapter reports improved washing-machine F1 over CNN/LSTM/GRU baselines.

Significance. If the dataset contribution stands, it is genuinely useful: Plegma is, as claimed, one of the first public 10-second residential datasets from Greece and the Mediterranean, with underrepresented appliances such as air conditioners and electric water boilers, and the chapter is careful about missing data (6.86% NaN, 0.82% issue flags) and about reproducible preprocessing. The pruning chapters address a real deployment bottleneck and the before-training pruning idea is worth pursuing. However, the significance of the modeling claims is currently conditional: the GCN result is not a valid NILM result as evaluated, because it uses test-time appliance entropy, and the pruning 'optimal threshold' is selected on the same test-house curves used for the final reported numbers. These are load-bearing issues, not presentation details.

major comments (3)
  1. [§4.3.3 and §4.1] The GCN encoder constructs graph nodes by assigning aggregate windows to clusters 'based on the entropy metric of the corresponding appliance's consumption windows' (§4.3.3). At inference in the NILM setting defined by Eq. (4.1), only the aggregate x(t) is observed, so the target appliance's entropy is not available; §4.1 concedes that the approach 'could only be used to disaggregate known datasets.' Consequently, the washing-machine comparison in Table 4.2 (F1 0.7805 vs 0.7261 for CNN) is obtained with access to ground-truth appliance information and is not a valid non-intrusive disaggregation result. The claimed advantage over sequential baselines therefore does not support the contribution as stated; a leakage-free variant (e.g., clustering on aggregate-only features or a state estimator trained without test labels) would be needed.
  2. [§5.3.1, Eq. (5.4); §8.4.3] The optimal pruning threshold p̂t is selected by minimizing the distance in Eq. (5.3) over the 'F1 score - MACs' curves, and the experimental setup states that models are trained on houses 1,3,4,5 and tested on house 2. As presented, the F1-MAC curves in Fig. 5.1 are the performance-degradation curves on which the final results in Figs. 5.2-5.3 and Table 6.1 are reported, so p̂t is an in-sample selection rather than an independent model-selection choice. The same issue appears in Chapter 8, where Eq. (8.10) sets Pthr = Popt from the test-house curves in Fig. 8.1. Please use a validation-house split for threshold selection and report test results only for the final threshold, or explicitly label the reported numbers as oracle-selected.
  3. [§5.3.1, Eq. (5.3)] The proposed trade-off metric mixes the F1 score, which lies in [0,1], with raw MAC counts, which in Fig. 5.1 are of order 10^6. As printed, Eq. (5.3) is dist = sqrt((1-F1)^2 + (0-MACs(pt))^2); without a normalization factor for MACs, the distance is dominated by the MAC term, so the 'optimal' threshold is essentially the most aggressive sparsity that keeps F1 above zero. Please specify the scaling (e.g., normalize MACs by the baseline model's MACs) and state the exact form used to produce the reported thresholds; the current formula also appears to contain a typo (missing closing parenthesis).
minor comments (5)
  1. [§3.4, Table 3.6] The table titled 'Monitored appliances in each house' actually reports aggregate consumption, sub-metered consumption, and the percentage of sub-metered consumption; the title appears to be copied from Table 3.3 and should be corrected.
  2. [§4.4.2, Eq. (4.10)] The MRE and MAE definitions are corrupted in the typesetting; as printed, MRE lacks an explicit summation/division structure and the MAE expression is incomplete. Please provide clean equations.
  3. [§1.4 and §3.1] The Plegma dataset is cited as [9] in Chapter 1 but as [10] in Chapter 3; please harmonize the citation numbering throughout the thesis.
  4. [§3.4, Table 3.7] The fridge row reports the number of activations as 'continuous' in a column labeled 'Total number of appliance activations'; please clarify how 'continuous' is defined and whether it is used in downstream analyses.
  5. [§4.4.3, Table 4.1] Table 4.1 reports F1 = 0.00000 for kettle and microwave; the text explains that the model is unsuited to short-duration devices, but reporting precision and recall or activation-level detection metrics would make the failure mode more informative.

Circularity Check

2 steps flagged · score 6.0 of 10

The pruning 'optimal threshold' is selected from the same test performance curves on which final results are reported, and the GCN encoder requires the target appliance's entropy at test time despite NILM being defined from the aggregate alone.

  1. self definitional [Sec. 4.1, Sec. 4.3.3, Eq. (4.1)]
    "The goal of energy disaggregation is to solve the inverse problem and determine the individual consumption ym of appliance m based exclusively on the measurement of the aggregate signal. ... these windows are being assigned to specific clusters based on the entropy metric of the corresponding appliance's consumption windows. ... our approach could only be used to disaggregate known datasets as it provides the limitation of knowing the entropy of the appliance's signal even on the testing data."

    The paper formulates NILM as recovering the appliance signal ym solely from the aggregate signal xagg, but the encoder's graph construction is defined by the entropy of the target appliance's own consumption windows. The clusters that form the graph nodes, the Markov transition weights, and hence the GCN embeddings are all built from the very signal the model is supposed to predict. The reported washing-machine advantage in Table 4.2 is therefore measured with the ground-truth appliance signal available at test time, so the claimed disaggregation result is not derived from the aggregate input alone; the target is built into the encoder by construction.

  2. fitted input called prediction [Sec. 5.3.1 Eqs. (5.3)-(5.4); results in Sec. 5.3.2]
    "First the Euclidean distance for all pt is calculated given an 'F1 score - MACs' diagram as: dist(F1, MACs(pt)) = sqrt((1 - F1)^2 + (0 - MACs(pt))^2) ... p_hat_t = arg min_{pt in (0,0.95)} dist(F1, MACs(pt)). ... The pruning thresholds were set equal to the p_hat_t of the proposed approach: 95% for the kettle, 80% for the dishwasher, 85% for the fridge and 60% for the washing machine."

    The 'optimal pruning threshold' is chosen by minimizing a distance to perfect F1, using F1 values computed from the same test-house performance curves on which the final results are reported. The headline compression claims (95% parameter reduction with negligible degradation) are thus in-sample selections: the threshold is fitted to the test F1 curve and the subsequently reported F1 at that threshold is the same quantity used in the selection rule. The comparison between pre-training and after-training pruning is also evaluated at per-method in-sample-optimal thresholds, so the claimed performance advantage is partly an artifact of each method being scored at its own test-set-selected operating point rather than at an independently fixed sparsity level.

full rationale

The dataset chapter is self-contained against external benchmarks and is not circular: the Plegma dataset is a new measurement campaign with documented hardware, pre-processing, and FAIR hosting, and its uniqueness claims are falsifiable external facts. The GCN chapter, however, contains a load-bearing self-definitional step: the graph encoder needs the entropy of the target appliance's consumption windows even on test data, which contradicts the paper's own Eq. (4.1) definition of disaggregation as recovering ym from the aggregate alone. The paper explicitly admits this limitation, yet the reported superiority over CNN/LSTM/GRU is presented as a NILM result. The pruning chapters select the pruning threshold via Eq. (5.4) from F1 values on the same test curves later used as the reported result; this is a fitted selection criterion renamed as an 'optimal pruning threshold' and then re-reported as independent performance. Chapters 7 and 8 reuse the same Eq. (5.4) selection mechanism, so the issue propagates. I did not count the author's self-citations as circular: they are normal bibliographic markers, and no machine-checked or externally verified 'uniqueness theorem' is invoked to forbid alternatives. The non-finding portions (dataset, hardware architecture) remain independent evidence. Overall, the central predictive claims for the GCN architecture and for the optimal-threshold pruning results reduce in part to their own inputs, giving a partial circularity score of 6.

Assumptions & free parameters 4 free parameters · 5 assumptions · 0 invented entities

The pruning chapters carry two fitted quantities (the per-appliance thresholds and the metric's ideal-point scaling) and import the lottery ticket hypothesis as a domain assumption. The GCN chapter contributes an ad hoc test-time information assumption. No new physical entities, forces, particles, or conserved quantities are introduced; the graph structure and binary masks are methodological constructs, not entities with independent falsifiable handles.

free parameters (4)
  • Per-appliance pruning threshold p̂t (Ch. 5), p̂opt (Ch. 6), ŝ (Ch. 7-8) = 95% (kettle), 80% (dishwasher), 85% (fridge), 60% (washing machine); per-layer ratios in Ch. 7-8 not itemized
    Chosen by minimizing Eq. 5.3-5.4, a Euclidean distance to the ideal point (F1=1, MACs=0) computed on the test-house F1-versus-MACs curves, so the threshold is fitted to the evaluation data.
  • Number of graph clusters (operational states) N for the GCN encoder = not reported
    Sec. 4.3.3 assigns windows to clusters via Lloyd's algorithm but never states the cluster count, which fixes the graph's node set and therefore the learned embeddings; the central Chapter 4 result is not reproducible from the text.
  • Ideal-point scale in the trade-off metric (Eq. 5.3) = (F1=1, MACs=0) with unscaled axes
    Because MACs values are of order 10^6 and F1 is in [0,1], the term (0-MACs)^2 dominates (1-F1)^2, making the argmin in Eq. 5.4 collapse to the maximum pruning level; the normalization, not the data, picks the threshold.
  • GCN layers and transformer layers (Ch. 4) = 8 GCN layers, 2 transformer layers
    Architecture hyperparameters in Sec. 4.3.3 and 4.3.4 chosen without a stated tuning protocol or sensitivity analysis.
assumptions (5)
  • domain assumption Aggregate power is the sum of appliance contributions plus noise (Eq. 4.1, Eq. 5.1, Eq. 6.1)
    Standard NILM additive model adopted without derivation; it is the invertibility premise on which disaggregation rests.
  • domain assumption Randomly initialized NILM networks contain sparse subnetworks that reach full-network accuracy when trained (lottery ticket hypothesis)
    Ch. 5.2.3 states this as 'the foundational hypothesis underpinning our approach', imported from the cited work [60]; the thesis's own experiments are the only NILM-specific evidence provided.
  • ad hoc to paper Appliance operational-state entropy is available at test time for the disaggregated appliance
    Sec. 4.3.3 clusters aggregate windows using the target appliance's own consumption windows, and Sec. 4.1 admits this limits the method to known datasets; this assumption is what makes the GCN encoder constructible.
  • standard math Markov property: the next appliance state depends only on the current state
    Used in Sec. 4.3.3 to build the transition matrix P from cluster sequences; a standard stochastic modeling assumption.
  • standard math GCN message passing (Kipf-Welling layers, Eq. 4.6-4.7) produces representations useful for regression decoding
    Adopted architectural machinery with no proof of sufficiency, consistent with common practice in the graph learning literature.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Edge-Optimized Deep Learning & Pattern Recognition Techniques for Non-Intrusive Load Monitoring of Energy Time Series." pith.science (2026). https://pith.science/paper/VL6L4JU7

@misc{pith2026250506289,
  author       = {Pith},
  title        = {Pith review of: Edge-Optimized Deep Learning & Pattern Recognition Techniques for Non-Intrusive Load Monitoring of Energy Time Series},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/VL6L4JU7}},
  note         = {Machine review of arXiv:2505.06289}
}
read the original abstract

The growing global energy demand and the urgent need for sustainability call for innovative ways to boost energy efficiency. While advanced energy-saving systems exist, they often fall short without user engagement. Providing feedback on energy consumption behavior is key to promoting sustainable practices. Non-Intrusive Load Monitoring (NILM) offers a promising solution by disaggregating total household energy usage, recorded by a central smart meter, into appliance-level data. This empowers users to optimize consumption. Advances in AI, IoT, and smart meter adoption have further enhanced NILM's potential. Despite this promise, real-world NILM deployment faces major challenges. First, existing datasets mainly represent regions like the USA and UK, leaving places like the Mediterranean underrepresented. This limits understanding of regional consumption patterns, such as heavy use of air conditioners and electric water heaters. Second, deep learning models used in NILM require high computational power, often relying on cloud services. This increases costs, raises privacy concerns, and limits scalability, especially for households with poor connectivity. This thesis tackles these issues with key contributions. It presents an interoperable data collection framework and introduces the Plegma Dataset, focused on underrepresented Mediterranean energy patterns. It also explores advanced deep neural networks and model compression techniques for efficient edge deployment. By bridging theoretical advances with practical needs, this work aims to make NILM scalable, efficient, and adaptable for global energy sustainability.

Figures

Figures reproduced from arXiv: 2505.06289 by the authors.

Figure 3
Figure 3. Overview of synchronization issue. Each line represents a sensor, P_agg being [PITH_FULL_IMAGE:figures/full_fig_p052_3.png] view at source ↗
Figure 3
Figure 3. Plegma Data Availability [PITH_FULL_IMAGE:figures/full_fig_p057_3.png] view at source ↗
Figure 3
Figure 3. The power usage for House 1 on the 22nd of August 2023. The space between [PITH_FULL_IMAGE:figures/full_fig_p058_3.png] view at source ↗
Figures from the paper (11 more)
Figure 3
Figure 3. Figure 3: This visualization showcases environmental data collected from House 3 across [PITH_FULL_IMAGE:figures/full_fig_p060_3.png]
Figure 5
Figure 5. Figure 5: Diagrams comparing computational complexity (MACs) for different pruning [PITH_FULL_IMAGE:figures/full_fig_p079_5.png]
Figure 5
Figure 5. Figure 5: Prediction consumption diagrams using the proposed, the after [PITH_FULL_IMAGE:figures/full_fig_p080_5.png]
Figure 6
Figure 6. Figure 6: The comparison of the conventional pruning process (upper) and the proposed [PITH_FULL_IMAGE:figures/full_fig_p083_6.png]
Figure 6
Figure 6. Figure 6: Overview of the OPT [PITH_FULL_IMAGE:figures/full_fig_p090_6.png]
Figure 6
Figure 6. Figure 6: Example of the proposed trade [PITH_FULL_IMAGE:figures/full_fig_p092_6.png]
Figure 6
Figure 6. Figure 6: The proposed CNN seq2seq architecture [PITH_FULL_IMAGE:figures/full_fig_p095_6.png]
Figure 7
Figure 7. Figure 7: Comparison of computational complexity (sparsity %) for different pruning [PITH_FULL_IMAGE:figures/full_fig_p109_7.png]
Figure 7
Figure 7. Figure 7: Prediction consumption using the optimized structured pruning vs. the unstruc [PITH_FULL_IMAGE:figures/full_fig_p110_7.png]
Figure 8
Figure 8. Figure 8: Diagrams compare computational complexity for different pruning thresholds [PITH_FULL_IMAGE:figures/full_fig_p120_8.png]
Figure 8
Figure 8. Figure 8: Comparison of predicted consumption diagrams using [PITH_FULL_IMAGE:figures/full_fig_p121_8.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

202 extracted references · 74 canonical work pages

  1. [1]

    Aeotec home energy meter

    Aeotec (2023). Aeotec home energy meter. https://aeotec.com/products/ aeotec-home-energy-meter

  2. [2]

    and Bons, M

    Ahmed, S. and Bons, M. (2020). Edge Computed NILM: A Phone-Based Implementation Using MobileNet Compressed by Tensorflow Lite, page 44–48. ACM, New York, NY, USA

  3. [3]

    Ai, S., Chakravorty, A., and Rong, C. (2019). Household power demand prediction using evolutionary ensemble neural network pool with multiple network structures. Sensors, 19(3):721

  4. [4]

    A., and Martins, J

    Akbari, S., Lopes, R. A., and Martins, J. (2024). The potential of residential load flexibility: An approach for assessing operational flexibility. International Journal of Electrical Power Energy Systems, 158:109918

  5. [5]

    M., Shahjalal, M., Rahman, M

    Alam, M. M., Shahjalal, M., Rahman, M. H., Nurcahyanto, H., Prihatno, A. T., Kim, Y., and Jang, Y. M. (2022). An energy and leakage current monitoring system for abnormality detection in electrical appliances. Scientific Reports, 12(1):18520

  6. [6]

    Alsalemi, A., Himeur, Y., Bensaali, F., Amira, A., Sardianos, C., Varlamis, I., and Dimitrakopou- los, G. (2020). Achieving domestic energy efficiency using micro -moments and intelligent recommendations. IEEE Access, 8:15047–15055

  7. [7]

    -F., Timplalexis, C., Krinidis, S., Ioannidis, D., and Tzovaras, D

    Angelis, G. -F., Timplalexis, C., Krinidis, S., Ioannidis, D., and Tzovaras, D. (2022). Nilm applications: Literature review of learning approaches, recent developments and challenges. Energy and Buildings, 261:111951

  8. [8]

    Anwar, S., Hwang, K., and Sung, W. (2017). Structured pruning of deep convolutional neural networks. ACM Journal on Emerging Technologies in Computing Systems (JETC), 13(3):1–18

Show all 202 references
  1. [9]

    Athanasoulias, S., Guasselli, F., Doulamis, N., Doulamis, A., Ipiotis, N., Katsari, A., Stankovic, L., and Stankovic, V. (2024a). The plegma dataset: Domestic appliance -level and aggregate electricity demand with metadata from greece. Scientific Data, 11(1):376

  2. [10]

    Athanasoulias, S., Katsari, A., Murray, D., Guasseli, F., Doulamis, A., Doulamis, N., Ipiotis, N., Stankovic, L., and Stankovic, V. (2023a). University of Strathclyde, PURE. https://doi.org/10. 15129/3b01a6c6-2efd-424a-b8b8-5fe7fa445ded

  3. [11]

    Athanasoulias, S., Katsari, A., Savvakis, M., Kalogridis, S., and Ipiotis, N. (2023b). An interoperable and cost-effective iot-based framework for household energy monitoring and analysis. In Proceedings of the 16th International Conference on PErvasive Technologies Related to...

  4. [12]

    Athanasoulias, S., Sykiotis, S., Kaselimi, M., Doulamis, A., Doulamis, N., and Ipiotis, N. (2023c). Opt-nilm: An iterative prior-to-full-training pruning approach for cost-effective user side energy disaggregation. IEEE Transactions on Consumer Electronics, pages 1–1

  5. [13]

    Athanasoulias, S., Sykiotis, S., Kaselimi, M., Doulamis, A., Doulamis, N., and Ipiotis, N. (2024b). Opt-nilm: An iterative prior-to-full-training pruning approach for cost-effective user side energy disaggregation. IEEE Transactions on Consumer Electronics, 70(1):4435–4446

  6. [14]

    Athanasoulias, S., Sykiotis, S., Kaselimi, M., Protopapadakis, E., and Ipiotis, N. (2022). A first approach using graph neural networks on non-intrusive-load-monitoring. In Proceedings of the 15th International Conference on PErvasive Technologies Related to Assistive Environm...

  7. [15]

    Athanasoulias, S., Sykiotis, S., Temenos, N., Doulamis, A., and Doulamis, N. (2024c). A pre- training pruning strategy for enabling lightweight non-intrusive load monitoring on edge devices. In 2024 IEEE International Conference on Acoustics, Speech, and Signal Processing Work...

  8. [16]

    Athanasoulias, S., Temenos, N., Doulamis, N., Doulamis, A., Kokos, I., and Ipiotis, N. (2024d). Towards edge-computed nilm: Insights from a mediterranean use case. In 2024 3rd International Conference on Energy Transition in the Mediterranean Area (SyNERGY MED), pages 1–5

  9. [17]

    Azarian, K., Bhalgat, Y., Lee, J., and Blankevoort, T. (2020). Learned threshold pruning. arXiv preprint arXiv:2003.00075, 1

  10. [18]

    Bakalos, N., Voulodimos, A., Doulamis, N., Doulamis, A., Ostfeld, A., Salomons, E., Caubet, J., Jimenez, V., and Li, P. (2019). Protecting water infrastructure from cyber and physical threats: Using multimodal data fusion and adaptive deep learning to monitor critical systems....

  11. [19]

    Barber, J., Cuayáhuitl, H., Zhong, M., and Luan, W. (2020). Lightweight NILM Employing Pruned Sequence-to-Point Learning, volume 1, page 11–15. Association for Computing Machinery, New York, NY, USA

  12. [20]

    Bastings, J., Titov, I., Aziz, W., Marcheggiani, D., and Sima’an, K. (2017). Graph convolutional encoders for syntax-aware neural machine translation. arXiv preprint arXiv:1704.04675

  13. [21]

    Batic, D., Stankovic, V., and Stankovic, L. (2024). Toward transparent load disaggregation—a framework for quantitative evaluation of explainability using explainable ai. IEEE Transactions on Consumer Electronics, 70(1):4345–4356

  14. [22]

    Batic, D., Tanoni, G., Stankovic, L., Stankovic, V., and Principi, E. (2023). Improving knowledge distillation for non-intrusive load monitoring through explainability guided learning. In ICASSP 2023 - 2023 IEEE International Conference on Acoustics, Speech and Signal Processi...

  15. [23]

    Batra, N., Kukunuri, R., Pandey, A., Malakar, R., Kumar, R., Krystalakos, O., Zhong, M., Meira, P., and Parson, O. (2019). Towards reproducible state -of-the-art energy disaggregation. In Proceedings of the 6th ACM Int Conf on Systems for Energy -Efficient Buildings, Cities, a...

  16. [24]

    M., Gutierrez-Villalobos, J

    Bautista-Villalon, M. M., Gutierrez-Villalobos, J. M., and Rivas -Araiza, E. (2018). Iot-based system to monitor and control household lighting and appliance power consumption and water demand. In 2018 7th International Conference on Renewable Energy Research and Applications ...

  17. [25]

    Beckel, C., Kleiminger, W., Cicchetti, R., Staake, T., and Santini, S. (2014). The eco data set and the performance of non-intrusive load monitoring algorithms. In Proceedings of the 1st ACM conference on embedded systems for energy-efficient buildings, pages 80–89

  18. [26]

    Bengio, Y., Lecun, Y., and Hinton, G. (2021). Deep learning for ai. Communications of the ACM, 64(7):58–65

  19. [27]

    and Plungklang, B

    Biansoongnern, S. and Plungklang, B. (2016). Non-intrusive appliances load monitoring (nilm) for energy conservation in household with low sampling rate. Procedia Computer Science, 86:172– 175

  20. [28]

    and Lewis, A

    Brandon, G. and Lewis, A. (1999). Reducing household energy consumption: A qualitative and quantitative field study. Journal of Environmental Psychology, 19(1):75–85

  21. [29]

    Buddhahai, B., Wongseree, W., and Rakkwamsuk, P. (2020). An energy prediction approach for a nonintrusive load monitoring in home appliances. IEEE Transactions on Consumer Electronics, 66(1):96–105

  22. [30]

    Cai, W., Wang, L., Li, L., Xie, J., Jia, S., Zhang, X., Jiang, Z., and Lai, K.-h. (2022). A review on methods of energy performance improvement towards sustainable manufacturing from perspectives of energy monitoring, evaluation, optimization and benchmarking. Renewable and Su...

  23. [31]

    Carrie Armel, K., Gupta, A., Shrimali, G., and Albert, A. (2013). Is disaggregation the holy grail of energy efficiency? the case of electricity. Energy Policy, 52:213–234. Special Section: Transition Pathways to a Low Carbon Economy

  24. [32]

    A., Camarda, C., Macii, E., and Patti, E

    Castangia, M., Sappa, R., Girmay, A. A., Camarda, C., Macii, E., and Patti, E. (2022). Anomaly detection on household appliances based on variational autoencoders. Sustainable Energy, Grids and Networks, 32:100823

  25. [33]

    M., and Pelillo, M

    Castellano, G., Fanelli, A. M., and Pelillo, M. (1997). An iterative pruning algorithm for feedforward neural networks. IEEE transactions on Neural networks, 8(3):519–531

  26. [34]

    Çavdar, ˙I. H. and Faryad, V. (2019). New design of a supervised energy disaggregation model based on the deep neural network for a smart grid. Energies, 12(7):1217

  27. [35]

    Chadoulos, S., Koutsopoulos, I., and Polyzos, G. C. (2021). One model fits all: Individualized household energy demand forecasting with a single deep learning model. In Proceedings of the Twelfth ACM International Conference on Future Energy Systems, e-Energy ’21, page 466–474...

  28. [36]

    Chalmers, C., Hurst, W., Mackay, M., and Fergus, P. (2016). Smart monitoring: an intelligent system to facilitate health care across an ageing population. In EMERGING 2016: The Eighth International Conference on Emerging Networks and Systems Intelligence, pages 34–39. IARIA XPS Press

  29. [37]

    Chavan, D. R. and More, D. S. (2022). A systematic review on low-resolution nilm: Datasets, algorithms, and challenges. Electronic Systems and Intelligent Computing: Proceedings of ESIC 2021, pages 101–120. 117

  30. [38]

    R., More, D

    Chavan, D. R., More, D. S., and Khot, A. M. (2022). Iedl: Indian energy dataset with low frequency for nilm. Energy Reports, 8:701–709

  31. [39]

    Chavat, J., Nesmachnow, S., Graneri, J., and Alvez, G. (2022). Ecd-uy, detailed household electricity consumption dataset of Uruguay. Scientific Data, 9(1):21

  32. [40]

    M., Yu, X., and Jalili, M

    Chen, Z., Amani, A. M., Yu, X., and Jalili, M. (2023). Control and optimisation of power grids using smart meter data: A review. Sensors, 23(4):2118

  33. [41]

    Cheng, Y., Wang, D., Zhou, P., and Zhang, T. (2017). A Survey of Model Compression and Acceleration for Deep Neural Networks. IEEE Signal Processing Magazine, 35

  34. [42]

    Commission, E., for Energy, D.-G., Alaton, C., and Tounquet, F. (2020). Benchmarking smart metering deployment in the EU-28 – Final report. Publications Office

  35. [43]

    Fit for 55: Delivering the eu’s 2030 climate target on the way to climate neutrality

    Council of the European Union (2024). Fit for 55: Delivering the eu’s 2030 climate target on the way to climate neutrality. Accessed: 2024-11-19

  36. [44]

    P., and Parizi, R

    Dabbagh, M., Lee, S. P., and Parizi, R. M. (2016). Functional and non-functional requirements prioritization: empirical evaluation of ipa, ahp-based, and ham-based approaches. Soft computing, 20:4497–4520

  37. [45]

    K., Zohourian, A., Truong, K

    Dadkhah, S., Mahdikhani, H., Danso, P. K., Zohourian, A., Truong, K. A., and Ghorbani, A. A. (2022). Towards the development of a realistic multidimensional iot profiling dataset. In 2022 19th Annual International Conference on Privacy, Security & Trust (PST), pages 1–11. IEEE

  38. [46]

    Danbatta, S. J. and Varol, A. (2019). Comparison of zigbee, z-wave, wi-fi, and bluetooth wireless technologies used in home automation. In 2019 7th International Symposium on Digital Forensics and Security (ISDFS), pages 1–5

  39. [47]

    Deese, A. S. and Daum, J. (2018). Application of zigbee-based internet of things technology to demand response in smart grids. IFAC-PapersOnLine, 51(28):43–48. 10th IFAC Symposium on Control of Power and Energy Systems CPES 2018

  40. [48]

    Dey, B., Basak, S., and Bhattacharyya, B. (2023). Demand-side-management-based bi-level intelligent optimal approach for cost-centric energy management of a microgrid system. Arabian Journal for Science and Engineering, pages 1–12

  41. [49]

    and Cheng, X

    Dinculeana˘, D. and Cheng, X. (2019). Vulnerabilities and limitations of mqtt protocol used between iot devices. Applied Sciences, 9(5):848

  42. [50]

    and Cheng, X

    Dinculeana˘, D. and Cheng, X. (2019). Vulnerabilities and limitations of mqtt protocol used between IoT devices. Applied Sciences, 9(5)

  43. [51]

    Doulamis, A. D. and Doulamis, N. D. (2004). Generalized nonlinear relevance feedback for interactive content-based retrieval and organization. IEEE transactions on circuits and systems for video technology, 14(5):656–671

  44. [52]

    D., Doulamis, A

    Doulamis, N. D., Doulamis, A. D., and Varvarigos, E. (2018). Virtual associations of prosumers for smart energy networks under a renewable split market. IEEE Transactions on Smart Grid , 9(6):6069–6083

  45. [53]

    Elmaz, F., Eyckerman, R., Casteels, W., Latré, S., and Hellinckx, P. (2021). Cnn-lstm architecture for predictive indoor temperature modeling. Building and Environment, 206:108327. 118

  46. [54]

    Guidelines on open access to scientific publications and research data in horizon 2020

    European Commission (2020). Guidelines on open access to scientific publications and research data in horizon 2020. European Commission Research & Innovation

  47. [55]

    Eurostat news: Energy statistics update - june 2024

    Eurostat (2024). Eurostat news: Energy statistics update - june 2024. Accessed: 2024-11-19

  48. [56]

    Fan, C., Li, J., Zhang, T., Ao, X., Wu, F., Meng, Y., and Sun, X. (2021). Layer-wise Model Pruning based on Mutual Information. In EMNLP2021, pages 3079–3090

  49. [57]

    B., and Wang, X

    Fang, G., Ma, X., Song, M., Mi, M. B., and Wang, X. (2023). Depgraph: Towards any structural pruning. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 16091–16101

  50. [58]

    Filip, A. et al. (2011). Blued: A fully labeled public dataset for event-based nonintrusive load monitoring research. In 2nd workshop on data mining applications in sustainability (SustKDD), volume 2012

  51. [59]

    Fischer, C. (2008). Feedback on household electricity consumption: a tool for saving energy? Energy efficiency, 1(1):79–104

  52. [60]

    and Carbin, M

    Frankle, J. and Carbin, M. (2018). The lottery ticket hypothesis: Finding sparse, trainable neural networks. arXiv preprint arXiv:1803.03635

  53. [61]

    Ghorbani, F., Ahmadi, A., Kia, M., Rahman, Q., and Delrobaei, M. (2023). A decision-aware ambient assisted living system with IoT embedded device for in-home monitoring of older adults. Sensors, 23(5):2673

  54. [62]

    Glória, A., Cercas, F., and Souto, N. (2017). Design and implementation of an iot gateway to create smart environments. Procedia Computer Science , 109:568 –575. 8th International Conference on Ambient Systems, Networks and Technologies, ANT-2017 and the 7th International Conf...

  55. [63]

    and Kumar, M

    Gopinath, R. and Kumar, M. (2023). Deepedge-nilm: A case study of non -intrusive load monitoring edge device in commercial building. Energy and Buildings, 294:113226

  56. [64]

    Han, B., Zahraoui, Y., Mubin, M., Mekhilef, S., Seyedmahmoudian, M., and Stojcevski, A. (2023). Home energy management systems: A review of the concept, architecture, and scheduling strategies. IEEE Access

  57. [65]

    Han, J., Choi, C.-s., Park, W.-k., Lee, I., and Kim, S.-h. (2014). Smart home energy management system including renewable energy based on zigbee and plc. IEEE Transactions on Consumer Electronics, 60(2):198–202

  58. [66]

    Han, S., Pool, J., Tran, J., and Dally, W. J. (2015). Learning both weights and connections for efficient neural networks. In Proceedings of the 28th International Conference on Neural Information Processing Systems - Volume 1, NIPS’15, page 1135–1143, Cambridge, MA, USA. MIT Press

  59. [67]

    Hargreaves, T., Nye, M., and Burgess, J. (2010). Making energy visible: A qualitative field study of how householders interact with feedback from smart energy monitors. Energy policy, 38(10):6111–6119

  60. [68]

    Hart, G. W. (1992). Nonintrusive appliance load monitoring. Proceedings of the IEEE , 80(12):1870–1891. 119

  61. [69]

    Hawkins, J. (2017). Special report : Can we copy the brain? IEEE Spectrum, 54(6):34–71

  62. [70]

    Hebrail, G. E. R. and Barard, A. E. R. (2006). Individual household electric power consumption data set (ihepcds). https://archive.ics.uci.edu/ml/datasets/Individual+household+electric+ power+consumption

  63. [71]

    Hernandez, A., Nieto, R., Fuentes, D., and Urena, J. (2020). Design of a SoC Architecture for the Edge Computing of NILM Techniques. In 2020 XXXV Conference on Design of Circuits and Integrated Systems (DCIS), pages 1–6. IEEE

  64. [72]

    and Roy, R

    Herring, H. and Roy, R. (2007). Technological innovation, energy efficient design and the rebound effect. Technovation, 27(4):194–203

  65. [73]

    Hesselink, L. X. and Chappin, E. J. (2019). Adoption of energy efficient technologies by households – barriers, policies and agent -based modelling studies. Renewable and Sustainable Energy Reviews, 99:29–41

  66. [74]

    Himeur, Y., Alsalemi, A., Bensaali, F., and Amira, A. (2020). Building power consumption datasets: Survey, taxonomy and future directions. Energy and Buildings, 227:110404

  67. [75]

    and Schmidhuber, J

    Hochreiter, S. and Schmidhuber, J. (1997). Long short-term memory. Neural computation , 9(8):1735–1780

  68. [76]

    H., Oreszczyn, T., Ridley, I., Group, W

    Hong, S. H., Oreszczyn, T., Ridley, I., Group, W. F. S., et al. (2006). The impact of energy efficient refurbishment on the space heating fuel consumption in english dwellings. Energy and buildings, 38(10):1171–1181

  69. [77]

    Horyachyy, O. (2017). Comparison of wireless communication technologies used in a smart home: Analysis of wireless sensor node based on arduino in home automation scenario. In Master of Science in Electrical Engineering with emphasis on Telecommunication Systems Faculty of Com...

  70. [78]

    H., Tsolakis, A., Alagumalai, A., Mahian, O., Lam, S

    Hosseini, S. H., Tsolakis, A., Alagumalai, A., Mahian, O., Lam, S. S., Pan, J., Peng, W., Tabatabaei, M., and Aghbashlo, M. (2023). Use of hydrogen in dual-fuel diesel engines. Progress in Energy and Combustion Science, 98:101100

  71. [79]

    Huber, P., Calatroni, A., Rumsch, A., and Paice, A. (2021). Review on deep neural networks applied to low-frequency nilm. Energies, 14(9):2390

  72. [80]

    C., Park, J., and Shon, J

    Hwang, H. C., Park, J., and Shon, J. G. (2016). Design and implementation of a reliable message transmission system based on mqtt protocol in IoT. Wireless Personal Communications, 91(4):1765–1777

  73. [81]

    Electricity 2024 - executive summary

    International Energy Agency (IEA) (2024). Electricity 2024 - executive summary. Accessed: 2024-11-19

  74. [82]

    K., Malik, F

    Iqbal, H. K., Malik, F. H., Muhammad, A., Qureshi, M. A., Abbasi, M. N., and Chishti, A. R. (2021). A critical review of state-of-the-art non-intrusive load monitoring datasets. Electric Power Systems Research, 192:106921

  75. [83]

    James, G., Witten, D., Hastie, T., Tibshirani, R., and Taylor, J. (2023). Resampling methods. In An Introduction to Statistical Learning: with Applications in Python, pages 201–228. Springer. 120

  76. [84]

    B., and Yan, G

    Jiao, R., Li, C., Xun, G., Zhang, T., Gupta, B. B., and Yan, G. (2023). A context-aware multi- event identification method for non-intrusive load monitoring. IEEE Transactions on Consumer Electronics, pages 1–1

  77. [85]

    Jones, J. S. (2022). Europe’s smart electricity meter penetration reaches 56%. https://www. smart-energy.com/

  78. [86]

    Kane, T., Cockbill, S., May, A., Mitchell, V., Wilson, C., Dimitriou, V., Liao, J., Murray, D., Stankovic, V., Stankovic, L., et al. (2015). Supporting retrofit decisions using smart metering data: A multi-disciplinary approach. In Energy use in buildings. ECEEE

  79. [87]

    Kaselimi, M., Doulamis, N., Doulamis, A., Voulodimos, A., and Protopapadakis, E. (2019). Bayesian-optimized bidirectional lstm regression model for nilm. In ICASSP 2019 - 2019 IEEE (ICASSP), pages 2747–2751

  80. [88]

    Kaselimi, M., Doulamis, N., Voulodimos, A., Doulamis, A., and Protopapadakis, E. (2021a). Energan++: A generative adversarial gated recurrent network for robust energy disaggregation. IEEE Open Journal of Signal Processing, 2:1–16

  81. [89]

    Kaselimi, M., Doulamis, N., Voulodimos, A., Protopapadakis, E., and Doulamis, A. (2020a). Context aware energy disaggregation using adaptive bidirectional lstm models. IEEE Transactions on Smart Grid, 11(4):3054–3067

  82. [90]

    Kaselimi, M., Protopapadakis, E., Voulodimos, A., Doulamis, N., and Doulamis, A. (2022). Towards trustworthy energy disaggregation: A review of challenges, methods, and perspectives for non-intrusive load monitoring. Sensors, 22(15)

  83. [91]

    Kaselimi, M., Voulodimos, A., Doulamis, N., Doulamis, A., and Protopapadakis, E. (2021b). A robust to noise adversarial recurrent model for non-intrusive load monitoring. In ICASSP 2021 - 2021 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), p...

  84. [92]

    Kaselimi, M., Voulodimos, A., Protopapadakis, E., Doulamis, N., and Doulamis, A. (2020b). Energan: A generative adversarial network for energy disaggregation. In ICASSP 2020 - 2020 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) , pages 1578–1582

  85. [93]

    S., Wong, J., and Chua, K

    Kee, K.-K., Lim, Y. S., Wong, J., and Chua, K. H. (2019). Non-intrusive load monitoring (nilm) – a recent review with cloud computing. In 2019 IEEE International Conference on Smart Instrumentation, Measurement and Application (ICSIMA), pages 1–6

  86. [94]

    and Knottenbelt, W

    Kelly, J. and Knottenbelt, W. (2015a). The UK -DALE dataset, domestic appliance -level electricity demand and whole-house demand from five UK homes. Scientific Data, 2

  87. [95]

    and Knottenbelt, W

    Kelly, J. and Knottenbelt, W. (2015b). The uk-dale dataset, domestic appliance-level electricity demand and whole-house demand from five uk homes. Scientific data, 2(1):1–14

  88. [96]

    Khajenasiri, I., Estebsari, A., Verhelst, M., and Gielen, G. (2017). A review on internet of things solutions for intelligent energy control in buildings for smart city applications. Energy Procedia, 111:770–779

  89. [97]

    A., Sajjad, I

    Khan, M. A., Sajjad, I. A., Tahir, M., and Haseeb, A. (2022). Iot application for energy management in smart homes. Engineering Proceedings, 20(1):43. 121

  90. [98]

    and Cho, S.-B

    Kim, T.-Y. and Cho, S.-B. (2018). Predicting the household power consumption using cnn-lstm hybrid networks. In Yin, H., Camacho, D., Novais, P., and Tallón -Ballesteros, A. J., editors, Intelligent Data Engineering and Automated Learning – IDEAL 2018, pages 481 –490, Cham. Sp...

  91. [99]

    Kipf, T. N. and Welling, M. (2016). Semi-supervised classification with graph convolutional networks. arXiv preprint arXiv:1609.02907

  92. [100]

    Barriers to energy efficiency adoption in low - income communities

    Kleinman Center for Energy Policy (2024). Barriers to energy efficiency adoption in low - income communities. Accessed: 2024-11-19

  93. [101]

    Klemenjak, C., Reinhardt, A., Pereira, L., Makonin, S., Bergés, M., and Elmenreich, W. (2019). Electricity consumption data sets: Pitfalls and opportunities. In Proceedings of the 6th ACM international conference on systems for energy-efficient buildings, cities, and transport...

  94. [102]

    and Johnson, M

    Kolter, J. and Johnson, M. (2011). REDD: A Public Data Set for Energy Disaggregation Research. In IN SUSTKDD, volume 25

  95. [103]

    C., and Karybali, I

    Kotsilitis, S., Kalligeros, E., Marcoulaki, E. C., and Karybali, I. G. (2023). An efficient lightweight event detection algorithm for on-site non-intrusive load monitoring. IEEE Transactions on Instrumentation and Measurement, 72:1–13

  96. [104]

    and Movellan, J

    Kruschke, J. and Movellan, J. (1991). Benefits of gain: speeded learning and minimal hidden layers in back -propagation networks. IEEE Transactions on Systems, Man, and Cybernetics , 21(1):273–280

  97. [105]

    Kukunuri, R., Aglawe, A., Chauhan, J., Bhagtani, K., Patil, R., Walia, S., and Batra, N. (2020). Edgenilm: Towards nilm on edge devices. In Proceedings of the 7th ACM International Conference on Systems for Energy-Efficient Buildings, Cities, and Transportation, BuildSys ’20, ...

  98. [106]

    S., Kumudham, R., Kumar, D

    Kumar, P. S., Kumudham, R., Kumar, D. R., Dhamodharan, M., and Vetrivel, S. (2022). Smart home automation using raspberry pi 4. In AIP Conference Proceedings , volume 2461 of AIP Conference Proceedings, page 020012. AIP Publishing LLC

  99. [107]

    LeCun, Y., Bengio, Y., and Hinton, G. (2015). Deep learning. nature, 521(7553):436–444

  100. [108]

    Lee, Y.-L., Tsung, P.-K., and Wu, M. (2018). Techology trend of edge ai. In 2018 International Symposium on VLSI Design, Automation and Test (VLSI -DAT), pages 1 –2, Ambassador Hotel, Hsinchu, Taiwan. IEEE

  101. [109]

    Li, H., Kadav, A., Durdanovic, I., Samet, H., and Graf, H. P. (2016). Pruning filters for efficient convnets. arXiv preprint arXiv:1608.08710

  102. [110]

    and Yang, B

    Liebgott, F. and Yang, B. (2017). Active learning with cross-dataset validation in event-based non-intrusive load monitoring. In 2017 25th European Signal Processing Conference (EUSIPCO), pages 296–300

  103. [111]

    Linh An, P. m. and Kim, T. (2018). A study of the z-wave protocol: Implementing your own smart home gateway. In 2018 3rd International Conference on Computer and Communication Systems (ICCCS), pages 411–415. 122

  104. [112]

    R., Taksinwarajan, D., and Seneviratne, S

    Liou, J.-C., Jain, S., Singh, S. R., Taksinwarajan, D., and Seneviratne, S. (2020). Side-channel information leaks of z-wave smart home iot devices: Demo abstract. In Proceedings of the 18th Conference on Embedded Networked Sensor Systems, pages 637–638

  105. [113]

    Liu, Z., Li, J., Shen, Z., Huang, G., Yan, S., and Zhang, C. (2017). Learning efficient convolutional networks through network slimming. In ICCV 2017, pages 2736–2744

  106. [114]

    Lloyd, S. (1982). Least squares quantization in pcm. IEEE transactions on information theory, 28(2):129–137

  107. [115]

    Lopes, M., Antunes, C., and Martins, N. (2012). Energy behaviours as promoters of energy efficiency: A 21st century review. Renewable and Sustainable Energy Reviews, 16(6):4095–4104

  108. [116]

    Lu, C., Li, S., and Lu, Z. (2022). Building energy prediction using artificial neural networks: A literature survey. Energy and Buildings, 262:111718

  109. [117]

    Machlev, R., Malka, A., Perl, M., Levron, Y., and Belikov, J. (2022). Explaining the decisions of deep learning models for load disaggregation (nilm) based on xai. In 2022 IEEE Power Energy Society General Meeting (PESGM), pages 1–5

  110. [118]

    V., and Popowich, F

    Makonin, S., Ellert, B., Bajic´, I. V., and Popowich, F. (2016). Electricity, water, and natural gas consumption of a residential house in Canada from 2012 to 2014. Scientific data, 3(1):1–12

  111. [119]

    V., Gill, B., and Bartram, L

    Makonin, S., Popowich, F., Bajic´, I. V., Gill, B., and Bartram, L. (2015). Exploiting hmm sparsity to perform online real-time nilm. IEEE Transactions on smart grid, 7(6):2575–2585

  112. [120]

    Marcheggiani, D., Bastings, J., and Titov, I. (2018). Exploiting semantics in neural machine translation with graph convolutional networks. arXiv preprint arXiv:1804.08313

  113. [121]

    Electricity demand in europe: Growing or going? Accessed: 2024-11-19

    McKinsey Company (2024). Electricity demand in europe: Growing or going? Accessed: 2024-11-19

  114. [122]

    Monacchi, A., Egarter, D., Elmenreich, W., D’Alessandro, S., and Tonello, A. M. (2014). Greend: An energy consumption dataset of households in Italy and Austria. In 2014 IEEE International Conference on Smart Grid Communications (SmartGridComm) , pages 511 –516. IEEE

  115. [123]

    Monarch, R. M. (2021). Human-in-the-Loop Machine Learning: Active learning and annota- tion for human-centered AI. Simon and Schuster

  116. [124]

    Murray, D., Liao, J., Stankovic, L., and Stankovic, V. (2016). Understanding usage patterns of electric kettle and energy saving potential. Applied Energy, 171:231–242

  117. [125]

    Murray, D., Stankovic, L., and Stankovic, V. (2017a). An electrical load measurements dataset of United Kingdom households from a two-year longitudinal study. Scientific data, 4(1):1–12

  118. [126]

    Murray, D., Stankovic, L., and Stankovic, V. (2017b). An electrical load measurements dataset of united kingdom households from a two-year longitudinal study. Scientific Data, 4:160122

  119. [127]

    Murray, D., Stankovic, L., and Stankovic, V. (2020). Explainable nilm networks. In Proceedings of the 5th International Workshop on Non-Intrusive Load Monitoring, NILM’20, page 64–69, New York, NY, USA. Association for Computing Machinery. 123

  120. [128]

    Murray, D., Stankovic, L., Stankovic, V., and Espinoza-Orias, N. (2018). Appliance electrical consumption modelling at scale using smart meter data. Journal of Cleaner Production, 187:237– 249

  121. [129]

    Murray, D., Stankovic, L., Stankovic, V., Lulic, S., and Sladojevic, S. (2019). Transferability of neural network approaches for low -rate energy disaggregation. In ICASSP 2019-2019 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), pages 8330–8...

  122. [130]

    Nebey, A. H. (2024). Recent advancement in demand side energy management system for optimal energy utilization. Energy Reports, 11:5422–5435

  123. [131]

    A., Colantuono, G., and Georges, J.-P

    Ortiz, P., Kubler, S., Éric Rondeau, McConky, K., Shukhobodskiy, A. A., Colantuono, G., and Georges, J.-P. (2022). Greenhouse gas emission reduction in residential buildings: A lightweight model to be deployed on edge devices. Journal of Cleaner Production, 368:133092

  124. [132]

    and Socolow, R

    Pacala, S. and Socolow, R. (2004). Stabilization wedges: solving the climate problem for the next 50 years with current technologies. science, 305(5686):968–972

  125. [133]

    G., Christoforidis, G

    Papageorgiou, P. G., Christoforidis, G. C., and Bouhouras, A. S. (2025). Nilm in high frequency domain: A critical review on recent trends and practical challenges. Renewable and Sustainable Energy Reviews, 213:115497

  126. [134]

    Pau, G., Collotta, M., Ruano, A., and Qin, J. (2017). Smart home energy management. Energies, 10(3)

  127. [135]

    Pereira, L., Costa, D., and Ribeiro, M. (2022). A residential labeled dataset for smart meter data analytics. Scientific Data, 9(1):134

  128. [136]

    and Nunes, N

    Pereira, L. and Nunes, N. (2018). Performance evaluation in non -intrusive load monitor - ing: Datasets, metrics, and tools —a review. Wiley Interdisciplinary Reviews: data mining and knowledge discovery, 8(6):e1265

  129. [137]

    Porteiro, R., Hernandez-Callejo, L., and Nesmachnow, S. (2022). Electricity demand forecast- ing in industrial and residential facilities using ensemble machine learning. Revista Facultad de Ingeniera Universidad de Antioquia, 350(102):9–25

  130. [138]

    Pullinger, M., Kilgour, J., Goddard, N., Berliner, N., Webb, L., Dzikovska, M., Lovell, H., Mann, J., Sutton, C., Webb, J., et al. (2021). The ideal household energy dataset, electricity, gas, contextual sensor data and survey data for 255 UK homes. Scientific Data, 8(1):146

  131. [139]

    Rafiq, H., Zhang, H., Li, H., and Ochani, M. K. (2018). Regularized lstm based deep learning model: First step towards real -time non -intrusive load monitoring. 2018 IEEE International Conference on Smart Energy Grid Engineering (SEGE), pages 234–239

  132. [140]

    Ramadan, R., Huang, Q., Bamisile, O., and Zalhaf, A. S. (2022). Intelligent home energy management using internet of things platform based on nilm technique. Sustainable Energy, Grids and Networks, 31:100785

  133. [141]

    Rashid, A. B. and Kausik, M. A. K. (2024). Ai revolutionizing industries worldwide: A comprehensive overview of its diverse applications. Hybrid Advances, 7:100277

  134. [142]

    Rashid, H., Singh, P., Stankovic, V., and Stankovic, L. (2019). Can non -intrusive load monitoring be used for identifying an appliance’s anomalous behaviour? Applied energy, 238:796– 805. 124

  135. [143]

    Raspberry pi 4

    Raspberry (2023). Raspberry pi 4. https://www.raspberrypi.com/products/ raspberry-pi-4-model-b/

  136. [144]

    Rathnayaka, A. J. D., Potdar, V. M., and Kuruppu, S. J. (2011). Evaluation of wireless home automation technologies. In 5th IEEE International Conference on Digital Ecosystems and Technologies (IEEE DEST 2011), pages 76–81

  137. [145]

    M., Raza, M

    Rind, Y. M., Raza, M. H., Zubair, M., Mehmood, M. Q., and Massoud, Y. (2023). Smart energy meters for smart grids, an internet of things perspective. Energies, 16(4)

  138. [146]

    and Casella, G

    Robert, C. and Casella, G. (2011). A short history of markov chain monte carlo: Subjective recollections from incomplete data. Statistical Science, 26(1):102–115

  139. [147]

    and Caird, S

    Roy, R. and Caird, S. (2006). Designing low and zero carbon products and systems–adoption, effective use and innovation. The Open University

  140. [148]

    Ruano, A., Hernandez, A., Ureña, J., Ruano, M., and Garcia, J. (2019). Nilm techniques for intelligent home energy management and ambient assisted living: A review. Energies, 12(11)

  141. [149]

    and Jiao, Z

    Shi, R. and Jiao, Z. (2023). Individual household demand response potential evaluation and identification based on machine learning algorithms. Energy, 266:126505

  142. [150]

    and Surantha, N

    Simadiputra, V. and Surantha, N. (2021). Rasefiberry: Secure and efficient raspberry-pi based gateway for smarthome IoT architecture. Bulletin of Electrical Engineering and Informatics , 10(2):1035–1045

  143. [151]

    M., and Harijyothi, M

    Sivapriyan, R., Rao, K. M., and Harijyothi, M. (2020). Literature review of iot based home automation system. In 2020 Fourth International Conference on Inventive Systems and Control (ICISC), pages 101–105

  144. [152]

    and Makwana, A

    Soni, D. and Makwana, A. (2017). A survey on mqtt: a protocol of internet of things (iot). In International conference on telecommunication, power analysis and computing techniques (ICTPACT-2017), volume 20, pages 173–177

  145. [153]

    Srinivas, S., Subramanya, A., and Babu, R. V. (2017). Training sparse neural networks. In 2017 IEEE Conference on Computer Vision and Pattern Recognition Workshops (CVPRW), pages 455–462

  146. [154]

    G., and Kadhe, S

    Srinivasarengan, K., Goutam, Y., Chandra, M. G., and Kadhe, S. (2013a). A framework for nilm using bayesian inference. In 2013 Conference on Innovative Mobile and Internet Services , pages 427–432. IEEE

  147. [155]

    G., and Kadhe, S

    Srinivasarengan, K., Goutam, Y., Chandra, M. G., and Kadhe, S. (2013b). A framework for non intrusive load monitoring using bayesian inference. In 2013 Seventh International Conference on Innovative Mobile and Internet Services in Ubiquitous Computing, pages 427–432. IEEE

  148. [156]

    Stankovic, L., Stankovic, V., Liao, J., and Wilson, C. (2016). Measuring the energy intensity of domestic activities from smart meter data. Applied Energy, 183:1565–1580

  149. [157]

    Global electricity consumption from 1980 to 2023

    Statista (2024). Global electricity consumption from 1980 to 2023. Accessed: 2024-11-19

  150. [158]

    Sun, Y., Zheng, L., Wang, Q., Ye, X., Huang, Y., Yao, P., Liao, X., and Jin, H. (2022). Accelerating sparse deep neural network inference using gpu tensor cores. In 2022 IEEE High Performance Extreme Computing Conference (HPEC), pages 1–7. 125

  151. [159]

    Sutskever, I., Vinyals, O., and Le, Q. V. (2014). Sequence to sequence learning with neural networks. Advances in neural information processing systems, 27

  152. [160]

    Sykiotis, S., Athanasoulias, S., Kaselimi, M., Doulamis, A., Doulamis, N., Stankovic, L., and Stankovic, V. (2023). Performance -aware nilm model optimization for edge deployment. IEEE Transactions on Green Communications and Networking, pages 1–1

  153. [161]

    Sykiotis, S., Kaselimi, M., Doulamis, A., and Doulamis, N. (2022). Electricity: An efficient transformer for nilm. Sensors, 22(8)

  154. [162]

    Tabanelli, E., Brunelli, D., Acquaviva, A., and Benini, L. (2022). Trimming Feature Extraction and Inference for MCU -based Edge NILM: A Systematic Approach. IEEE Transactions on Industrial Informatics, 18(2):943–952

  155. [163]

    Tabanelli, E., Brunelli, D., and Benini, L. (2020). A feature reduction strategy for enabling lightweight non -intrusive load monitoring on edge devices. In 2020 IEEE 29th International Symposium on Industrial Electronics (ISIE), pages 805–810. IEEE

  156. [164]

    Tiwari, G., Sharma, A., Sahotra, A., and Kapoor, R. (2020). English-hindi neural machine translation-lstm seq2seq and convs2s. In 2020 International Conference on Communication and Signal Processing (ICCSP), pages 871–875. IEEE

  157. [165]

    Todic, T., Stankovic, V., and Stankovic, L. (2023). An active learning framework for the low-frequency non-intrusive load monitoring problem. Applied Energy, 341:121078

  158. [166]

    Trotta, G., Hansen, A., Aagaard, L., and Gram -Hanssen, K. (2023). SURVEY QUESTION- NAIRE ON HOUSEHOLDS’ USE OF SMART HOME TECHNOLOGY AND THEIR TIME OF USE OF ELECTRIC APPLIANCES. Institut for Byggeri, By og Miljø (BUILD), Aalborg Universitet

  159. [167]

    DECC, UK government

    UK Government (2023). DECC, UK government. smart metering equipment technical specifications: Version 2. https://www.gov.uk/government/publications/ smart-metering-implementation-programme-technical-specifications. Accessed: 2023- 05-22

  160. [168]

    Ullah, M., Javaid, N., Khan, I., Mahmood, A., and Farooq, M. (2013). Residential energy consumption controlling techniques to enable autonomous demand side management in future smart grid communications. In 2013 Eighth International Conference on Broadband and Wireless Computi...

  161. [169]

    Energy Information Administration (EIA) (2013)

    U.S. Energy Information Administration (EIA) (2013). International energy outlook 2013. Accessed: 2024-11-19

  162. [170]

    S., Reyes Lua, A., and Prasad, V

    Uttama Nambi, A. S., Reyes Lua, A., and Prasad, V. R. (2015). Loced: Location-aware energy disaggregation framework. In Proceedings of the 2nd acm international conference on embedded systems for energy-efficient built environments, pages 45–54

  163. [171]

    and Ameen, S

    Vadera, S. and Ameen, S. (2022). Methods for pruning deep neural networks. IEEE Access, 10:63280–63300

  164. [172]

    N., Kaiser, Ł., and Polosukhin, I

    Vaswani, A., Shazeer, N., Parmar, N., Uszkoreit, J., Jones, L., Gomez, A. N., Kaiser, Ł., and Polosukhin, I. (2017). Attention is all you need. Advances in neural information processing systems, 30

  165. [173]

    Vázquez-Canteli, J. R. and Nagy, Z. (2019). Reinforcement learning for demand response: A review of algorithms and modeling techniques. Applied energy, 235:1072–1089. 126

  166. [174]

    Vitiello, S., Andreadou, N., Ardelean, M., and Fulli, G. (2022). Smart metering roll-out in Europe: Where do we stand? cost benefit analyses in the clean energy package and research trends in the green deal. Energies, 15(7)

  167. [175]

    Völker, B., Reinhardt, A., Faustine, A., and Pereira, L. (2021). Watt’s up at home? smart meter data analytics from a consumer-centric perspective. Energies, 14(3)

  168. [176]

    Wang, L., Mao, S., and Nelms, R. M. (2022). Transformer for nilm: Complexity reduction and transferability. IEEE Internet of Things Journal, 9(19):18987–18997

  169. [177]

    Wang, P., Ye, F., and Chen, X. (2018). A smart home gateway platform for data collection and awareness. IEEE Communications Magazine, 56(9):87–93

  170. [178]

    Wang, Y., Chen, Q., Hong, T., and Kang, C. (2019). Review of smart meter data analytics: Applications, methodologies, and challenges. IEEE Transactions on Smart Grid, 10(3):3125–3148

  171. [179]

    Wang, Z. (2020). SparseRT: Accelerating Unstructured Sparsity on GPUs for Deep Learning Inference. PACT ’20, page 31–42, New York, NY, USA. ACM

  172. [180]

    Wen, Z., O’Neill, D., and Maei, H. (2015). Optimal demand response using device -based reinforcement learning. IEEE Transactions on Smart Grid, 6(5):2312–2324

  173. [181]

    D., Dumontier, M., Aalbersberg, I

    Wilkinson, M. D., Dumontier, M., Aalbersberg, I. J., Appleton, G., Axton, M., Baak, A., Blomberg, N., Boiten, J.-W., da Silva Santos, L. B., Bourne, P. E., et al. (2016). The fair guiding principles for scientific data management and stewardship. Scientific Data, 3(1):160018

  174. [182]

    Wu, Z., Wang, C., Peng, W., Liu, W., and Zhang, H. (2021). Non-intrusive load monitoring using factorial hidden markov model based on adaptive density peak clustering. Energy and Buildings, 244:111025

  175. [183]

    C., Yang, T., and Zomaya, A

    Xia, C., Li, W., Chang, X., Delicato, F. C., Yang, T., and Zomaya, A. Y. (2018). Edge-based energy management for smart homes. In IEEE, pages 849–856. IEEE

  176. [184]

    S., et al

    Xu, M., Li, L., Wong, D., Liu, Q., Chao, L. S., et al. (2020). Document graph for neural machine translation. arXiv preprint arXiv:2012.03477

  177. [185]

    Xu, Q., Liu, Y., and Luan, K. (2022). Edge-Based NILM System with MDMR Filter -Based Feature Selection. In 2022 IEEE 5th International Electrical and Energy Conference (CIEEC) , pages 5015–5020. IEEE

  178. [186]

    B., Mardini, W., and Khalil, A

    Yassein, M. B., Mardini, W., and Khalil, A. (2016). Smart homes automation using z -wave protocol. In 2016 International Conference on Engineering MIS (ICEMIS), pages 1–6

  179. [187]

    Yu, H., Jiang, Z., Li, Y., Zhou, J., Wang, K., Cheng, Z., and Gu, Q. (2019). A multi-objective non-intrusive load monitoring method based on deep learning. In IOP Conference Series: Materials Science and Engineering, volume 486, page 012110. IOP Publishing

  180. [188]

    Yuan, J., Jin, R., Wang, L., and Wang, T. (2024). A non-intrusive load identification method based on dual-branch attention gru fusion network. IEEE Transactions on Instrumentation and Measurement

  181. [189]

    R., Jorde, D., and Jacobsen, H

    Yue, Z., Witzig, C. R., Jorde, D., and Jacobsen, H. -A. (2020). Bert4nilm: A bidirectional transformer model for non -intrusive load monitoring. In Proceedings of the 5th International Workshop on Non -Intrusive Load Monitoring , NILM’20, page 89 –93, New York, NY, USA. Associ...

  182. [190]

    Zboˇril, J., Hujnˇák, O., and Malinka, K. (2023). Iot gateways network communication analysis. In 2023 International Conference on Information Networking (ICOIN), pages 334–339. IEEE

  183. [191]

    Zhang, C., Zhong, M., Wang, Z., Goddard, N., and Sutton, C. (2018). Sequence-to-point learn- ing with neural networks for non-intrusive load monitoring. In Proceedings of the Thirty-Second AAAI Conference on Artificial Intelligence and Thirtieth Innovative Applications of Arti...

  184. [192]

    Zhang, Y., Tang, G., Huang, Q., Wang, Y., Wu, K., Yu, K., and Shao, X. (2023). Fednilm: Applying federated learning to nilm applications at the edge. IEEE Transactions on Green Communications and Networking, 7(2):857–868

  185. [193]

    Zhang, Y., Yang, G., and Ma, S. (2019). Non-intrusive load monitoring based on convolutional neural network with differential input. Procedia CIRP, 83:670–674. 11th CIRP

  186. [194]

    Zhao, B., He, K., Stankovic, L., and Stankovic, V. (2018). Improving event-based non-intrusive load monitoring using graph signal processing. IEEE Access, 6:53944–53959

  187. [195]

    Zhou, J., Cui, G., Hu, S., Zhang, Z., Yang, C., Liu, Z., Wang, L., Li, C., and Sun, M. (2020). Graph neural networks: A review of methods and applications. AI Open, 1:57–81

  188. [196]

    and Zhang, L

    Zhou, S. and Zhang, L. (2018). Smart home electricity demand forecasting system based on edge computing. In 2018 IEEE 9th International Conference on Software Engineering and Service Science (ICSESS), pages 164–167

  189. [197]

    Zhuang, Y.-t., Wu, F., Chen, C., and Pan, Y. -h. (2017). Challenges and opportunities: from big data to knowledge in ai 2.0. Frontiers of Information Technology & Electronic Engineering, 18:3–14

  190. [198]

    A., Suryanarayanan, S., Roche, R., Earle, L., Christensen, D., Bauleo, P., and Zimmerle, D

    Zipperer, A., Aloise-Young, P. A., Suryanarayanan, S., Roche, R., Earle, L., Christensen, D., Bauleo, P., and Zimmerle, D. (2013). Electric energy management in the smart home: Perspectives on enabling technologies and consumer behavior. Proceedings of the IEEE, 101(11):2397–2408

  191. [199]

    A., and Rajasegarar, S

    Zoha, A., Gluhak, A., Imran, M. A., and Rajasegarar, S. (2012). Non-intrusive load monitoring approaches for disaggregated energy sensing: A survey. Sensors, 12(12):16838–16866

  192. [200]

    Z-wave daughter card

    ZWave (2023). Z-wave daughter card. https://z-wave.me/products/razberry/slide-2

  193. [201]

    C., and Guerrero, J

    Çimen, H., Çetinkaya, N., Vasquez, J. C., and Guerrero, J. M. (2021). A microgrid energy man- agement system based on non-intrusive load monitoring via multitask learning. IEEE Transactions on Smart Grid, 12(2):977–987

  194. [202]

    Žitnik, S., Jankovic´, M., Petrovcˇicˇ, K., and Bajec, M. (2016). Architecture of standard-based, interoperable and extensible iot platform. In 2016 24th Telecommunications Forum (TELFOR) , pages 1–4

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.