Table 1 Studies on the application of...

Table 1

Studies on the application of ANN in predicting properties of stabilized soil

Reference	Features	Target	Stabilizing agent/material	No. of data used	Train:test:validation data(%)	Network/training algorithm	No. of input layer neurons	Optimum No. of neurons in hidden layer	No. of output layer neurons	Feature importance/sensitivity analysis		Model performance (testing)
Reference	#	@	$				No. of input layer neurons	Optimum No. of neurons in hidden layer	No. of output layer neurons	Method *	Result	Model performance (testing)
Heshmati et al. (2009)	LL, PL, PI, LS, Gr, S, C, L, CC, AC	UCS	CC, L, AC	219	50:25:25	FFN/RBF	8	32	1	Parametric study	UCS insensitive to LC and CC	R = 0.9667,MSE = 0.071,MAE = 0.188
Alavi et al. (2010)	LL, PI, LS,C, S, L, CC, AC	MDD and OMC	L, CC, AC	192	52:24:24	MLP/GaA	9	12 (MDD), 10 (OMC)	1	Parametric study	G > LL>PI (MDD) G > PI>LS(OMC)	R² = 0.8916 MSE = 0.895 MAE = 1.13
Das et al. (2011)	LL, PI, C, S, Gr, WC, CC	MDD and OMC	CC	55	67:33:00	BRNN, LMNN, DENN	7	4	1	GA and CWA	CC > PI>C > LL>S > Gr>WC	MAE = 1.82 AAE = 0.61 RMSE = 0.80
Ouf (2012)	RC, OPC, GGBS, L, CA	UCS and FSP	GGBS, L,RC,OPC	38		GeA	5		2			Min error value < 4% for UCS and <8% for FSP
Mozumder and Laskar (2015)	LL, PI, GGBS, FA, MoC, a/B, Na/Al, Si/Al	UCS(28 days)	Geopolymers (GGBS,FA)	142	70:30:00	MLP/BRNN-BP	8	9	1	GA and CWA	%S>%FA > Na/Al > Si/Al > A/B > MoC>PI > LL	R = 0.982 MSE = 1.50 MAE = 8.34
Sabat (2015)	L, QD, OMC, MDD, CA	CBR (28 days soaked)	L, QD	90	76:24:00	DENN,LMNN,BRNN	5	5	1	GA	MDD > OMC>L > CA>QD	R2 = 0.981 RMSE = 1.187 MAE = 1.75 AAE = 1.07
Vakili et al. (2015)	P_0.005, PI, MDD, L, FC, PoP, CA	Coefficient of permeability (k)	L, pozzolan	69	70:15:15	MLP/BP	6	9	1			R = 0.99
Ayeldeen et al. (2016)	CA, CoE	Gain in strength(%)	CC	80	70:15:15	FFN/BPA	2	2 layer,8 in each	1			R² = 0.908
Belal et al. (2016)	RHA, L, CA, OMC, MDD	CBR (28 days soaked)	L, RHA	48	70:15:15	FFN/BPA	5	12	1			R = 0.9889 MAE = 0.1644
Vinodhkumar and Balaji (2016)	LL, PI, FA, OMC, MDD, NoGl	Soaked CBR	FA, geotextile			LMNN	6	2–30	1			R = 0.9869 MSE = 8.024 × 10-11
Ghanizadeh and Rahrovan (2016)	WD cycle, L/SAF, MDD/OMC, σ3, σd	Mr	CKD, FA, FBA	704	60:30:10	FFN/BPA	5	24	1			R²=0.9857 MSE = 49784(Overall)
Bahmed et al., 2017	LL, PL, LC	PI, MDD and OMC	L	280(PI), 122 (MDD and OMC)	70:15:15	LMNN	3	7(PI), 11(MDD), 9(OMC)	1			R = 0.9373(PI) R = 0.9356(MDD) R = 0.9406(OMC)
Salahudeen et al. (2018)	G, LS, FSP, D10, D30, D60, Cu, Cc, LL, PL	MDD and OMC	CKD	90	70:15:15	MLP/BPA	10	1–10	2			OMC,MDD R = 0.983,9884 MSE = 0.0013,0.001 MAE = 0.0208,0.0321
Taha et al. (2018)	C, M, S, G, PI, LL, NM	MDD and OMC	Nano-Cu,Nano-Al,Nano-clay	75	80:20:00	FFN/BPA	7	1–20	1			Nano clay, Nano Al, Nano-copper(R2=0.987,0.991,0.989)
Salahudeen and Sadeeq, (2019)	PL, LL, G, LS, Cu, Cc, OMC and MDD	CBR (soaked and unsoaked)	CKD	72	70:15:15	FFN	8	8(Soaked),17(unsoaked)	1			Soaked,Unsoaked R = 0.9986,991 MSE = 0.00013,0.00109 MAE = 0.008,0.012
Tinoco et al. (2019)	WC, C, M, S, W/C, CC, OM, CA, CoB, CoSB	UCS	CC	444	70:30:00	DMA	10		1	GSA	W/C > CC>OM > CA	R2=0.94 ± 0.001 RMSE = 0.69 ± 0.05 MAE = 0.46 ± 0.02
Onyelowe et al. (2023)	CC, L, LL, PI, OMC, MDD	UCS	CC,L	190	74:26		9	5	1		MDD > LL,PL > CC>L > OMC
Hanandeha et al. (2020)	CC, L, PI, M, C, FA, OMC, WC	Resilient modulus (Mr)	L and FA	125		BPA	8	9	1	Parmeteric study by changing input parameter	C > L>FA > PI>C > WC>OMC > M	R²=0.97(training)
Priyadarshee et al. (2020)	C, RHA, CC, PA, CA	UCS	PA,RHA,CC	129	70:15:15	FFN/LMNN-BP	5	2–11	1	CWA	RHA > PA>C, CC > CA	R = 0.9856 R²=0.9714 MSE = 51.34 RMSE = 7.1651
Rajakumar and Babu (2020)	AsT, AsC, LL, PL, MDD, OMC and NoGeL	Soaked CBR	CoA, BA, GSA and geogrid	210		CBLF,GDBLF,LMNN,HWLR	7	7	1			R²=0.97
Salahudeen et al. (2020)	G, LS, Cu, Cc, LL, PI, OMC, MDD	UCS(7, 14, 28 days)	CKD	72	70:15:15	MLP/BPA	8	1–15	1			Soaked,Unsoaked R = 0.99976,9806 MSE = 0.000079,0.000079
Shah et al. (2020)	AE, AS, PI, G, OMC, MDD, AASTO, UCS, GI	CBR (10, 30, 65 blows compaction)	Alum sludge waste	36	78:22		9	1 layer	3			10,30,65 blows R²=0.9892,0.9963,0.9749 RMSE = 0.1586,0.1244,0.5921
Hu and Solanki (2021)	25 properties of cementitiously stabilized subgrade soils.	Mr	CC			RBF,MLP/LM,BRNN,SCG	(Best)25	1 and 2 layers(MLP)/(Best)15 neurons	1			MSE = 0.5644 (25 - 15-1 layer)
Salahudeen et al. (2020)	G, LS, Cu, Cc, LL, PI, OMC, MDD	Durability, Mr, resistance value	CKD	72	70:15:15	FFN/BPA	8	17(durability),24(resilient modulus),18(resistance value)	1			Durability,M_r, resistance R = 0.8388,0.8433,0.7572 MSE = 0.01258,0.01446,0.02368 RMSE = 0.112,0.12,0.154
Ngo et al. (2021)	ST, WC, We, CC, CeT D, Sl, Sd, Sa, Sv, ma, De, CA, CuC	UCS	CC	216	80:20:00	FFN/BPA	14	2–50	1	GB model	We > CC>Ma > WC>De > D>CuC > Sa>Sv > S>CeT > CA>Sd > Sl	UCS 7,14,28 days R = 0.9812,0.9783,0.9942
Tabarsa et al. (2021)	ST, CA, DUW, CC, RHA, L	UCS	CC, L, RHA	137	60:20:20	MLP/LMNN-BPA	6	7 and 4	1	Statistical software	CC,L > other	R = 0.9957,AAPE = 12.349,AIC = 4.289
Tran (2021)	CC, WC, AF, WFNC	UCS	CC, AF, WFN	51	70:30:00	LMNN	4	2 layers, 12 and 10 in each	1	Relative importance	CC > AF>WC > WFNC	R = 0.94 MAE = 8.6535 RMSE = 10.3390
Pham et al. (2021)	F0.5, F0.25, F0.1, SiO₂,Fe₂O₃,Al₂O₃,SO₃,K₂O,CaO,Ti₂OCC, CA	UCS	CC	80		FFN/LMNN-BPA	12		1	GA and CWA	CC > F0.5	R²=0.997 MSE = 0.0415 RMSE = 0.0614
Chen et al. (2022)	WC, T, σ_N, ωt	Shear stress	AGF	40	80:20	BPNN	4	3 layer, (best-20,20,20)	1	Based on BPNN	ωt>σ_N>T > W	R²=0.948 RMSE = 16.153 MAPE = 24.230
Mustafa et al. (2022)	G, S, M + C, LL, PI, LS,% stabilizer, stabilizer type, OMC, MDD, AR, DW	UCS	Unstabilized, L,CC	488	70:15:15	LMNN	12	1 layer (24,41,44 for Three dataset)	1	Modifying soil chracteristics		R²=0.9883(overall)
Abdallah et al. (2023)	CC, MDD, IWC, CA	UCS, WC and suction	CC		80:20	BRBP	4	2 layers, 10 and 5 in each	3			UCS, WC, suction R²=0.972,0.932,0.995
Aljanabi and Salih (2023)	% of soil, LL, PL, SL, additives(%); additive type	UCS	UIR, UAO, ShP, LGC, WPP, RH, SSP	74	60:20:20, 70:15:15, 80:10:10	MLP	6	3 layers, 12; (1–7);1	1	Normalized importance	Additive type > additives(%); >PL > LL> SL>% of soil	R = 0.98
Krishna et al. (2023)	LL, PL, OMC, MDD, L, CC, FA (for UCS) & UCS (for CBR)	CBR and UCS	L, CC, FA	125	80:10:10	FFN/LMNN-BPA	7(UCS)&8(CBR)	2 layer, 10 in each	1			CBR,UCS R²=0.924,0.95 MAE = 0.45228,0.0166 RMSE = 0.00537,0.0012
Kumar et al. (2023)	Soil(%), CC, L, LL, PL, PI, MDD, OMC	UCS	L,CC	100	60:10:30	LMNN-BP	8	2 layer, 16 and 32 neurons each(best)	1			R = 0.70
Lu et al., 2023	S, M, C, LI, Wc/Cc	UCS	CC	80	80:20	LM, BR	5	1,2,3,4,5 layers and 60 neurons	1	Varying value of input parameter	Wc/Cc > M > S > C>LI	R²=0.896 MSE = 13667.232 RMSE = 116.907 MAE = 102.584 MSLE = 0.042 RMSLE = 0.206
Onyelowe et al. (2024)	WGP, NACL, PA, LL, PL, FSI, OMC, MDD	CBR,UCS	WGP,NACL,PA	25	80:20	BPNN	8	1 layer (1,2,3 neurons)	2	Relative importance	OMC > WGP>PA > NACL>FSI > LL>PL > MDD	CBR,UCS SSE = 1.5%,2% R²=0.9979,0.9973
Anh et al. (2024)	WC, LL, PL, Wab	Aps	PSAS	147	60:20:20	MLP/BP ANN and GA ANN	4	2–12-BP ANN, 8-GA ANN	1			BP, GA R²=0.9269,0.9121 MSE = 0.0299,0.0244
Baldovino et al. (2024)	CA, MDD, WGP(%), CC, VWC, VCC, WGP + CC, P/CCi,P/Bi	UCS(7 and 28 days), Stiffness	WGP, CC	72	70:15:15	BP		1 layer,9	1			Stiffness, UCS R2=1,1 RMSE = 0.0384,0.0021
Goutham and Krishnaiah (2024)	BA, L, LL, PL, SL, MDD, OMC	UCS(28 days)	BA, L	79	70:30	MLP/LMNN-BPA	7	7	1	GA and CWA	LL > L>SL > BA>PL > MDD > OMC (GaA), BA > OMC>MDD > SL>LL > L>PL (CWA)	R²=0.99
Mojtahedi et al. (2024)	CA, W/C, CC, PI, G, S, C	UCS by deep mixing	CC	192	75:25	FFBP	7	1 layer, 7 &12	1	Analyzing current weights	G > W/CC > C > S > CA>CC > PI	R2=0.992
Thapa et al. (2024)	C,M, NS, BC, SI, MDD, WC, omc	CBR	NS, BC	175		LMNN-BPA	6	1layer, 10 neurons	1	XAI (SHAP and LIME)	C > M>MC > NS/BC > SI>MDD	R²=0.958 MSE = 0.02 RMSE = 0.049 MAE = 0.085
Wani and Thagunna (2024)	GSD, PI, LL, PL, WC, FA, CA	Shear strength	FA		85:15		7	5 layer	1		FA > CA>other	R²=0.69 MSE = 0.01
Mohammed et al. (2025)	CA, MDD, OMC, GGBS, PL, LL, PI	UCS	GGBS	200	80:20		7		1	SHAP and LIME	CA > OMC>MDD > GGBS>PL > LL>PI (SHAP)	R²=0.94 RMSE = 0.15 MAE = 0.037
Sharma et al. (2024)	L, CA, PI, pH, Vp, OMC, MDD	UCS, E, c, ϕ	L	54	80:20		7	10	4			UCS,E,c,ϕ R²=0.887,0.926,0.805,0.859 RMSE = 20.984,3.654,6.019,2.338 MAPE-7.722,13.338,10.325,6.803

Reference	Features	Target	Stabilizing agent/material	No. of data used	Train:test:validation data(%)	Network/training algorithm	No. of input layer neurons	Optimum No. of neurons in hidden layer	No. of output layer neurons	Feature importance/sensitivity analysis		Model performance (testing)
Reference	#	@	$				No. of input layer neurons	Optimum No. of neurons in hidden layer	No. of output layer neurons	Method *	Result	Model performance (testing)
Heshmati et al. (2009)	LL, PL, PI, LS, Gr, S, C, L, CC, AC	UCS	CC, L, AC	219	50:25:25	FFN/RBF	8	32	1	Parametric study	UCS insensitive to LC and CC	R = 0.9667,MSE = 0.071,MAE = 0.188
Alavi et al. (2010)	LL, PI, LS,C, S, L, CC, AC	MDD and OMC	L, CC, AC	192	52:24:24	MLP/GaA	9	12 (MDD), 10 (OMC)	1	Parametric study	G > LL>PI (MDD) G > PI>LS(OMC)	R² = 0.8916 MSE = 0.895 MAE = 1.13
Das et al. (2011)	LL, PI, C, S, Gr, WC, CC	MDD and OMC	CC	55	67:33:00	BRNN, LMNN, DENN	7	4	1	GA and CWA	CC > PI>C > LL>S > Gr>WC	MAE = 1.82 AAE = 0.61 RMSE = 0.80
Ouf (2012)	RC, OPC, GGBS, L, CA	UCS and FSP	GGBS, L,RC,OPC	38		GeA	5		2			Min error value < 4% for UCS and <8% for FSP
Mozumder and Laskar (2015)	LL, PI, GGBS, FA, MoC, a/B, Na/Al, Si/Al	UCS(28 days)	Geopolymers (GGBS,FA)	142	70:30:00	MLP/BRNN-BP	8	9	1	GA and CWA	%S>%FA > Na/Al > Si/Al > A/B > MoC>PI > LL	R = 0.982 MSE = 1.50 MAE = 8.34
Sabat (2015)	L, QD, OMC, MDD, CA	CBR (28 days soaked)	L, QD	90	76:24:00	DENN,LMNN,BRNN	5	5	1	GA	MDD > OMC>L > CA>QD	R2 = 0.981 RMSE = 1.187 MAE = 1.75 AAE = 1.07
Vakili et al. (2015)	P_0.005, PI, MDD, L, FC, PoP, CA	Coefficient of permeability (k)	L, pozzolan	69	70:15:15	MLP/BP	6	9	1			R = 0.99
Ayeldeen et al. (2016)	CA, CoE	Gain in strength(%)	CC	80	70:15:15	FFN/BPA	2	2 layer,8 in each	1			R² = 0.908
Belal et al. (2016)	RHA, L, CA, OMC, MDD	CBR (28 days soaked)	L, RHA	48	70:15:15	FFN/BPA	5	12	1			R = 0.9889 MAE = 0.1644
Vinodhkumar and Balaji (2016)	LL, PI, FA, OMC, MDD, NoGl	Soaked CBR	FA, geotextile			LMNN	6	2–30	1			R = 0.9869 MSE = 8.024 × 10-11
Ghanizadeh and Rahrovan (2016)	WD cycle, L/SAF, MDD/OMC, σ3, σd	Mr	CKD, FA, FBA	704	60:30:10	FFN/BPA	5	24	1			R²=0.9857 MSE = 49784(Overall)
Bahmed et al., 2017	LL, PL, LC	PI, MDD and OMC	L	280(PI), 122 (MDD and OMC)	70:15:15	LMNN	3	7(PI), 11(MDD), 9(OMC)	1			R = 0.9373(PI) R = 0.9356(MDD) R = 0.9406(OMC)
Salahudeen et al. (2018)	G, LS, FSP, D10, D30, D60, Cu, Cc, LL, PL	MDD and OMC	CKD	90	70:15:15	MLP/BPA	10	1–10	2			OMC,MDD R = 0.983,9884 MSE = 0.0013,0.001 MAE = 0.0208,0.0321
Taha et al. (2018)	C, M, S, G, PI, LL, NM	MDD and OMC	Nano-Cu,Nano-Al,Nano-clay	75	80:20:00	FFN/BPA	7	1–20	1			Nano clay, Nano Al, Nano-copper(R2=0.987,0.991,0.989)
Salahudeen and Sadeeq, (2019)	PL, LL, G, LS, Cu, Cc, OMC and MDD	CBR (soaked and unsoaked)	CKD	72	70:15:15	FFN	8	8(Soaked),17(unsoaked)	1			Soaked,Unsoaked R = 0.9986,991 MSE = 0.00013,0.00109 MAE = 0.008,0.012
Tinoco et al. (2019)	WC, C, M, S, W/C, CC, OM, CA, CoB, CoSB	UCS	CC	444	70:30:00	DMA	10		1	GSA	W/C > CC>OM > CA	R2=0.94 ± 0.001 RMSE = 0.69 ± 0.05 MAE = 0.46 ± 0.02
Onyelowe et al. (2023)	CC, L, LL, PI, OMC, MDD	UCS	CC,L	190	74:26		9	5	1		MDD > LL,PL > CC>L > OMC
Hanandeha et al. (2020)	CC, L, PI, M, C, FA, OMC, WC	Resilient modulus (Mr)	L and FA	125		BPA	8	9	1	Parmeteric study by changing input parameter	C > L>FA > PI>C > WC>OMC > M	R²=0.97(training)
Priyadarshee et al. (2020)	C, RHA, CC, PA, CA	UCS	PA,RHA,CC	129	70:15:15	FFN/LMNN-BP	5	2–11	1	CWA	RHA > PA>C, CC > CA	R = 0.9856 R²=0.9714 MSE = 51.34 RMSE = 7.1651
Rajakumar and Babu (2020)	AsT, AsC, LL, PL, MDD, OMC and NoGeL	Soaked CBR	CoA, BA, GSA and geogrid	210		CBLF,GDBLF,LMNN,HWLR	7	7	1			R²=0.97
Salahudeen et al. (2020)	G, LS, Cu, Cc, LL, PI, OMC, MDD	UCS(7, 14, 28 days)	CKD	72	70:15:15	MLP/BPA	8	1–15	1			Soaked,Unsoaked R = 0.99976,9806 MSE = 0.000079,0.000079
Shah et al. (2020)	AE, AS, PI, G, OMC, MDD, AASTO, UCS, GI	CBR (10, 30, 65 blows compaction)	Alum sludge waste	36	78:22		9	1 layer	3			10,30,65 blows R²=0.9892,0.9963,0.9749 RMSE = 0.1586,0.1244,0.5921
Hu and Solanki (2021)	25 properties of cementitiously stabilized subgrade soils.	Mr	CC			RBF,MLP/LM,BRNN,SCG	(Best)25	1 and 2 layers(MLP)/(Best)15 neurons	1			MSE = 0.5644 (25 - 15-1 layer)
Salahudeen et al. (2020)	G, LS, Cu, Cc, LL, PI, OMC, MDD	Durability, Mr, resistance value	CKD	72	70:15:15	FFN/BPA	8	17(durability),24(resilient modulus),18(resistance value)	1			Durability,M_r, resistance R = 0.8388,0.8433,0.7572 MSE = 0.01258,0.01446,0.02368 RMSE = 0.112,0.12,0.154
Ngo et al. (2021)	ST, WC, We, CC, CeT D, Sl, Sd, Sa, Sv, ma, De, CA, CuC	UCS	CC	216	80:20:00	FFN/BPA	14	2–50	1	GB model	We > CC>Ma > WC>De > D>CuC > Sa>Sv > S>CeT > CA>Sd > Sl	UCS 7,14,28 days R = 0.9812,0.9783,0.9942
Tabarsa et al. (2021)	ST, CA, DUW, CC, RHA, L	UCS	CC, L, RHA	137	60:20:20	MLP/LMNN-BPA	6	7 and 4	1	Statistical software	CC,L > other	R = 0.9957,AAPE = 12.349,AIC = 4.289
Tran (2021)	CC, WC, AF, WFNC	UCS	CC, AF, WFN	51	70:30:00	LMNN	4	2 layers, 12 and 10 in each	1	Relative importance	CC > AF>WC > WFNC	R = 0.94 MAE = 8.6535 RMSE = 10.3390
Pham et al. (2021)	F0.5, F0.25, F0.1, SiO₂,Fe₂O₃,Al₂O₃,SO₃,K₂O,CaO,Ti₂OCC, CA	UCS	CC	80		FFN/LMNN-BPA	12		1	GA and CWA	CC > F0.5	R²=0.997 MSE = 0.0415 RMSE = 0.0614
Chen et al. (2022)	WC, T, σ_N, ωt	Shear stress	AGF	40	80:20	BPNN	4	3 layer, (best-20,20,20)	1	Based on BPNN	ωt>σ_N>T > W	R²=0.948 RMSE = 16.153 MAPE = 24.230
Mustafa et al. (2022)	G, S, M + C, LL, PI, LS,% stabilizer, stabilizer type, OMC, MDD, AR, DW	UCS	Unstabilized, L,CC	488	70:15:15	LMNN	12	1 layer (24,41,44 for Three dataset)	1	Modifying soil chracteristics		R²=0.9883(overall)
Abdallah et al. (2023)	CC, MDD, IWC, CA	UCS, WC and suction	CC		80:20	BRBP	4	2 layers, 10 and 5 in each	3			UCS, WC, suction R²=0.972,0.932,0.995
Aljanabi and Salih (2023)	% of soil, LL, PL, SL, additives(%); additive type	UCS	UIR, UAO, ShP, LGC, WPP, RH, SSP	74	60:20:20, 70:15:15, 80:10:10	MLP	6	3 layers, 12; (1–7);1	1	Normalized importance	Additive type > additives(%); >PL > LL> SL>% of soil	R = 0.98
Krishna et al. (2023)	LL, PL, OMC, MDD, L, CC, FA (for UCS) & UCS (for CBR)	CBR and UCS	L, CC, FA	125	80:10:10	FFN/LMNN-BPA	7(UCS)&8(CBR)	2 layer, 10 in each	1			CBR,UCS R²=0.924,0.95 MAE = 0.45228,0.0166 RMSE = 0.00537,0.0012
Kumar et al. (2023)	Soil(%), CC, L, LL, PL, PI, MDD, OMC	UCS	L,CC	100	60:10:30	LMNN-BP	8	2 layer, 16 and 32 neurons each(best)	1			R = 0.70
Lu et al., 2023	S, M, C, LI, Wc/Cc	UCS	CC	80	80:20	LM, BR	5	1,2,3,4,5 layers and 60 neurons	1	Varying value of input parameter	Wc/Cc > M > S > C>LI	R²=0.896 MSE = 13667.232 RMSE = 116.907 MAE = 102.584 MSLE = 0.042 RMSLE = 0.206
Onyelowe et al. (2024)	WGP, NACL, PA, LL, PL, FSI, OMC, MDD	CBR,UCS	WGP,NACL,PA	25	80:20	BPNN	8	1 layer (1,2,3 neurons)	2	Relative importance	OMC > WGP>PA > NACL>FSI > LL>PL > MDD	CBR,UCS SSE = 1.5%,2% R²=0.9979,0.9973
Anh et al. (2024)	WC, LL, PL, Wab	Aps	PSAS	147	60:20:20	MLP/BP ANN and GA ANN	4	2–12-BP ANN, 8-GA ANN	1			BP, GA R²=0.9269,0.9121 MSE = 0.0299,0.0244
Baldovino et al. (2024)	CA, MDD, WGP(%), CC, VWC, VCC, WGP + CC, P/CCi,P/Bi	UCS(7 and 28 days), Stiffness	WGP, CC	72	70:15:15	BP		1 layer,9	1			Stiffness, UCS R2=1,1 RMSE = 0.0384,0.0021
Goutham and Krishnaiah (2024)	BA, L, LL, PL, SL, MDD, OMC	UCS(28 days)	BA, L	79	70:30	MLP/LMNN-BPA	7	7	1	GA and CWA	LL > L>SL > BA>PL > MDD > OMC (GaA), BA > OMC>MDD > SL>LL > L>PL (CWA)	R²=0.99
Mojtahedi et al. (2024)	CA, W/C, CC, PI, G, S, C	UCS by deep mixing	CC	192	75:25	FFBP	7	1 layer, 7 &12	1	Analyzing current weights	G > W/CC > C > S > CA>CC > PI	R2=0.992
Thapa et al. (2024)	C,M, NS, BC, SI, MDD, WC, omc	CBR	NS, BC	175		LMNN-BPA	6	1layer, 10 neurons	1	XAI (SHAP and LIME)	C > M>MC > NS/BC > SI>MDD	R²=0.958 MSE = 0.02 RMSE = 0.049 MAE = 0.085
Wani and Thagunna (2024)	GSD, PI, LL, PL, WC, FA, CA	Shear strength	FA		85:15		7	5 layer	1		FA > CA>other	R²=0.69 MSE = 0.01
Mohammed et al. (2025)	CA, MDD, OMC, GGBS, PL, LL, PI	UCS	GGBS	200	80:20		7		1	SHAP and LIME	CA > OMC>MDD > GGBS>PL > LL>PI (SHAP)	R²=0.94 RMSE = 0.15 MAE = 0.037
Sharma et al. (2024)	L, CA, PI, pH, Vp, OMC, MDD	UCS, E, c, ϕ	L	54	80:20		7	10	4			UCS,E,c,ϕ R²=0.887,0.926,0.805,0.859 RMSE = 20.984,3.654,6.019,2.338 MAPE-7.722,13.338,10.325,6.803

Note(s):

# LL= Liquid Limit, PL= Plastic Limit, SL=Shrikage Limit, PI=Plasticity Index, C= clay content, M= silt content, S= sand content, Gr=gravel content, G=Specific gravity, WC=Water content, CC=Cement Content, CA=Curing age/time, LS=linear shrinkage, L=Lime, AC=Asphalt content, RC=Road Cement, OPC=Ordinary Portland cement, GGBS=Ground Granulated Blast Furnace Slag, FA=Fly ash, MoC=Molar concentration, A/B=Alkali to binder ratio, Na/Al=atomic ratio of sodium to aluminate, Si/Al=atomic ratio of silicate to aluminate, QD=Quarry dust, MDD=maximum dry density, OMC=optimum moisture content, P_0.005=Percentage finer than 0.005 mm, PoP=Pozzolon percentage, FC=fine content, CoE=Compaction energy, RHA=Rice husk ash, NoGl=Number of geotextile layer, G= specific gravity, grain sizes -(D10,D30,D60), Cu-uniformity coefficient, Cc-coefficient of curvature, Water/cement ratio=W/C, OM=Soil organic matter, Coefficient related with the binder type=CoB, Coefficient related with a secondary binder=CoSB, PA-Pond ash, OM=organic matter content, ST=Soil type, We=Wet density, CeT-Cement type, D-Soil sampling depth, Sl=specimen length, Sd=specimen diameter, Sa=specimen area, Sv=specimen volume, Ms=Mass of specimen, De=Density of specimen, CA, CuC=Curing condition, AsT=Ash type, AsC=Ash content, NoGel=Number of Geogrid layers, AF=Air foam, WFNC= Waste fishing net content, F0.5,F0.25,F0.1=Particle size distribution, SiO₂,Fe₂O₃,Al₂O₃,SO₃,K₂O,CaO,Ti₂O=Chemical composition, WD cycle=Wetting and Drying cycle, L/SAF=ratio of Free Lime to Silica, Alumina and Ferric Oxide compounds in the cementitious materials, MDD/OMC=ratio of maximum dry density to the optimum moisture content, σ3=confining stress, σd=deviator stress, BA=Bagasse ash, NM=Nano Material, DUW=dry unit weight, WGP=Waste glass powder, VGW=volumetric water content, VCC= volumetric cement content, WGP + CC=volumetric content of the sum of WGP and cement, P/Ci=porosity/cement index, P/Bi=porosity/binder index, IWC=Initial water content, LI=Liquidity Index, Wc/Cc = Water content/Cement content, Vp=primary ultrasonic wave velocity, pH=potential of hydrogen, T =Temperature, σ_N=Normal stress, ω^t= Shear displacement, AR=Aspect ratio, DW=Testing condition, GSD=Grain size distribution, AE=Applied energy, AS=Alum sludge, AASHTO =American Association of state highway and transportation, GI=Group index, NACL=Sodium Chloride content, FSI=Free swell Index, NS=Nano-silica, BC=Bio-char, SI=Soil index

@ OMC = optimum moisture content; MDD = maximum dry density; UCS = Unconfined compressive strength; CBR = California bearing ratio; Mr = Resilient Modulus, Aps = Amount of paper sludge ash stabilizer

! MSE = Mean squared error; MSLE = Mean squared logarithmic error; RMSE = Root mean squared error; RMSLE = Root mean squared logarithmic; E = modulus of elasticity; c = cohesion; Φ = angle of internal friction

$- CC = Cement Content, L = Lime, AC = Asphalt content, RC = Road Cement, OPC = Ordinary Portland cement, GGBS = Ground Granulated Blast Furnace Slag, QD = Quarry dust, PoP = Pozzolon percentage, FA = Fly ash, CKD-Cement klin dust, PA=Pond ash, RHA = Rice husk ash, CoA = Coal ash, BA = Bagasse ash, GSA = Groundnut shell ash, PSAS = Paper sludge ash stabilizer, FBA = Fluidized bed ash, UIR = iron ore, UAO = used automobile oil, ShP-shale rock powder, LGC = a mixture of hydrated lime, gypsum and cement, WPP = waste plastic pieces, RH = rice husk, SSP = sandstone powder, AGF = Artificial ground freezing

c FFN=Feed forward network, RBF=Radial Basis Function MLP=Multi-layer perceptron, BP=Backpropagation, BPA=Back propagation algorithm, GaA=Garson Algorithm, BRNN=Bayesian regularization method, LMN=Levenberg-Marquardt algorithm, DENN=differential evolution algorithm, GeA=General algorithm, DMA=Data Mining algorithm, CBLF=Conscience bias Learning function, GDBLF=Gradient descent weight and Bias learning function, HWLR=Hebb weight learning rule, RBF=Radial Bias Function, SCG=Scaled conjugate gradient, ABC-ANN=artificial bee colony-ANN hybrid

* GA = Garson Algorithm, CWA = Connection weight approach, GSA = Global sensitivity analysis, GB = Gradient boosting, BR = Bayesian Regularization, BRBP = Bayesian Regularization backpropogation, SHAP = Shapely Additive Explanation, LIME = Local Interpretable Model-agnostic Explanations, XAI = Explainable Artificial Intelligence

> indicates superior

Source(s): Author’s own work

[ViewLarge]

Sharing Unavailable