Mathematical reasoning is one of the most impressive achievements of human intellect but remains a formidable challenge for artificial intelligence systems. In this work we explore whether modern deep learning architectures can learn to solve a symbolic addition task by discovering effective arithmetic procedures. Although the problem might seem trivial at first glance, generalizing arithmetic knowledge to operations involving a higher number of terms, possibly composed by longer sequences of digits, has proven extremely challenging for neural networks. Here we show that universal transformers equipped with local attention and adaptive halting mechanisms can learn to exploit an external, grid-like memory to carry out multi-digit addition. The proposed model achieves remarkable accuracy even when tested with problems requiring extrapolation outside the training distribution; most notably, it does so by discovering human-like calculation strategies such as place value alignment.
Transformers discover an elementary calculation system exploiting local attention and grid-like problem representation
	
	
	
		
		
		
		
		
	
	
	
	
	
	
	
	
		
		
		
		
		
			
			
			
		
		
		
		
			
			
				
				
					
					
					
					
						
							
						
						
					
				
				
				
				
				
				
				
				
				
				
				
			
			
		
			
			
				
				
					
					
					
					
						
							
						
						
					
				
				
				
				
				
				
				
				
				
				
				
			
			
		
		
		
		
	
Cognolato S.;Testolin A.
			2022
Abstract
Mathematical reasoning is one of the most impressive achievements of human intellect but remains a formidable challenge for artificial intelligence systems. In this work we explore whether modern deep learning architectures can learn to solve a symbolic addition task by discovering effective arithmetic procedures. Although the problem might seem trivial at first glance, generalizing arithmetic knowledge to operations involving a higher number of terms, possibly composed by longer sequences of digits, has proven extremely challenging for neural networks. Here we show that universal transformers equipped with local attention and adaptive halting mechanisms can learn to exploit an external, grid-like memory to carry out multi-digit addition. The proposed model achieves remarkable accuracy even when tested with problems requiring extrapolation outside the training distribution; most notably, it does so by discovering human-like calculation strategies such as place value alignment.| File | Dimensione | Formato | |
|---|---|---|---|
| Cognolato and Testolin - IJCNN- 2022.pdf accesso aperto 
											Tipologia:
											Preprint (AM - Author's Manuscript - submitted)
										 
											Licenza:
											
											
												Accesso gratuito
												
												
												
											
										 
										Dimensione
										537.61 kB
									 
										Formato
										Adobe PDF
									 | 537.61 kB | Adobe PDF | Visualizza/Apri | 
Pubblicazioni consigliate
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.




