Yann Lecun : ESIEE skipped Préparatoire

December 10, 2025Chinoiseries2014 AI, Education Leave a comment

It is a significant distinction because it challenges the traditional narrative that top French scientists must come exclusively from the “royal path” (Polytechnique, ENS, etc.). LeCun’s journey is a powerful counter-example to the French élitisme of the Classes Préparatoires.
Here is a summary of the updated record regarding Yann LeCun’s education and his views on the French system, based on the details you provided:

The Rejection of the “Royal Path”
Contrary to the typical path of the French scientific elite, LeCun did not complete the grueling two-year Classes Préparatoires aux Grandes Écoles (CPGE).

The Drop-out: He enrolled in Maths Sup at Lycée Jacques Amyot in Melun in 1978 but left after only a few weeks.
The Reason: He explicitly rejected the “intellectual hazing” (bizutage intellectuel), the dogmatic teaching style, and the obsession with ranking rather than learning. He viewed the system as a mechanism for social reproduction and creating CEOs, rather than fostering scientific innovation.

The Strategic Choice of ESIEE
His move to ESIEE Paris (École Supérieure d’Ingénieurs en Électronique et Électrotechnique) was a deliberate pursuit of practical innovation over prestige.

Curriculum Gap: In the late 70s/early 80s, the “top” schools (X, Mines, Centrale) focused heavily on abstract mathematics and physics. Computer science was barely treated as a distinct discipline.
Hands-on Access: ESIEE offered access to actual microprocessors and allowed for hands-on programming. This environment was crucial for LeCun, who was interested in building machines and understanding intelligence, not just solving abstract equations. It was at ESIEE that he built his own computer and started reading about automata and early neural networks.

The “Godfather” of AI who avoided the System
LeCun’s success is often cited as a critique of the French academic system’s rigidity.

Innovation vs. Selection: He argues that the prépa system selects for conformity and speed in solving known problems, whereas research requires creativity, patience, and the ability to define new problems.
Peers: As you noted, he shares this trait with other top researchers like Yoshua Bengio, highlighting that the “elite” path is not the only (or even the best) generator of AI talent.
Updated Profile Summary

Yann LeCun

Education: Diplôme d’Ingénieur from ESIEE Paris (1983); PhD from Pierre and Marie Curie University (1987).

Key Narrative: Dropped out of Maths Sup to avoid the toxic competitive culture; chose ESIEE for its practical focus on electronics and computing, which allowed him to explore neural networks when prestigious schools ignored them.

Philosophy: Vocal critic of the French prépa/concours system, advocating for research-driven education over intense abstract selection.

——– Interesting to note those French who succeeded in their field (Science, Maths , Engineering ) outside the royal path of “Preparatoire – X / ENS”. Yann LeCun (inventor :Convolution Network) , who won “Turing Prize” aka “3 AI God Fathers” with his Postdoc Canadian Benjio (for “Backpropagation , Word2Vec”) , and his own mentor Hinton (for “Deeplearning”) is one such exception case from ESIEE (5 year private Grande Ecole sans Préparatoire), although he quit after 3 weeks from Preparatoire sick of Concours-focus Abstract Algebra. https://grok.com/share/bGVnYWN5LWNvcHk_4c460ec1-8956-4483-9b7a-660b2074f778

French Classe Préparatoire

October 28, 2025Chinoiseries2014 Education, Modern Math Leave a comment

If you think “Harvard Math 55” which had frightened Freshman Bill Gates who viewed it as the hardest undergrad Maths in the world, then the French Classe Préparatoire established since Napoléon is harder than “Harvard Math 55” – Proof: Galois had failed its Concours in X (Ecole Polytechnique), his 15-year High School junior Charles Hermite passed X in last position.

为啥法国数学这么强，平民的算数这么烂？

December 7, 2020Chinoiseries2014 Education, French 1 Comment

结账的算数：€ 313.5

华人：[给] €323.5 – [找] €10 ＝€313.5

法国人: [给] €320 – [找] ？＝€ 313.5

{? ＝€ 6.5 是这样算的：313.5+0.5 [凑整]＝314.0, 然后 314＋6＝320，得[找] 0.5＋6 ＝6.5}

https://m.toutiaocdn.com/i6896389485832389134/?app=news_article&timestamp=1607277607&use_new_style=1&req_id=202012070200070100140540151610B97E&group_id=6896389485832389134&tt_from=android_share&utm_medium=toutiao_android&utm_campaign=client_share

Rigorous Prépa Math Pedagogy

November 19, 2016tomcircle French, Modern Math 1 Comment

The Classe Prépa Math for Grandes Écoles is uniquely French pedagogy – very rigorous based on solid abstract theories.

In this lecture the young French professor demonstrates how to teach students the rigorous Math à la Française:

$\displaystyle {\lim_{n\to\infty} \bigl( 1 + \frac{1}{n} \bigr)^{n} = e}$

1st Mistake:
$\displaystyle { \bigl( 1 + \frac{1}{n} \bigr) \xrightarrow [n\to\infty] {} 1 \implies \boxed {{\bigl( 1 + \frac{1}{n} \bigr)}^{n} \xrightarrow [n\to\infty] {} 1^{n} =1}}$ (WRONG!!)

THEOREM 1 :
$\boxed { \text {If } {U}_{m} \xrightarrow [n\to\infty] {} \alpha \text{ and } f(x) \xrightarrow [x \to\alpha] {} \ell \text { then } f (U_{m}) \xrightarrow [n\to\infty] {} \ell}$

Note: The ‘x’ in f (x) is ${U}_{m}$ hence f is ‘fixed’ by a value $\alpha$

In the above mistake:
${U}_{m} = 1 + \frac{1}{n}$
$f _{n} (1 + \frac{1}{n} ) \text { where } f_{n} : x \mapsto x^{n}$
${f}_{n}$ is not fixed, but depends on ‘n’. It is wrong to apply Theorem 1.

2nd Mistake:
$\forall n \in \mathbb {N}^{*}, \bigl( 1 + \frac{1}{n} \bigr) > 1$

THEOREM 2:
$\boxed { \forall q > 1, q^{n} \xrightarrow [n\to +\infty] {} +\infty}$
Note: q is a fixed value.

$\text {Let } q_{n} = 1 + \frac{1}{n}$
Therefore,
$\boxed {{\bigl(1 + \frac{1}{n} \bigr)}^{n} \xrightarrow [n\to\infty] {} +\infty }$ (WRONG!!)

Reason: $q_{n} = \bigl( 1 + \frac{1}{n} \bigr)$ is not fixed value but depends on variable ‘n’. It is wrong to apply Theorem 2.

3rd Mistake: Binomial

$\boxed { \displaystyle \forall n \in \mathbb {N}^{*}, {\bigl( 1 + \frac{1}{n} \bigr)}^{n} = \sum_{k=0}^{n} \binom {n}{k} {(\frac {1}{n})}^{k} . (1)^{n-k} = \sum_{k=0}^{n} \binom {n}{k} {(\frac {1}{n})}^{k} }$

Note:
${\bigg[ \bigl( 1 + \frac{1}{n} \bigr)}^{n} = \binom {n}{0} {(\frac {1}{n})}^{0} + \binom {n}{1} {(\frac {1}{n})}^{1} + ... = 1 + 1 + ... > 2 \bigg]$
Expanding the binomial,
$\displaystyle \binom {n}{k} = \frac {n. (n-1). (n-2)... (n-k+1)}{k!} \sim_{n\to +\infty} \frac {n^k}{k!}$

$\boxed {\displaystyle\binom {n}{k}.{(\frac {1}{n})}^{k} \sim_{n\to +\infty}\frac {n^k}{k!}. {(\frac {1}{n})}^{k} =\frac {1}{k!}}$ Note: valid if k is fixed value.

$\boxed {\displaystyle {\bigl( 1 + \frac{1}{n} \bigr)}^{n} = \sum_{k=0}^{n} \binom {n}{k} {(\frac {1}{n})}^{k} \sim_{n\to +\infty} \sum_{k=0}^{n}\frac {1}{k!}}$ (WRONG !!)
Reason: k in the $\displaystyle \sum_{k=0}^{n}$ is not fixed, it varies from k = 0 to n

THEOREM 3:
$\displaystyle\boxed { {U}_{n} \sim_{n\to +\infty} {V}_{n}, \forall p \ge 1, ({U}_{n})^{p} \sim_{n\to +\infty} ({V}_{n})^{p} }$
Note: p is fixed value.
$\boxed {\displaystyle ({U}_{n})^{n} \sim_{n\to +\infty} ({V}_{n})^{n} }$ (WRONG!!)
Reason: n is variable to infinity.

Question:
$\text {If } {U}_{n}\xrightarrow [n\to\infty] {} 0, \text {then } {(1+{U}_{n})}^{\alpha} \xrightarrow [n\to\infty] {} \alpha.{U}_{n}$
Is below correct ?
${(1+\frac{1}{n})}^{n} \xrightarrow [n\to\infty] {??} n. \frac{1}{n} = 1$
[Hint] n is variable, $\alpha$ is fixed value.

Final Solution: Exponential

THEOREM 4:
$\boxed {\forall (a,b) \in \mathbb {R}_{+}^{*} \text { x }\mathbb {R}, a^{b} = e^{b.\ln {a}}}$

$\boxed { \displaystyle \forall n \in \mathbb {N}^{*}, {\bigl( 1 + \frac{1}{n} \bigr)}^{n} = e^{n. {\ln (1+ \frac {1}{n})}} }$ … [*]

THEOREM 5:
$\boxed {\text {If } {U}_{n}\xrightarrow [n\to\infty] {} 0, \text {then } \ln (1+{U}_{n}) \sim_{n\to\infty} {} {U}_{n} }$

Since
$\frac{1}{n} \xrightarrow [n\to\infty] {} 0, \text {then } \ln (1+ \frac{1}{n}) \sim_{n\to\infty} {} \frac{1}{n}$
Therefore,
$n.\ln (1+ \frac{1}{n}) \sim_{n\to\infty} {}1$

THEOREM 6: From equivalence ( $\sim_{n\to\infty} {}$ ) to find Limit ( $\xrightarrow [n\to\infty] {}$ ), and vice-versa
$\boxed { \text {If } {U}_{n} \sim_{n\to\infty} {} \ell \in \mathbb {R}, \text {then } {U}_{n} \xrightarrow [n\to\infty] {} \ell }$

Converse is true only if $\ell \neq 0$
eg. $\ell = (\frac {1}{n})_{n \in {N}^{*}} \neq 0$
because $(\frac {1}{n}) \xrightarrow [n\to\infty] {} 0$ but $\nsim_{n\to\infty} {} 0$

Therefore (from Theorem 6):
$\displaystyle \lim_{n\to\infty} n.\ln (1+ \frac{1}{n})= 1$

Since exponential function is continuous at 1 (why must state the condition of Continuity? )
hence, from [*], we have:
$\boxed { \displaystyle {\lim_{n\to\infty} \bigl( 1 + \frac{1}{n} \bigr)^{n} = e}}$

French Flawed Elite Higher Education: Classes Préparatoires (CP)

September 22, 2016tomcircle Education Leave a comment

3 top Classes Préparatoires (CP) (Baccalaureate + 2 years, equivalent to Bachelor of Math & Science) supply more than 50% of the elite Ecole Polytechnique (X) & Ecole Normale Supérieure (ENS) (Masters degree in Engineering or Business) students each year:
1. Lycée Louis-Le-Grande (LLG) – Paris
2. Lycée Privé Sainte-Geneviève – Paris (Versailles)
3. Lycée Henri IV – Paris

These students come from families in the rich Parisian regions, whose parents and grand-parents are also the alumni of X or ENS — this education phenomenon is called “In-Breeding”, or in Mathematical parlence “Closure”. The flaw of this “Social Immobility” is that “The rich gets richer, the poor gets poorer“, because poor provincial French families have no resources to send their children to these top “CPs” to better prepare for the elite Grandes Ecoles. Although the “Concours” (Entrance Exams 科举) is meritocratic and fair, the preparation for “Concours” in the 2-year excellent top “CP” is not a level-playing field.

This Concours which was inherited from ancient China had already manifested for thousand years in the flawed elitism of Chinese “Mandarin” bureaucracy (官僚制度) — the top scholars were from the same closed circles of elite families, either mandarin parents or rich merchants. China had few major revolutions initiated by failed “Concours” ( 科举) scholars of poor provincial sons, notably “黄巢”之乱 (Hwang Chao Revolution) in 9 CE which brought down the glorious Tang (唐) Dynasty, the 洪秀全 (太平天国 Taiping Revolution, 19 CE) also preluded 40 years before the demise of Qing (清) Dynasty. In certain resemblance, the poor farmer’s non-university-graduate son Mao ZeDong launched the “Cultural Revolution” to purge the “Smelly Elite Scholars” (臭老九) inherited from the “Old China”.

It is no surprise for France, after the glorious 30-year “5th Republique” (1950 – 1980), to see so many social unrests – “non-elite” university graduate high unemployment, Arabic-African Muslim immigrant descendents’ riots, ISIS terrorism …, while the mostly-white Grandes Ecoles elites monopolise the top social echelons in government, civil service, big state and private enterprises. The previous French President faced strong resistance from the elites to introduce “diversity” (ie reserve certain student quota for poor family children and minority races) in these top Grandes Ecoles.

http://mobile.lemonde.fr/societe/article/2011/10/12/ces-lycees-qui-monopolisent-la-fabrique-des-elites_1586236_3224.html?xtref=acc_dir

French Taupe: 3/2 & 5/2

May 4, 2013tomcircle Education, French 1 Comment

French elite Grandes Écoles (Engineering College), established since Napoleon with the first Military College (1794) École Polytechnique (nickname X because the College logo shows two crossed swords like X), entry only through very competitive ‘Concours’ Entrance Exams – to gauge its difficulty, Évariste Galois failed in two consecutive years.

Before taking Concours, there are two years of Prépas, or Classe Préparatoire (Preparatory class) housed in a Lycée (High school) to prepare the top Math / Science post-Baccalaureat students. These two undergraduate years are so torturous that French call these students Taupes (Moles) – they don’t see sunlight because most of the time they are studying 24×7, minus sleeping and meal time.

Most students take 2 years to prepare (Year 1: Mathématiques Supérieures, Year 2: Mathématiques Spéciales) for the Concours in order to enter X. These students are nicknamed 3/2 (Trois-Demi), so called playfully by the integration of X:

$\displaystyle\int_{1}^{2} xdx= \frac{1}{2}x^{2}\Bigr|_{1}^{2}=\frac{3}{2}$

If by the end of second year some students fail the Concours, they can repeat the second year, then these repeat students are called 5/2 (Cinq-Demi) – integrating X from Year 2 to Year 3:

$\displaystyle\int_{2}^{3} x dx=\frac{1}{2}x^{2}\Bigr\vert_{2}^{3}=\frac{5}{2}$

Évariste Galois was 5/2 yet he still failed X, not because of his intelligence but the incompetent X Examiner at whom the angry Galois threw the chalk duster. (Well done !)

Another famous 5/2 is René Thom (Fields medal 1958) who discovered ‘Chaos Theory’.

There are few rare cases of 7/2 (Sept-Demi):
$\displaystyle\int_{3}^{4} x dx=\frac{1}{2}x^{2}\Bigr\vert_{3}^{4}=\frac{7}{2}$
for those who insist on attempting 3 times to enter X or other elite Grandes Écoles. Equally good – if not better – is École Normale Supérieure (ENS) where Galois finally entered after having failed X twice. The tragic Galois was expelled by ENS for his involvement in the Revolution.

Note: Only 200 years later that ENS officially apologized in recent year, during the Évariste Galois Anniversary ceremony, for wrongfully expelled the greatest Math genius of France and mankind.

One of the top Classe Préparatoire “Lycée Pierre de Fermat” named after the 17th century great Mathematician of the “Last Theorem of Fermat”, in his hometown Toulouse, Southern France.

Math Online Tom Circle

Ideal Math = [(中文 + English) * Français] ^ IT

Tag Archives: Classe Préparatoire